Scope: This dossier assesses the distinct trajectories of quantum computing, quantum communications and cryptography, and quantum sensing through 2031, separating civilian and military applications and examining their technical evidence, institutional ownership, industrial dependencies and consequences for government decisions.

Executive Summary / BLUF

[D/P] The controlling judgment is that quantum policy requires separate investment and assurance decisions for computational experiments, cryptographic migration and sensing systems, because the verified record establishes different kinds of progress across these activities rather than a common transition to operational capability. Quantum error correction below the surface code threshold — Nature — Dec 2024; Announcing Approval of Three Federal Information Processing Standards (FIPS) for Post-Quantum Cryptography — NIST — Aug 2024. Nature

[P] The immediate executive obligation is to govern cryptographic migration against data lifetimes and replacement cycles, with European transition milestones extending from initiation by the end of 2026 to critical-infrastructure protection by the end of 2030, while the United Kingdom distinguishes discovery by 2028, priority migration by 2031 and completion by 2035. EU reinforces its cybersecurity with post-quantum cryptography — European Commission — Jun 2025; Timelines for migration to post-quantum cryptography — NCSC — Mar 2025. Shaping Europe’s digital future

[D/R] Error-correction experiments support continued investment in computing, but a demonstrated memory or logical-operation protocol must remain distinct from a vendor commitment to deliver a machine, including IBM’s target of 200 logical qubits and 100 million gates in 2029. A fault-tolerant neutral-atom architecture for universal quantum computation — Nature/NIST — Nov 2025; Quantum 2030 — IBM Technology Atlas — updated Mar 2026. NIST

[C/P] Government-reported navigation flight trials justify further evaluation rather than fleet-readiness claims, while the NSA’s position published on 1 October 2026 continues to reject QKD for national-security transmission unless specified limitations are overcome. Un-jammable quantum tech takes flight to boost UK’s resilience against hostile actors — DSIT/UKRI — May 2024; Cryptograph Solutions are Limited Compared to Quantum-Resistant Algorithms — NSA — Oct 2026. GOV.UK

Quantum Migration: The Hidden Industrial Bill

Quantum Policy Must Buy Protection and Evidence

Governments should finance quantum policy according to what it can protect or deliver, because the technology’s security timetable has already diverged from its computing timetable. NIST’s adoption of FIPS 203, FIPS 204 and FIPS 205 on 13 August 2024 gave institutions standards for preparing ordinary digital systems against future quantum attacks; it did not demonstrate a quantum computer capable of executing those attacks. That distinction should determine expenditure: ministries must replace vulnerable security dependencies while buying computing progress against verifiable milestones and measurement equipment against defined tasks. Confusing these responsibilities would expose long-lived information, misstate industrial achievement and reward suppliers for a label rather than a delivered function. The governing test is whether each commitment purchases protection, knowledge or an accepted service, with evidence appropriate to that purpose.

A security upgrade cannot certify a quantum computer

Post-quantum cryptography describes mathematical protection performed on conventional systems, rather than quantum processing inside the institution adopting it. ML-KEM establishes shared secret material; ML-DSA and SLH-DSA provide digital signatures. A successful implementation therefore demonstrates a cryptographic operation within its tested scope, while establishing nothing about the reliability of a quantum processor. The distinction matters politically because the phrase “quantum-safe” can make a cybersecurity purchase sound like the acquisition of quantum computing capacity, allowing an accurately described security improvement to become an unsupported industrial claim.

The Willow protected-memory experiment provides evidence of a different kind: physical quantum information preserved through error correction under specified conditions. Its significance lies in the engineering result, whose limits must remain visible, rather than in the adoption of a security algorithm elsewhere. Governments should assess these achievements through separate acceptance tests, because the information-protection programme has to reach the actual services holding sensitive data, while the computing programme has to establish the calculations its equipment can complete.

The figures measure different promises and obligations

Britain’s £2.5 billion quantum research and innovation commitment covers ten years from 2024; France’s May 2026 announcement adds €1 billion for 2026–2030; Australia’s A$940 million commitment concerns a particular computing project. These amounts cannot be ranked as comparable annual expenditure or completed capacity. Their different periods and scopes determine what each government has undertaken, while contract performance and operational acceptance determine what it ultimately receives. India’s March 2026 disclosure makes the distinction concrete: for IIT Bombay, the reported release was ₹114.89 crore and utilisation ₹22.31 crore, within a selected institutional row rather than the whole National Quantum Mission. Authorisation, payment and use remain different financial facts.

The technical calendars are equally incompatible. IBM’s Starling roadmap targets 200 logical qubits and 100 million processing operations in 2029; France’s PROQCIMA programme now targets 1,024 logical qubits in 2032, increased from its earlier target of 128. DARPA’s Quantum Benchmarking Initiative asks whether computational value can exceed cost by 2033. None establishes a completed service within the report’s 2026–2031 horizon. Britain’s National Cyber Security Centre, meanwhile, sets discovery and planning milestones for 2028, highest-priority migration work for 2031 and completion for 2035. Those dates govern organisational preparation rather than predict an attacker’s arrival, so a security deadline can precede computing maturity without contradiction.

The application evidence further narrows what these commitments can promise. The GIRAFE airborne gravimetry study, published in April 2025 from a June–July 2023 campaign, reported approximately 1–2 milligals accuracy for both quantum and conventional instruments; a milligal measures small gravity differences. The comparison supports evaluation of a defined instrument rather than automatic technological superiority. A July 2026 reinforcement-learning study reported approximately 20% additional logical-error suppression in specified Willow experiments, supporting a measured AI contribution to control rather than a claim that AI has completed the engineering of a reliable computing service.

Procurement must follow the result into operation

The Royal Navy’s August 2026 announcement concerning AQlock clocks and Saab radar equipment supports examination of a named timing trial. It does not establish complete navigation independence or fleet-wide readiness, because maintaining time and determining position are different functions. The procurement question is consequently whether the tested instrument can maintain the required service throughout the intended disruption, with supporting equipment, trained users and a workable recovery procedure. Buying a device without specifying that responsibility would leave acceptance dependent on the supplier’s preferred description.

The GIRAFE airborne gravimetry study offers a stronger comparison framework because it examined a quantum instrument alongside conventional equipment in an actual survey campaign. Its lesson for procurement is that the alternative must be competitive and the task explicit: measurement quality, processing requirements and maintenance determine usefulness more directly than the technology label. Quantum computing contracts need the same discipline, with a named calculation, complete execution cost and independent assessment. DARPA’s Quantum Benchmarking Initiative institutionalises that question by testing utility against cost, rather than treating access to equipment as sufficient evidence of value.

Standards open a transition without certifying the estate

FIPS 203, FIPS 204 and FIPS 205 specify mechanisms; they do not certify every product, replace every certificate hierarchy or verify every software-update path. NIST’s migration work distinguishes discovery of cryptographic dependencies from interoperability testing, exposing the management burden that lies between an approved algorithm and a protected service. A ministry has to establish where the older methods operate, which suppliers control replacement and whether both ends of a connection can support the change, while maintaining continuity during deployment. A procurement announcement cannot substitute for those acceptance results.

NSA’s published position on quantum key distribution also limits the claim that special quantum equipment provides a complete security answer. Its objections include authentication, trusted relays, implementation assurance and availability, with conditional rejection for the national-security transmission use discussed. That position should not be turned into a universal prohibition, but it requires a buyer to explain the threat model and the trust arrangement before crediting an additional benefit. Mathematical migration and specialised communications infrastructure need separate business cases, preventing a hardware proposal from inheriting the justification for a different security mechanism.

European coordination cannot replace control of components

The European Commission’s Quantum Europe Strategy, adopted in July 2025, identifies fragmentation and the difficulty of converting scientific strengths into market opportunities. Italy’s strategy adoption that month establishes a coordinating framework, while France’s PROQCIMA programme sets a computing ambition that still requires verified delivery. Germany’s quantum action concept links development to industrial implementation, and Britain’s National Cyber Security Centre gives system owners a staged security transition. These instruments serve different responsibilities; treating them as interchangeable national victories would obscure whether public money is building production, supporting research or changing operational systems.

Japan’s May 2025 ecosystem policy identifies concerns involving specialist lasers, detectors, diamond materials and refrigeration-related equipment, illustrating where industrial dependence can survive the purchase of a finished machine. The EU’s sovereignty objective therefore has to reach design, production, maintenance and replacement options, with contracts identifying indispensable suppliers. Cooperation can improve access, but the buyer still needs to know who repairs the equipment and who controls the components that keep it operating. National installation counts cannot answer those questions, and a domestic research achievement cannot by itself establish continuity of supply.

The next two years will charge delay to system owners

Between October 2026 and October 2028, the practical test should be whether institutions have converted NIST standards and the National Cyber Security Centre’s discovery milestone into funded inventories, tested changes and accountable service plans. Delay would leave the work with security teams, infrastructure operators and future procurement budgets, while long-lived information would continue to depend on the protection already in place. The cost is conditional exposure and a deferred replacement burden, rather than an invented estimate of losses from a quantum attack whose operational arrival remains unestablished.

IBM’s roadmap, PROQCIMA and the Quantum Benchmarking Initiative should face the corresponding industrial test: each next funding decision should identify which uncertainty has been reduced and which capability has been demonstrated. If governments report post-quantum cryptography as proof of quantum computing maturity, taxpayers would finance an overstated achievement and operational users would inherit an unverified promise. The decision over the next 24 months is to complete measurable security preparation and make further industrial commitments conditional on evidence; announcements alone would leave both responsibilities with the institutions expected to operate the result.

High-Speed UAS Evolution: Quantum Systems Apex & Interceptors

---

Navigational Index

The dossier retains the eleven substantive sections requested in the topic document within the governing protocol’s architecture of exactly three thematic pillars; chapter identifiers are retained in the index, while substantive headings remain unnumbered.

Pillar 0 — Quantum Technologies Explained for Political Leaders and Nontechnical Readers

What Works Today, What Still Prevents Wider Use, and What Governments Should Prepare for by 2031

Pillar I — Mechanisms, Hardware and Computational Evidence

Dossier section	Required coverage
Chapter 1 — Executive Adjudication	Five findings that change ministerial decisions; the evidence supporting each finding; the consequences of confusing a demonstration, programme, roadmap or unsupported claim.
Chapter 2 — The Taxonomy	The three technology stacks; mechanism-level interpretation of public terminology; a comparison of technical and institutional readiness; forbidden synonyms and misleading equivalences.
Chapter 3 — Quantum Computing: Hardware Truth	Superconducting, trapped-ion, neutral-atom, photonic, silicon-spin and topological platforms; physical and logical resources; fidelity definitions; demonstrated error correction; manufacturing and sovereign supply-chain constraints.
Chapter 4 — What Quantum Computers Can and Cannot Do in 2026	Chemistry, materials, optimisation, sampling, factoring and discrete logarithms, assessed against problem size, classical comparators, verification cost and the resources required for fault tolerance.

Pillar II — Security Migration and Operational Applications

Dossier section	Required coverage
Chapter 5 — Cryptography	QKD, quantum-secure direct communication, quantum networks, PQC, symmetric cryptography, harvest-now-decrypt-later exposure, named cryptanalytic resource estimates and jurisdiction-specific migration requirements.
Chapter 6 — Civilian Uses That Are Real	Finance, telecommunications, pharmaceuticals, energy, metrology and civil positioning, navigation and timing, retaining only applications supported by a named deployment, pilot, experiment or standard.
Chapter 7 — Military and Intelligence Uses That Are Real	Sensing and timing, communications and computing assessed separately; contested-spectrum PNT, magnetic-anomaly missions, subsurface detection, distributed timing, national-security cryptographic migration and experimental networks.

Pillar III — Industrial Power, Conditional Outlook and Decisions

Dossier section	Required coverage
Chapter 8 — The Industrial and Geopolitical Layer	United States, China, European Union, United Kingdom, Japan, India, Israel, Australia and Canada, with separate Italian, French and German lenses; transparent expenditure, hardware sovereignty, standards, talent and operative export controls.
Chapter 9 — Five-Year Outlook Under AI Acceleration	Constrained, base and accelerated pathways across 2026–27, 2028–29 and 2030–31, conditional on error correction, logical resources, decoder latency, manufacturing yield and cryptanalytic developments.
Chapter 10 — Implications for Decision-Makers	Application-level action matrix for civil government, defence, critical infrastructure and enterprise, distinguishing immediate action, monitored investment, unsuitable procurement and claims requiring retirement.
Chapter 11 — Annexes	Glossary and forbidden synonyms; source register with retrieval dates; ten claim corrections; uncertainty register and records capable of changing the assessment.

---

The Quantum Shock: Strategic Synthesis of Cryptographic Vulnerabilities and Multi-Domain Intelligence Resilience in the Era of Qubit Optimization

Front Matter

Method and Evidentiary Boundary

This assessment treats the supplied text file as the governing report protocol and the supplied Markdown file as the subject-specific specification, admitting peer-reviewed technical papers and explicitly identified preprints where the latter document theoretical estimates rather than experimentally established capability.

The governing question is: which decisions should institutions take between 2026 and 2031 when hardware demonstrations, security requirements, funded programmes and commercial forecasts advance on different schedules?

All sources cited in this opening delivery were retrieved on 4 October 2026; their publication dates identify the age of the evidence rather than implying that the underlying experiment, policy position or roadmap was newly established on the assessment date.

Tag Legend

Tag	Meaning and application
[D] Demonstrated	A peer-reviewed hardware result, bounded by its platform, experimental conditions, measured quantity and limitations; publication does not by itself establish independent reproduction or operational accreditation.
[P] Programme	An adopted standard, funded activity, contractual requirement or institutional migration framework, whose existence does not establish that implementation has been completed.
[R] Roadmap	A dated future target that remains conditional on milestones and delivery evidence.
[C] Claim	An attributed institutional or company assertion, including a theoretical resource projection that has not been demonstrated on the complete proposed hardware.
[H] Hype	A proposition that confuses mechanisms, removes a decisive limitation or assigns capability without a corresponding technical basis.

Where a sentence draws an analytical conclusion from evidence, its tag identifies the underlying evidence category, while wording such as “the assessment therefore” makes clear that the conclusion is a judgment rather than an additional experimental result.

What This Report Will Not Do

The dossier will not assign a universal maturity score to “quantum,” infer cryptanalytic capability from a physical-qubit headline, turn an experimental sensor into a military detection range, count announced investment as expenditure, or present a procurement recommendation without identifying the mechanism, mission, acceptance criteria and responsible institution.

Executive Adjudication

Cryptographic Migration Requires an Owner Before It Requires a Quantum Computer

[P] Finding — Decision changed: establish a funded migration authority, an asset inventory and replacement obligations rather than making preparation conditional on an agreed date for a cryptographically relevant quantum computer. The January 2026 G7 Cyber Expert Group statement addresses the financial sector’s transition as a coordinated governance problem, explicitly incorporating the exposure created when information intercepted before a future quantum capability remains sensitive when that capability becomes available; its timetable is described as non-authoritative and responsive to changing risk, which prevents its use as either a prediction of machine arrival or a universal regulatory deadline. G7 Cyber Expert Group Statement on Advancing a Coordinated Roadmap for the Transition to Post-Quantum Cryptography in the Financial Sector: January 2026 — G7 Cyber Expert Group/HM Treasury — Jan 2026. GOV.UK

The recommended governance response is to assign responsibility for confidentiality, authentication and software-signing dependencies separately, because the replacement plan must account for the lifetime of information, the lifetime of deployed equipment and the authority to change the relevant cryptographic implementation.

Computing Procurement Must Purchase a Workload Result

[D] Finding — Decision changed: condition computing expenditure on reproducible logical performance and a specified workload, with physical-qubit count treated as a hardware descriptor rather than an acceptance test. Google’s Willow paper establishes below-threshold surface-code memory experiments on processors containing 72 and 105 physical qubits, with a distance-5 experiment using an integrated real-time decoder and a distance-7 memory experiment; the authors explicitly identify additional challenges in logical computation, so this result supports an error-correction milestone without establishing a cryptanalytic machine or a general industrial advantage. Quantum error correction below the surface code threshold — Nature — Dec 2024. Nature

The proposed purchasing rule is to require the supplier to identify the computational task, logical operations, complete execution time, verification method and classical comparator before a machine-level performance claim enters an institutional business case.

QKD Requires a Specific Threat Model and Assurance Route

[P] Finding — Decision changed: separate QKD research or specialist-link evaluation from national-security cryptographic migration, rather than allowing a physical-layer security argument to determine system accreditation. In its October 2026 publication, the NSA retains a conditional position against QKD for national-security transmission and identifies absent source authentication, dedicated equipment, trusted-relay exposure, implementation vulnerabilities and denial-of-service sensitivity as material limitations; its wording includes “does not recommend using quantum key distribution,” while its broader position remains conditional on overcoming the listed problems. Cryptograph Solutions are Limited Compared to Quantum-Resistant Algorithms — NSA — Oct 2026. nsa.gov

The recommended approval test is to require a named authentication mechanism, relay-trust model, physical-security boundary, implementation evaluation and availability requirement, with the approving authority determining whether those properties satisfy the actual mission.

Sensing Investment Should Advance Through Mission Trials

[C/R] Finding — Decision changed: finance instrument integration and mission-relevant evaluation while withholding fleet-capability claims until published performance or accreditation evidence supports them. The British government reported flight trials involving Infleqtion, BAE Systems and QinetiQ at Boscombe Down in May 2024, supported by nearly £8 million in government backing, while describing aircraft deployment by 2030 as a national mission objective; the public announcement therefore establishes an officially reported trial and future target, rather than a disclosed navigation-error envelope or a completed operational deployment. Un-jammable quantum tech takes flight to boost UK’s resilience against hostile actors — DSIT/UKRI — May 2024. GOV.UK

The proposed acceptance programme should specify navigation drift, duration without external correction, environmental tolerance, maintenance burden and integration with existing navigation systems before the instrument is credited with a mission effect.

AI Funding Requires a Named Bottleneck and a Measured Gain

[D] Finding — Decision changed: fund AI as an experimentally evaluated control or decoding tool, with expenditure tied to the bottleneck it addresses rather than to a general claim of accelerated quantum maturity. A July 2026 Nature paper reports approximately 20% additional logical-error suppression from reinforcement-learning fine-tuning after conventional calibration, together with a distance-7 surface-code logical-error rate of 7.72(9) × 10⁻⁴ using the AlphaQubit2 decoder; this supports a specific improvement in the tested control-and-decoding setting, while providing no basis for substituting AI for the physical operations required by the architecture. Reinforcement learning control of quantum error correction — Nature — Jul 2026. Nature

Google’s 2029 Post-Quantum Deadline Is a Migration Target, Not the Arrival Date of Q-Day

Master Abstract

Three Stacks Require Separate Evidence and Separate Decisions

[D/P] The central distinction is between computing systems that manipulate and protect quantum information, communication systems that distribute quantum states or keying material, and classical cryptographic mechanisms standardised to resist specified classes of quantum attack; NIST’s FIPS 203 establishes a key-encapsulation mechanism, while FIPS 204 and FIPS 205 establish digital-signature mechanisms, so their adoption must be assessed as cryptographic implementation rather than evidence that a quantum communication or computing platform has become operational. Module-Lattice-Based Key-Encapsulation Mechanism Standard — NIST, FIPS 203 — Aug 2024; Module-Lattice-Based Digital Signature Standard — NIST, FIPS 204 — Aug 2024; Stateless Hash-Based Digital Signature Standard — NIST, FIPS 205 — Aug 2024. CSRC

The dossier’s recommended institutional architecture therefore assigns computing evaluation to workload and hardware specialists, cryptographic migration to accountable security and infrastructure owners, and sensing acquisition to the organisation that can define and test the intended mission; common scientific foundations do not remove the need for separate acceptance criteria.

Error-Corrected Operations Are Real, While Scale Remains an Engineering Question

[D] Neutral-atom evidence extends beyond a simple memory milestone: the paper published online in November 2025 and appearing in Nature’s January 2026 volume describes reconfigurable arrays of up to 448 atoms, below-threshold behaviour in a four-round characterisation circuit, logical entanglement operations and protocols involving dozens of logical qubits and hundreds of logical teleportations; the demonstrated elements support the feasibility of an architecture, while the experimental scope must remain explicit whenever the result is used to discuss scalable universal processing. A fault-tolerant neutral-atom architecture for universal quantum computation — Nature/NIST — Nov 2025. NIST

[C] The theoretical attraction of quantum low-density parity-check codes is their potential to reduce encoding overhead, but the peer-reviewed bivariate-bicycle work is a code-and-protocol analysis whose benefits depend on the assumed hardware and connectivity, so a favourable encoding rate must not be reported as a fielded processor possessing the complete required gate set, decoding performance and operating reliability. High-threshold and low-overhead fault-tolerant quantum memory — Nature — Mar 2024. Nature

The assessment recommends evaluating fault tolerance through the full execution chain, including state preparation, logical operations, measurement, decoding and the resources consumed by the chosen error-correction architecture, because a result established for one element cannot serve as evidence that all remaining elements have met the same performance standard.

Cryptanalytic Estimates Must Retain Their Complete Assumptions

[C] Gidney’s May 2025 preprint supplies a concrete, architecture-dependent RSA-2048 resource model rather than a demonstrated attack, with the following quantities reported in its logical-cost table and physical-cost analysis. How to factor 2048 bit RSA integers with less than a million noisy qubits — Craig Gidney, arXiv preprint — May 2025. arxiv.org

Resource or assumption	Value in the cited RSA-2048 model
Status	[C] Theoretical resource estimate; no complete attack demonstrated on the proposed hardware.
Logical resources	Physical-layout analysis allocates 1,537 logical patches including workspace, while the logical-cost table lists 1,399 algorithmic qubits and the layout text discusses 1,409 active logical qubits; these quantities describe different accounting boundaries.
Non-Clifford workload	Approximately 6.5 × 10⁹ expected Toffoli gates per factoring, as reported in Table 5.
Error-correction distance	Distance 25 for ordinary hot-storage surface-code patches; cold storage uses a yoked construction.
Encoding overhead	1,352 physical qubits per hot logical patch, approximately 430 per cold logical qubit, with separate computation and factory resources.
Complete physical layout	897,864 physical qubits, with additional slack motivating the headline near one million.
Physical noise assumption	Uniform depolarising noise strength of 10⁻³.
Timing and connectivity	1 μs surface-code cycle, 10 μs control reaction time, and a nearest-neighbour square-grid architecture.
Runtime	Approximately 4.96 days after the stated logical-error adjustment, rounded upward to less than one week.

[C] The assessment must also retain newer architectural alternatives: the March 2026 Cain and co-authors preprint examines high-rate codes, logical instruction sets and reconfigurable atomic hardware in an architecture distinct from the surface-code model above, which makes it relevant to the uncertainty register without permitting a lower headline qubit count to be treated as measured attack capability or as a directly comparable revision of the same machine. Shor’s algorithm is possible with as few as 10,000 reconfigurable atomic qubits — Cain et al., arXiv preprint — Mar 2026. arxiv.org

The decision consequence is to keep cryptanalytic resource estimates under technical review while progressing migration against the institution’s exposure, rather than deriving a single arrival year from estimates that assume different codes, connectivity, clock rates and implementation conditions.

Standards Provide a Migration Basis Without Providing Absolute Assurance

[P/R] NIST selected HQC for standardisation in March 2025 as an alternative mathematical basis to ML-KEM, while its announcement described finalisation in 2027 as an expected future milestone; the selection therefore supplies evidence of diversification in the standards programme, while that announcement alone cannot establish the subsequent completion or deployment status of an HQC standard. NIST Selects HQC as Fifth Algorithm for Post-Quantum Encryption — NIST — Mar 2025. NIST

[P] Symmetric cryptography requires a separate risk assessment: NIST’s FAQ distinguishes the dramatic algorithmic threat to susceptible public-key schemes from Grover-type search, emphasising the sequential operations and parallelisation constraints that limit the practical interpretation of a quadratic speed-up; consequently, this assessment rejects the use of an RSA resource model as evidence of a comparable attack on AES-256. Post-Quantum Cryptography: FAQs — NIST — undated, retrieved Oct 2026. CSRC

The recommended security case should therefore identify the actual primitive, parameter set, protocol and implementation, with confidentiality migration, signature migration and symmetric-key policy receiving separate acceptance tests.

European Coordination Must Preserve National Assurance Differences

[P] Italy’s institutional baseline includes the adoption of its national quantum strategy by the Interministerial Committee for the Digital Transition, recorded in July 2025, which establishes a coordinating framework without itself demonstrating that every strategic objective has an appropriated budget, awarded contract or delivered capability. Tecnologie quantistiche: una Strategia per l’Italia — Dipartimento per la trasformazione digitale — Jul 2025. innovazione.gov.it

[P/R] France’s ANSSI addendum, dated December 2023 and published through its service portal in January 2024, addresses hybridisation and an accelerated assurance agenda, while the stated expectation of initial security visas around 2024–25 remains a historical target unless a product-specific certificate establishes issuance. Avis de l’ANSSI sur la migration vers la cryptographie post-quantique (suivi 2023) — ANSSI — Dec 2023, p. 1. messervices.cyber.gouv.fr

[P/C] Germany’s government response of July 2025 records support for migration and the development of a national roadmap on the basis of the European framework, so the document supports institutional intent and stated policy without establishing completion of the later roadmap or migration outcomes. Aktuelle und zukünftige Sicherheitsrisiken für Bestände an Kryptowährungen durch Quantencomputer und Hackergruppen — Bundesregierung/Deutscher Bundestag, Drucksache 21/928 — Jul 2025, responses on migration. dserver.bundestag.de

The recommended coordination principle is to compare these jurisdictions through equivalent records—operative guidance, certification, expenditure, procurement and implementation evidence—while preserving differences between a national strategy, an assurance position and a sectoral migration timetable.

Key Evidence Table

Indicator	Value/status	Reference date	Definition/scope	Issuer	Exact source
[P] Standardised key establishment	FIPS 203; ML-KEM	August 2024	Key-encapsulation mechanism; the standard does not establish accreditation of every implementation	NIST	Module-Lattice-Based Key-Encapsulation Mechanism Standard — NIST — Aug 2024
[P] Standardised digital signatures	FIPS 204; ML-DSA	August 2024	Module-lattice signature mechanism	NIST	Module-Lattice-Based Digital Signature Standard — NIST — Aug 2024
[P] Alternative standardised signatures	FIPS 205; SLH-DSA	August 2024	Stateless hash-based signature mechanism	NIST	Stateless Hash-Based Digital Signature Standard — NIST — Aug 2024
[R] IBM computing target	200 logical qubits; 100 million gates; target availability in 2029	June 2025 announcement	Vendor delivery objective, requiring subsequent hardware and performance evidence	IBM	IBM Sets the Course to Build World’s First Large-Scale, Fault-Tolerant Quantum Computer at New IBM Quantum Data Center — IBM — Jun 2025
[P] European transition milestone	Initiation by end-2026; critical infrastructure by end-2030	June 2025	Coordinated roadmap expectations; national legal implementation must be examined separately	European Commission/NIS Cooperation Group	EU reinforces its cybersecurity with post-quantum cryptography — European Commission — Jun 2025
[P] UK migration milestones	Discovery and planning by 2028; priority migration by 2031; completion by 2035	March 2025	Indicative institutional migration targets, with qualifications in the guidance	NCSC	Timelines for migration to post-quantum cryptography — NCSC — Mar 2025

Technical and Institutional Readiness

The following classifications are dossier judgments tied to the cited evidence, rather than numerical scores or a claim that every application within a technology family has reached the same stage.

Application examined	Technical-readiness judgment	Institutional-readiness judgment	Evidence boundary
[D] Willow surface-code memory	Component demonstrated	Research programme	The experiment establishes protected-memory behaviour; the paper identifies further logical-computation requirements. Quantum error correction below the surface code threshold — Nature — Dec 2024
[D] Neutral-atom logical-processing architecture	Integrated experimental prototype	Research programme	The cited work demonstrates architectural elements and bounded protocols rather than an accredited industrial service. A fault-tolerant neutral-atom architecture for universal quantum computation — Nature/NIST — Nov 2025
[P] PQC standards and migration	Standardised mechanisms; deployment readiness requires implementation-specific evidence	Guidance and requirements established in named jurisdictions	Algorithm approval must remain distinct from product assurance and completed estate migration. Announcing Approval of Three Federal Information Processing Standards (FIPS) for Post-Quantum Cryptography — NIST — Aug 2024
[C/R] UK airborne quantum-navigation activity	Officially reported limited flight trial	Funded programme with future deployment objective	Publicly disclosed trial information does not establish a complete operational performance envelope. Un-jammable quantum tech takes flight to boost UK’s resilience against hostile actors — DSIT/UKRI — May 2024

Principal Gaps and Watch Indicators

Assessment-changing gap	Required record or observation	Decision threshold
[D] Sustained logical-computation performance	Reproducible workload results showing logical operations, complete runtime, error budget and all material overheads	Expand application procurement only when the tested system meets a predeclared workload requirement.
[C] Cryptanalytic resource-model feasibility	Architecture-specific validation of the assumed codes, connectivity, timing, logical operations and factory or equivalent resource provision	Reassess exposure when validated engineering evidence materially changes an end-to-end attack estimate.
[P] Migration completion	Asset-level evidence covering protocol, algorithm, parameter set, implementation, assurance and retirement of vulnerable dependencies	Report completion against protected assets and functions rather than the number of plans or products purchased.
[C/R] Navigation mission performance	Trial reports disclosing drift, duration, environmental conditions, external updates, integration and maintenance	Move from experimental evaluation to acquisition only against mission-specific acceptance criteria.
[R] Roadmap delivery	Dated hardware delivery and performance evidence corresponding to the stated milestone	Release later-stage funding against verified achievement rather than elapsed calendar time.

These collection priorities follow the gaps between demonstrated experiments, theoretical estimates and delivery targets identified above; they are proposed evidence requirements, not assertions that the relevant capabilities are absent outside the public record.

Conditional Pathways to 2031

The opening assessment uses alternatives and indicators rather than assigning probabilities, because a common numerical model for hardware progress, industrial scaling and institutional migration has not been established.

Pathway	Branch conditions	Decision consequence
Constrained pathway	Larger systems fail to preserve their demonstrated logical performance, or decoding, interconnection and manufacturing prevent sustained workload execution.	Continue cryptographic migration and mission-specific sensing evaluation while limiting computing expenditure to measurable research and enabling infrastructure.
Base pathway	Protected logical operations become more reproducible and systems support selected workloads, while remaining short of a validated cryptanalytic architecture.	Expand computing pilots through independent comparison with classical methods, retaining separate budgets and acceptance criteria for security migration.
Accelerated pathway	Error correction, logical operations, control latency, manufacturing and architecture-specific resource reductions succeed together rather than through isolated milestones.	Reassess long-lived information exposure and deployment priorities while demanding complete attack-model evidence before attributing cryptanalytic capability.

[R] IBM’s current roadmap illustrates the need for this conditional treatment, since it explicitly describes its information as intent subject to change and places Starling availability in 2029, while positioning a larger Blue Jay system at 2033 and beyond; the roadmap is a relevant industrial commitment, but it does not establish which branch will occur or demonstrate a cryptographically relevant computer within the dossier’s window. Quantum 2030 — IBM Technology Atlas — updated Mar 2026.

---

Pillar 0 — Quantum Technologies Explained for Political Leaders and Nontechnical Readers

What Works Today, What Still Prevents Wider Use, and What Governments Should Prepare for by 2031

Scope and Method

This opening pillar explains the findings of the report without requiring the reader to understand advanced physics, mathematics or computer engineering. Its purpose is to make the political choices understandable: what deserves immediate implementation, what deserves controlled experimentation, what requires patient research, and what should not be presented to the public as an existing capability.

It introduces the meaning of the technical chapters rather than reproducing their complete inventories of machines, national budgets or cryptographic resource estimates. Where a number is necessary to understand the situation, its source and limitations are stated.

The evidence labels used throughout the report retain their original meaning:

Label	Meaning in plain English	How the reader should interpret it
[D] Demonstrated	A result supported by a scientific experiment or independently reproduced technical evidence	Something worked under the stated conditions; wider usefulness still needs to be assessed
[P] Programme	A standard, funded programme, contract or formal requirement exists	An institution has authorised or organised work; this does not establish a completed operational service
[R] Roadmap	A future target or conditional technical estimate	Something may become possible if the necessary milestones are achieved
[C] Claim	A company or public authority reports a result without sufficient independent technical evidence in the cited record	The statement has an identifiable source, but its claimed performance remains to be independently established
[H] Hype	A misleading label, unsupported generalisation or confusion between different technologies	The statement should not guide procurement or public policy in its present form

Definitions, teaching examples and proposed policy controls are identified as explanation, illustration or analytical recommendation. They do not claim that a particular product has achieved a particular performance.

All sources linked in this pillar were retrieved on 4 October 2026. Publication dates and future target dates are distinguished throughout.

0.1 — The Situation in Plain English

The principal political difficulty is that the word “quantum” is used for several different technologies, each with a different state of development. An announcement about a computer, a communications link or a precision clock may sound like evidence of the same technological revolution, although it concerns a different practical problem and a different level of readiness.

The evidence reviewed in this report supports an uneven picture. [P] New standards for protecting information against future quantum attacks already exist. [D] Scientists have demonstrated important improvements in the reliability of quantum information stored in experimental processors. [D] Certain quantum measurement instruments have been tested outside the laboratory. [R] Large, dependable quantum computers capable of carrying out demanding applications remain the subject of future engineering targets. These findings come from different bodies of evidence and should lead to different public decisions. Sources: NIST standards announcement, quantum error-correction experiment, airborne gravity measurements and IBM hardware roadmap. NIST

For a political reader, the essential distinction is between an instrument that can perform a defined task, an experiment that establishes a scientific result, and a future system expected to transform an industry. Each may justify public attention, but each requires a different explanation of what the expenditure will buy.

The present situation at a glance

Area	Evidence-based position	Practical interpretation for a political reader
Protection of digital information	[P] NIST finalised ML-KEM, ML-DSA and SLH-DSA standards on 13 August 2024	Organisations can begin adopting new cryptographic methods without owning a quantum computer
Quantum computing reliability	[D] Published experiments show that carefully organised groups of physical qubits can preserve quantum information more reliably than individual components	A necessary engineering step has been demonstrated; a complete industrial service requires much more
Large quantum computing systems	[R] Public roadmaps specify future targets for reliable calculations	A target date should be treated as a milestone to verify, rather than as a guaranteed delivery date
Quantum communications	[P] Security authorities distinguish special quantum links from software-based protection and identify substantial deployment limitations	A secure communications proposal must explain its actual mechanism and the systems it leaves exposed
New quantum measurement instruments	[D] Some instruments have produced results in real surveys and field experiments	Suitability should be assessed application by application, including cost, maintenance and comparison with existing equipment
New defence timing applications	[C] Named military authorities report trials with specified equipment	A reported trial supports further evaluation; it does not establish fleet-wide deployment or complete mission independence

Sources: NIST, Nature: error correction, IBM, NSA: quantum key distribution, Earth System Science Data: airborne gravimetry and Royal Navy: August 2026 clock trials. NIST

Analytical judgment. A government should therefore manage quantum technologies as several connected policy responsibilities, with separate delivery tests. A national strategy can bring these responsibilities together, but a single promotional label should never replace an assessment of whether a particular capability works.

Why Post-Quantum Cryptography Works on Ordinary Computers and Does Not Demonstrate the Maturity of Quantum Computing

Section	Subject
The central distinction	What post-quantum cryptography actually is
How the name creates confusion	Why “post-quantum” describes a security objective
What happens inside the system	Ordinary computers performing different mathematics
What successful implementation proves	Security progress and its evidentiary limits
Why preparation precedes the threat	The separate timelines of protection and computing
Why quantum communications are different	Distinguishing mathematical protection from quantum equipment
The political consequences	Accurate procurement, budgets and public communication
Key judgments and open record	What should be verified before making a capability claim

The Central Distinction

Post-quantum cryptography is cryptography that ordinary computers can perform, designed to resist attacks from both conventional computers and sufficiently powerful future quantum computers. Its successful implementation demonstrates that a particular security mechanism can operate; it does not demonstrate that a quantum computer performed that operation or that quantum computing has reached industrial maturity. [P] NIST’s National Cybersecurity Center of Excellence explicitly describes standards intended to work with current classical computers while resisting future quantum machines. Source: Migration to Post-Quantum Cryptography — NIST NCCoE — maintained programme resource. NCCoE

The misunderstanding arises when the name of the anticipated threat is mistaken for the operating mechanism of the protection. A reader hears that a bank, ministry or telecommunications provider has adopted “post-quantum security” and concludes that the institution is already using quantum computing to secure its information. That conclusion does not follow from the announcement, because a post-quantum implementation can carry out its cryptographic work entirely through conventional digital processing.

The distinction is particularly important when assessing claims that “quantum technology is already working”. That expression needs an identified subject: a computing experiment, a measurement instrument, a communications device or a cryptographic implementation. Evidence for one cannot automatically establish the performance of another, and the adoption of post-quantum cryptography cannot be counted as evidence that a quantum processor has completed a useful calculation.

Analytical judgment. The legitimate achievement is the development and implementation of protection against a class of anticipated attacks. Describing that achievement precisely preserves its value while preventing it from becoming evidence for an unrelated technological claim.

How the Name Creates Confusion

“Post-quantum” describes the intended resistance

[P] In NIST’s explanation, post-quantum cryptographic methods are selected around mathematical problems believed to remain difficult for conventional and quantum attackers. The term therefore concerns the protection sought against an attacker’s capabilities, rather than a requirement that the legitimate user operate a quantum computer. Source: What Is Post-Quantum Cryptography? — NIST — official explanatory resource. NIST

Explanation. A useful analogy is a fire-resistant door: the description identifies a hazard against which the door is designed to provide protection, rather than something the door uses to function. In the same way, the word “quantum” in “post-quantum cryptography” identifies an attacker considered in the security design; it does not identify the machinery required to perform the protective calculation.

The prefix “post” can create a second misunderstanding by suggesting that the technology belongs to a period after powerful quantum computers have already arrived. That interpretation is also unnecessary. A protective method can be developed, standardised and implemented before the threat against which it is designed becomes operational.

Analytical implication. A political briefing should therefore explain the term as “cryptography designed to resist future quantum attacks” before discussing implementation. This wording gives the reader the relevant security objective without implying that the organisation owns quantum hardware.

The adjective does not certify the complete system

A supplier’s description of a product as “post-quantum” also leaves a further question unanswered: which functions have actually been changed? The label alone cannot establish that every connection, identity mechanism, certificate, software update and supporting device has been assessed.

[P] The NCCoE’s migration documentation distinguishes cryptographic discovery from interoperability testing, reflecting the need to understand where protection is used and whether updated components work together. Source: Frequently Asked Questions about Post-Quantum Cryptography — NIST NCCoE — migration documentation. nccoe.nist.gov

Analytical recommendation. The institution should state the scope of its implementation in operational terms, identifying the services covered and the dependencies that remain. An accurate limited claim provides more useful assurance than an expansive label whose coverage cannot be verified.

What Happens Inside the System

Ordinary processing carries out the protective mathematics

Explanation. In a conventional implementation, the legitimate computers execute instructions that generate cryptographic material, establish keys or produce and verify signatures. The calculations use ordinary digital processing; the change concerns the mathematical construction and its implementation within the security system.

[P] NSA makes the mechanism explicit by distinguishing quantum-resistant algorithms implemented on existing platforms from QKD equipment that uses quantum physical signals. For the algorithmic approach, security rests on the difficulty of the relevant mathematical problems. Source: Quantum Key Distribution and Quantum Cryptography — NSA — official agency position. nsa.gov

Consequently, an organisation does not need a quantum processor merely to perform a post-quantum cryptographic operation. It may need updated software, compatible protocols, suitable conventional hardware or replacement products, but those implementation requirements should be identified separately from the engineering requirements of a quantum computer.

The importance of this distinction should not make migration sound effortless. Ordinary computing can still face difficult changes involving performance, compatibility, long-lived equipment and coordinated replacement. The fact that the mechanism is conventional tells the reader what kind of technology is involved; it does not guarantee that every existing product can support it without modification.

Different standards perform different security functions

[P] NIST finalised three initial post-quantum standards on 13 August 2024, providing concrete mechanisms for key establishment and digital signatures. Their publication establishes the standards, while the security and suitability of a particular implementation require their own assessment. Source: NIST Releases First 3 Finalized Post-Quantum Encryption Standards — NIST — Aug 2024. NIST

Standardised mechanism	Plain-English function	What its use establishes	What it does not establish
[P] ML-KEM	Helps two systems establish shared secret key material	Implementation of a specified key-establishment method	That a quantum computer generated the key
[P] ML-DSA	Produces and verifies digital signatures	Implementation of a specified signature method	That quantum processing authenticated the document
[P] SLH-DSA	Provides a signature method based on hash constructions	Implementation of another specified signature approach	That the protected service operates through quantum hardware

Sources: FIPS 203, Module-Lattice-Based Key-Encapsulation Mechanism Standard — NIST — Aug 2024, FIPS 204, Module-Lattice-Based Digital Signature Standard — NIST — Aug 2024 and FIPS 205, Stateless Hash-Based Digital Signature Standard — NIST — Aug 2024. CSRC

[P] FIPS 203 describes ML-KEM as a mechanism for establishing a shared secret over a public channel and characterises its security in terms of a mathematical problem believed to be difficult even for quantum computers. This is a security assumption supporting a defined construction, rather than a claim of absolute immunity from every conceivable attack. Source: FIPS 203 — NIST — Aug 2024. CSRC

Analytical implication. Political communication should preserve that qualification. “Designed to resist quantum attacks using a specified standard” identifies an assessable protection objective; “unbreakable quantum security” converts it into a much broader assertion.

What Successful Implementation Proves

A security result has its own evidentiary meaning

Illustration — hypothetical institution. Suppose a ministry updates a communications service and demonstrates that the new key-establishment method works between its approved systems. The test produces relevant evidence about that implementation: the systems can negotiate the specified method and carry out the tested operation.

The same test does not show that the ministry owns a quantum computer, that a quantum processor has overcome its physical error problems or that a future attacker can execute a cryptographic attack. Those propositions concern different equipment and different experiments.

This distinction is logical rather than rhetorical: evidence that a defence operates cannot, by itself, establish the operational maturity of the technology against which it was designed. The protective system and the potential attacking system have separate development requirements.

Quantum computing requires evidence from quantum computing

[D] A Willow experiment first published in December 2024 demonstrated protected quantum memory using a defined superconducting processor arrangement. That evidence concerns physical quantum information and error correction, giving it a different evidentiary meaning from the implementation of a cryptographic algorithm on conventional systems. Source: Quantum error correction below the surface code threshold — Nature — Dec 2024; corrected Apr 2026. Nature

Analytical judgment. A report assessing computing maturity must examine such hardware results against their actual function and limits. A report assessing security migration must examine coverage, interoperability and implementation assurance. Combining the two into a single count of “quantum deployments” would obscure what each result demonstrates.

This correction should also avoid the opposite overstatement. Establishing that post-quantum cryptography uses conventional processing does not show that every quantum technology is unusable or that quantum computing research has produced no results. It establishes a narrower and essential point: post-quantum cryptography is not the evidence needed to substantiate a quantum computing capability claim.

Why Preparation Precedes the Threat

Future resistance can have present value

[P] NIST identifies the possibility of an adversary retaining encrypted information for later exploitation, commonly called “harvest now, decrypt later”. This concern is relevant where information remains sensitive over a long period, even though collecting it does not establish the collector’s present ability to decrypt it through quantum computing. Source: What Is Post-Quantum Cryptography? — NIST — official explanatory resource. NIST

Explanation. An institution therefore has to consider the lifetime of its information alongside the time required to change its systems. The practical question is whether protection can be improved before valuable information is exposed to a future capability, rather than whether a precise arrival date can already be stated.

[P] NCSC treats migration as a substantial organisational change involving discovery, planning, prioritisation, suppliers and continuity. Its guidance reflects the burden of changing a large digital environment, rather than a requirement to acquire a quantum computer. Source: Timelines for migration to post-quantum cryptography — NCSC — Mar 2025. National Cyber Security Centre

Analytical judgment. Early preparation is consequently compatible with uncertainty about quantum computing. A government can justify a security transition because information is long-lived and replacement is slow, while continuing to describe future attacking capability as conditional.

Adoption cannot establish an arrival date

[H] Inferring that an encryption-threatening quantum computer already exists because institutions are adopting post-quantum cryptography reverses the logic of precautionary preparation. The adoption concerns a protection decision, while the existence and performance of an attacking machine require direct evidence.

The opposite inference is equally unsound: the ability to deploy post-quantum cryptography does not establish that all relevant systems have been protected or that every future security risk has been removed. The implementation’s scope remains a separate question, and a completed operation in one service cannot establish coverage across an entire ministry, bank or industrial network.

Why Quantum Communications Are Different

Mathematical resistance and quantum signals are separate mechanisms

[P] QKD uses special-purpose technology and quantum physical signals to generate and distribute key material. NSA’s published discussion separates this mechanism from quantum-resistant algorithms and identifies further requirements involving authentication, equipment, relay trust and implementation security. Source: Quantum Key Distribution and Quantum Cryptography — NSA — official agency position. nsa.gov

Explanation. The distinction can be expressed through the legitimate user’s activity. In post-quantum cryptography, conventional systems perform cryptographic calculations designed around resistance to quantum attacks. In QKD, specialised equipment uses quantum signals for a key-distribution function. Both belong in a discussion of future information protection, but their mechanisms and deployment requirements differ.

Analytical recommendation. When a supplier uses expressions such as “quantum protection” or “quantum-safe communications”, the authority should require it to identify whether the proposal uses post-quantum algorithms, QKD, both, or another specified mechanism. Without that answer, the institution cannot explain accurately what it is buying.

The Political Consequences

Classify the expenditure according to the capability purchased

Analytical recommendation. A programme implementing post-quantum cryptography should be evaluated as a security transition, with success measured through protected functions, verified coverage, compatibility and maintained service. Its association with a quantum-related threat does not justify reporting its expenditure as the acquisition of quantum computing capacity.

Conversely, a quantum computing programme should be evaluated through the performance of its quantum equipment, the calculations it can execute and the conditions under which it delivers useful results. A shared strategic umbrella can coordinate the programmes, but their acceptance criteria must remain distinct.

Failure to preserve this distinction would allow a government to report progress in a category that its expenditure did not actually deliver. It could also mislead legislators into believing that a successful security upgrade has resolved the separate engineering difficulties of quantum computing.

Report what the institution has actually accomplished

Analytical recommendation. A defensible announcement would state that an institution has implemented a specified post-quantum mechanism in identified services, explain the coverage of the change and describe the remaining work. If the deployment uses conventional processing, that fact should be made explicit when there is a risk of confusion.

The question “Which quantum computer performs this protection?” should therefore receive a mechanism-based answer: the post-quantum cryptographic operation does not require one. If the overall product includes some additional quantum component, that component must be identified and assessed independently; its presence cannot be inferred from the name of the algorithm.

For nontechnical audiences, the clearest replacement for the potentially ambiguous definition is:

Post-quantum cryptography: cryptographic methods performed by ordinary computers and designed to resist attacks from conventional computers and future quantum computers; implementing these methods does not demonstrate the maturity of quantum computing.

Key Judgments

Analytical judgments. The first judgment is that post-quantum cryptography should be understood through its operating mechanism: conventional digital processing performs the protective mathematics. The second is that its implementation provides evidence about a security deployment, whose scope and correctness still require verification. The third is that this evidence cannot establish the performance or arrival date of a powerful quantum computer.

These judgments support action rather than postponement. An institution can pursue a justified security transition while demanding separate evidence for quantum computing claims, preserving both the urgency of information protection and the integrity of its technology assessment.

What Would Change the Assessment?

Analytical review criterion. The classification of a particular product would change if its technical documentation established an additional quantum operating mechanism, but that would be a finding about the specified product or component. It would not change the meaning of post-quantum cryptography or make a quantum processor necessary for performing the standardised algorithms.

A broader computing judgment would require direct evidence about the relevant quantum system, while a stronger security judgment would require verified deployment coverage and implementation assurance.

Open Official Record

The record needed to substantiate an institutional claim includes the algorithm and protocol used, the services covered, the implementation tests, the acceptance authority and the unresolved dependencies. A product label, purchase announcement or national programme title cannot supply those details on its own.

Final analytical assessment. Post-quantum cryptography is a concrete means of preparing ordinary digital systems for a potential quantum threat; recognising its conventional operating mechanism allows political leaders to value the security achievement accurately, without presenting it as proof that the separate engineering challenge of powerful quantum computing has been completed.

0.2 — The Different Technologies Hidden Behind the Word “Quantum”

A simple map

Technology	Plain-English explanation	The question it is intended to answer
Quantum computing	A different way of carrying out certain calculations using carefully controlled physical systems	Can a particular difficult calculation be performed better than with existing computers?
Post-quantum cryptography	New mathematical methods for protecting information, implemented on ordinary digital systems	Can information remain protected if a powerful quantum computer becomes available to an attacker?
Quantum key distribution, or QKD	Special equipment that uses quantum signals to help two parties establish secret key material	Can a particular communications link obtain keys under an explicitly defined security arrangement?
Quantum networking	Equipment designed to connect systems that store or process quantum information	Can quantum information or shared quantum states be distributed reliably between locations?
Quantum sensing	Instruments that use controlled physical effects to measure quantities such as gravity or magnetic fields	Can a measurement become more useful, accurate or dependable for a specific task?
Atomic timing	Clocks that use the stable behaviour of atoms as their reference	Can systems maintain the timing accuracy they need?

Explanation. These definitions identify mechanisms, not product maturity. A clock, a computer and a communications link can all involve quantum physics while performing entirely different functions.

The distinction matters because policy language often moves too quickly from a scientific term to a promised public benefit. “Quantum-safe” may refer to software using post-quantum cryptography. “Quantum communication” may refer to special optical equipment. “Quantum navigation” may refer to a clock, a motion sensor or an experimental combination of several instruments. A proposal that does not identify the mechanism cannot be assessed properly.

How to translate a headline into a useful question

Headline or phrase	The question a nontechnical decision-maker should ask
“The country has acquired a quantum computer”	What calculations can it complete, with what reliability, and who will use the results?
“The network is quantum-safe”	Which cryptographic methods protect its connections, identities and software updates?
“The link cannot be hacked”	What happens if an endpoint, administrator, relay or software component is compromised?
“The ship has quantum navigation”	Does the equipment provide time, position, motion measurements or a complete navigation solution?
“Quantum technology will accelerate drug discovery”	Which stage of discovery has been tested, at what scale, against which existing method?
“AI has solved the quantum problem”	Which measured engineering problem improved, and which physical obstacles remain?
“A national quantum programme is operational”	Is this a research programme, an installed prototype, an accepted service or a maintained deployment?

Analytical recommendation. Public communications should name the function before using the label. “A clock tested for maintaining timing during a satellite-signal disruption” gives a minister substantially more information than “a breakthrough in quantum defence”.

0.3 — Quantum Computers: Why They Matter and Why They Remain Difficult to Use

What a quantum computer is expected to contribute

Explanation. A quantum computer uses physical behaviour that ordinary digital computers do not use in the same way. Its potential value depends on designing a suitable calculation around that behaviour.

[H] The description that a quantum computer simply tries every answer and then reads out the correct one is misleading. NIST explains that the information available from the final measurement is limited; useful algorithms must arrange the calculation so that the measurement reveals something valuable. Source: NIST, Quantum Computing Explained. NIST

For a political reader, the consequence is straightforward: a new type of computer does not automatically improve every public service. A ministry should ask whether its specific problem has an appropriate algorithm, whether the available machine can execute it reliably, and whether the result improves on the best existing approach.

Why the number of qubits can mislead

Explanation. A physical qubit is a physical component used to hold or process quantum information. A logical qubit is a protected unit of information constructed using an error-correction arrangement. Several physical components may be required to support one protected unit.

An analogy is a public record stored with checks and recovery procedures. Counting the individual storage elements tells us something about the equipment, but it does not tell us whether the record can survive errors or remain usable throughout a long task.

[D] A study first published on 9 December 2024 reported a protected quantum memory on Google’s Willow processors. One arrangement used 101 physical qubits and achieved an error rate of approximately 0.143% per error-correction cycle. Increasing the protection improved reliability under the tested conditions. The demonstrated function was protected memory, which should not be described as an entire industrial computer completing arbitrary long applications. Source: Nature, Quantum error correction below the surface code threshold. Nature

What must be solved before a demanding calculation becomes dependable

Problem	Plain-English meaning	Why it matters to a user
Errors in the physical components	The equipment does not always perform exactly as intended	Incorrect operations can undermine the answer
Loss of stored quantum information	The useful state changes or becomes disturbed before the task finishes	A long calculation needs information to survive
Error correction	The system must repeatedly detect and manage faults	Protection has to work throughout the calculation
Protected operations	Reliable information must also be processed reliably	Preserving a record is different from carrying out a long sequence of correct changes
Control and feedback	Ordinary electronics and software must direct the quantum equipment	The supporting system must respond within the required time
Manufacturing consistency	Many components must meet demanding specifications	A good experimental device does not establish reliable production
Operational stability	Performance must remain acceptable over extended periods	Users need repeatable service, rather than an occasional successful run
Application performance	The complete workflow must produce a useful result	A scientific milestone may not yet improve a business or public task

Explanation. This table describes the questions that must be answered. It does not assign the same obstacle or the same cost to every hardware design.

[D] Research published in November 2025 established limits on noisy quantum calculations under specified mathematical models of errors. Its conclusions reinforce the importance of reliability, while its assumptions prevent it from being treated as proof that every near-term quantum application is impossible. Source: npj Quantum Information, Limitations of noisy quantum devices in computing and entangling power. npj Quantum Information

What “usable now” should mean

Analytical recommendation. Ministers should require suppliers and research bodies to distinguish four statements:

Statement	What it establishes	What remains to be demonstrated
“Researchers can access the machine”	Access to equipment exists	Whether it improves a particular task
“The machine completed an experiment”	A specified experiment ran	Whether the result is useful and repeatable
“The result exceeded a stated comparison”	An advantage was reported against that comparison	Whether the comparison remains competitive and the full cost is favourable
“The service improves our operation”	A user-facing benefit has been established	Whether it remains dependable as workload and deployment increase

Analytical judgment. Quantum computing can be worth using for research and carefully designed trials before it becomes a dependable tool for a demanding production task. Public policy should recognise both possibilities and require an honest description of which one public money is purchasing.

0.4 — Protecting Information: Why Action Is Necessary Before the Future Computer Arrives

The two different security clocks

[P] NIST’s explanation of post-quantum cryptography identifies a present planning problem: an adversary can retain protected communications in the hope that future computing capabilities will make some of them readable. This is called “harvest now, decrypt later”. It does not establish that the adversary currently possesses a computer capable of performing the future attack. Source: NIST, What Is Post-Quantum Cryptography?. NIST

Explanation. There are two separate clocks:

Clock	What it measures	Why a government should care
The secrecy clock	How long information must remain confidential	Some material could still cause damage many years after collection
The migration clock	How long an organisation needs to replace vulnerable protection	Discovery, procurement, testing and replacement may take several budget cycles

Illustration — hypothetical dates, not a forecast. Suppose information created in 2026 must remain confidential for fifteen years. Its protection matters until 2041. If the institution needs several years to upgrade its systems, beginning only after an attack-capable computer is publicly demonstrated could leave a long period of avoidable exposure.

The political decision therefore concerns the lifetime of the information and the time needed to improve its protection. It cannot sensibly be reduced to a confident prediction of the year in which a future computer will arrive.

What can be implemented without a quantum computer

[P] Post-quantum cryptography consists of mathematical methods intended for ordinary computing platforms. NIST’s final standards include ML-KEM, which helps establish shared secret keys, and ML-DSA and SLH-DSA, which provide digital signatures. Their existence gives organisations a concrete basis for implementation work. Sources: FIPS 203, FIPS 204 and FIPS 205. CSRC

Security function	Plain-English purpose	What the implementation programme should verify
Establishing secret keys	Allow legitimate systems to agree on secret material used to protect a connection	Compatible implementations, correct configuration and acceptable performance
Digital signatures	Allow a system to verify the origin and integrity of a message, document or software update	The complete chain of trust, including certificates, signing equipment and verification software
Replacing older methods	Remove vulnerable dependencies from the service	Whether every relevant connection and component has actually been addressed
Keeping future replacement possible	Avoid making the next change unnecessarily difficult	Whether contracts and system designs allow cryptographic methods to be updated

Analytical recommendation. A standard should be treated as the beginning of an implementation programme. An organisation must still verify that its products, operating procedures and suppliers use the standard correctly.

Why migration is a public-management problem

[P] The UK National Cyber Security Centre’s guidance sets milestones for discovery and planning by 2028, highest-priority migration work by 2031, and completion by 2035. These are migration targets within the guidance, rather than predictions of when an attacker will obtain a powerful quantum computer. Source: NCSC, Timelines for migration to post-quantum cryptography. National Cyber Security Centre

Analytical recommendation. A political authority should require a programme that answers the following questions:

Management question	Required answer
What information is most valuable over time?	A prioritised list with an identified owner
Which services depend on vulnerable cryptographic methods?	A documented assessment covering relevant systems and suppliers
Which equipment can be updated?	A tested upgrade path
Which equipment must be replaced?	A funded replacement plan
What happens during the transition?	A continuity plan and tested recovery arrangements
Who verifies completion?	A named authority responsible for acceptance
How are exceptions handled?	A recorded risk decision with a review date

The threat should remain specific

[H] “A quantum computer will break all encryption” is an inaccurate generalisation. NIST distinguishes the implications for public-key systems from those for symmetric methods and notes the substantial practical complications in applying quantum search attacks. Source: NIST post-quantum cryptography FAQs. CSRC

Analytical judgment. The appropriate response is a targeted security transition. A dramatic claim about the collapse of every form of encryption can obscure the actual systems that require attention.

0.5 — Quantum Communications: What They Protect and What They Leave Exposed

Understanding quantum key distribution

Explanation. A QKD system uses special physical equipment to help establish secret key material between locations. The resulting keys can be used within a communications security system. This function should be described separately from the protection of computers, users, software and buildings.

[P] NSA’s published position identifies limitations involving authentication, dedicated equipment, trusted relays, implementation security and denial of service. It states: “NSA does not recommend the usage of quantum key distribution and quantum cryptography” for the national-security systems discussed unless the stated limitations are overcome. Source: NSA, Quantum Key Distribution and Quantum Cryptography. nsa.gov

Questions that remain after a quantum link is installed

Question	Why it still matters
Are the communicating parties correctly identified?	A security system must establish who is entitled to participate
Are the endpoints protected?	Information can be exposed before protection is applied or after it is removed
Are intermediate locations trusted?	Some network designs place sensitive responsibilities in relay sites
Can the equipment be maintained and updated?	Security must survive faults, changes and newly discovered weaknesses
Can the link be interrupted?	Detecting interference does not guarantee that communication remains available
Does the whole system meet the required security standard?	A strong component cannot establish the security of every connected component

Analytical recommendation. QKD procurement should be justified by a clearly defined need, an explicit trust arrangement and a comparison with alternatives. The decision-maker should receive an explanation of what additional protection is obtained and what operational burden is created.

A quantum network is a different proposition

[D] A paper published in May 2024 demonstrated shared quantum states between experimental memory systems using 35 kilometres of deployed telecommunications fibre. The fibre route was a metropolitan loop; its length should not be presented as the straight-line separation between distant operational users. The result established a networking experiment, rather than a general replacement for ordinary internet services. Source: Nature, Entanglement of nanophotonic quantum memory nodes in a telecom network. Nature

Analytical judgment. Such results can justify research into future networks. They do not, by themselves, justify promising a population that a nationwide “quantum internet” is available or necessary for every public service.

0.6 — Quantum Sensors and Clocks: Where Practical Use Is Further Advanced

Why measuring is a different challenge from computing

Explanation. A sensor measures a physical quantity. A clock provides a timing reference. Neither must necessarily perform the long, complex sequence of protected calculations required by a large quantum computing application.

This distinction explains why a useful measurement instrument can emerge while a much more demanding computing system remains under development. Nevertheless, each instrument still needs to demonstrate that it can perform its intended task under the conditions in which users will operate it.

Established atomic timing and newer instruments should be separated

[D] Germany’s national metrology institute, PTB, operates primary atomic clocks to realise the official unit of time. Its explanation describes the atomic basis of the second and the clocks used for this public measurement responsibility. This is an established metrology function, which should be assessed separately from the readiness of newer compact or mobile instruments. Source: PTB, Realisation of the SI second. PTB.de

What named experiments actually establish

Application	Evidence	What a nontechnical reader should conclude
Airborne gravity measurement	[D] A paper published on 17 April 2025 reports flights conducted in Iceland and Greenland during June–July 2023, comparing the GIRAFE cold-atom instrument with a conventional gravimeter	A real airborne measurement campaign took place; the paper describes a technology still at the development stage
The same airborne comparison	[D] Both instruments achieved accuracy of approximately 1–2 milligals, a unit used for small gravity differences	The evidence does not support a blanket claim that the quantum instrument was superior in every respect
Underground structure measurement	[D] A 2022 study reported detection of a known two-metre tunnel through a defined gravity survey	A specific experiment worked; general discovery of arbitrary underground objects requires separate evidence
Defence timing	[C] The Royal Navy reported trials on 14 August 2026 involving two AQlock cold-atom clocks, Saab radar equipment and deliberately disrupted timing conditions	There is a named trial to examine; the announcement does not provide a complete independent account of operational performance

Sources: Airborne gravimetry study, Nature, Quantum sensing for gravity cartography and Royal Navy clock trials. essd.copernicus.org

Why a better clock does not automatically mean complete navigation

Explanation. Time, position and movement are related but different pieces of information. A clock can support systems that depend on precise timing, but it does not by itself report a vehicle’s complete position and route.

[H] Presenting a clock trial as proof that a ship or aircraft can navigate indefinitely without satellite signals would extend the claim beyond the function tested.

Analytical recommendation. Any proposal for navigation resilience should specify:

Required specification	Plain-English meaning
The service maintained	Timing, position, heading, velocity or a defined combination
The duration	How long the required performance can be sustained
The operating conditions	Motion, temperature, vibration and relevant interference
The supporting equipment	Maps, conventional sensors, communications and external references
The acceptable error	How much uncertainty the mission can tolerate
The fallback procedure	What the operator does when performance deteriorates

Civilian and military usefulness must be evaluated separately

Analytical judgment. A civilian survey and a military mission can use related measurement principles while requiring different evidence. A geological survey may tolerate processing after the flight. A navigation task may need a dependable answer immediately. A laboratory sensitivity result does not settle either operational question.

For civilian procurement, the principal test should be whether the instrument improves an identified service at an acceptable cost. For defence procurement, the same test must also address the mission environment, hostile interference, deployment procedures, training and recovery after failure.

0.7 — From an Experiment to a Dependable Public Service

A major source of misunderstanding is the assumption that a successful demonstration is the final step before widespread adoption. Political readers need a more complete picture of delivery.

The delivery stages

Analytical framework.

Stage	What has been achieved	What the public authority should require next
Scientific experiment	A result has been obtained under stated conditions	Independent examination and confirmation of the limits
Working component	One part of a system performs its intended function	Evidence that it can work with the other required components
Integrated prototype	Several parts work together	Testing against the intended user task
Limited field trial	Equipment has operated in a relevant setting	Repeatability, failure records and comparison with existing methods
Accepted deployment	A user has approved it for a defined responsibility	Maintenance, training, support and continued performance monitoring
Repeatable production	Multiple units can be delivered consistently	Quality control, supply continuity and manageable lifetime cost

Two kinds of readiness

Analytical framework. Technical readiness asks whether the equipment works. Institutional readiness asks whether an organisation can use and sustain it.

Technical situation	Institutional situation	Policy implication
Equipment works, but no organisation owns the task	No operator, budget or acceptance authority	Assign responsibility before describing a service as deliverable
A programme is authorised, but the equipment remains experimental	An institution has committed to development	Track technical milestones separately from spending
A trial succeeds, but maintenance is unresolved	The user cannot yet sustain deployment	Fund operational preparation before expanding installation
A service performs reliably and support is established	Operators, procedures and resources are in place	Consider broader adoption against evidence from actual use

Why the existing alternative remains important

Analytical recommendation. Every proposed application should be compared with the best relevant existing method. The comparison should include:

Comparison item	Why it matters
Accuracy or quality of the result	Establishes whether the new method performs the task better
Time to obtain a usable answer	Includes preparation and processing
Availability	Shows whether users can depend on access
Staffing	Captures training and specialist support
Maintenance	Identifies the effort needed to keep equipment working
Lifetime expenditure	Includes installation, operations and replacement
Failure consequences	Establishes the cost of an incorrect or unavailable result
Independent verification	Prevents the supplier’s preferred benchmark from becoming the only measure

[P] DARPA’s Quantum Benchmarking Initiative explicitly investigates whether quantum computing can reach a scale at which its computational value exceeds its cost, with an assessment horizon of 2033. This is an evaluation objective, rather than a promise that a useful system will necessarily be delivered by that date. Source: DARPA, Quantum Benchmarking Initiative. DARPA

0.8 — What Artificial Intelligence Can Realistically Improve

The useful political question is whether AI improves a particular bottleneck, by a measured amount, under relevant conditions.

A demonstrated contribution

[D] Research published on 8 July 2026 used reinforcement learning to adjust the controls of Google’s Willow processor. For specified error-correction experiments, further tuning reduced logical error rates by approximately 20% beyond the existing calibration. The study also distinguished experiments conducted through repeated runs from simulations exploring uninterrupted operation. Source: Nature, Reinforcement learning control of quantum error correction. Nature

Explanation. This is a practical example of AI helping engineers operate equipment more effectively. It is not evidence that the physical equipment, error correction or manufacturing requirements have become unnecessary.

How to assess an AI contribution

Proposed role for AI	Plain-English mechanism	Evidence a decision-maker should request
Adjusting equipment controls	Search for settings that improve measured performance	Improvement on actual hardware, stability and limits
Interpreting error signals	Identify likely faults from the information the machine produces	Accuracy and the time required to deliver the decision
Searching for materials or designs	Reduce the number of candidates engineers must examine	Physical validation of the suggested design
Testing security implementations	Find weaknesses in software or equipment	A reproducible weakness, its scope and a tested correction
Combining sensor information	Use several measurements to estimate a useful quantity	Comparison with existing methods in relevant field conditions

[D] The AlphaQubit study published on 20 November 2024 reported machine-learning improvements in interpreting quantum error information. It included data from physical processors and results from simulations, which must remain separately identified. Source: Nature, Learning high-accuracy error decoding for quantum processors. Nature

Analytical judgment. AI can improve the speed and quality of engineering work. A political authority should still demand evidence that the improvement survives integration into the complete system.

0.9 — What May Change Between 2026 and 2031

Forecasts should be conditional

[R] IBM’s public roadmap targets a system called Starling for 2029, with 200 logical qubits and the ability to run 100 million quantum gates, meaning individual processing operations. These are vendor targets. Their significance depends on whether the complete specified performance is demonstrated and independently assessed. Source: IBM Quantum hardware and roadmap. Hardware and roadmap

Analytical recommendation. Governments should plan around several possible outcomes and identify the evidence that would move the assessment from one outcome to another.

Three possible development paths

[R] Conditional analytical scenarios — not assigned probabilities and not delivery promises.

Scenario	What happens	What holds it back or enables it	Political response
Constrained development	Research advances, but dependable computing applications remain narrow; some specialised instruments progress	Reliability, manufacturing or integration improvements remain insufficient	Continue security migration, support rigorous research and expand only applications that pass their own acceptance tests
Steady development	More protected calculations become possible and selected applications begin to demonstrate useful results	Several engineering milestones are achieved together	Support independent benchmarking and controlled adoption by identified users
Accelerated development	Reliability and production improve faster, allowing broader and more demanding trials	Strong components, fast control, reliable operations and a credible application all succeed together	Prepare procurement and governance early while retaining full verification requirements

The practical questions for each period

[R] Conditional outlook. These are proposed review questions, not claims that particular capabilities will exist in the stated years.

Period	Quantum computing	Information protection	Sensors and timing	Required evidence
2026–2027	Can protected components support increasingly demanding integrated experiments?	Have organisations identified vulnerable dependencies and begun tested changes?	Can promising instruments repeat their results in relevant conditions?	Reproducible experiments, implementation tests and field records
2028–2029	Are roadmap targets achieved as complete systems?	Have priority services removed the relevant vulnerable dependencies?	Can multiple units be maintained by actual users?	Independent acceptance results, reliable operation and production records
2030–2031	Does a defined application deliver a dependable benefit over a competitive existing method?	Are transition results verified across the intended services?	Does deployment improve the mission throughout its operating conditions?	Measured outcomes, lifetime cost, support performance and accountable ownership

When might encryption-threatening computing arrive?

[R] A resource estimate first submitted by Craig Gidney on 21 May 2025 models factoring an RSA-2048 number in less than a week using fewer than one million noisy physical qubits. Its logical-cost table gives 1,399 logical qubits and approximately 6.5 billion Toffoli operations, a particular kind of processing operation. The physical model uses surface-code protection, including distance-25 arrangements, with differentiated storage overhead. Assumptions include 0.1% physical gate errors, one-microsecond correction cycles, ten-microsecond control reactions and nearest-neighbour connectivity. This is a clearly identified preprint resource estimate, not an executed attack. Source: Gidney, resource estimate and full technical model. arxiv.org

Explanation. The number of components is only one requirement. The machine would also need to perform the required operations reliably, retain information throughout the task and sustain the assumed control performance.

Analytical judgment. The selected evidence reviewed here does not establish that such an attack will be operational by 2031. It also does not justify a guarantee that the risk lies safely beyond every system’s service life. Governments should manage the uncertainty through migration, monitoring and technical review.

Useful progress can occur without a dramatic headline

[R] Conditional analytical outlook. By 2031, meaningful progress could include better protected information, dependable specialised instruments, more reliable computing experiments, stronger supply chains and better evidence about useful applications. None requires a claim that quantum computers have replaced ordinary computing across the economy.

0.10 — The Political and Industrial Stakes

What a government is really trying to secure

Analytical judgment. The industrial question concerns the ability to obtain, operate, repair and improve technology over time. Ownership of a machine is one part of that ability.

Form of capability	Plain-English meaning	Political relevance
Understanding	National institutions can assess performance and limitations	Reduces dependence on promotional claims
Design	Engineers can develop important parts of a system	Preserves the ability to adapt technology
Production	Suppliers can deliver components consistently	Supports continuity and industrial participation
Operation	Users can run equipment and interpret results	Turns an installation into a service
Maintenance	Faults can be diagnosed and repaired	Prevents dependence on a fragile support arrangement
Security assurance	Qualified bodies can examine protection and implementation	Supports trusted deployment
Standards participation	Institutions can contribute to common technical rules	Influences compatibility and future market access
Replacement options	Users can change suppliers or components	Reduces long-term dependence

Why supply chains matter

[P] Japan’s quantum ecosystem policy dated 30 May 2025 addresses dependence on important components and materials, including specialist lasers, detectors, diamond materials and refrigeration-related equipment. The document provides evidence of an official industrial concern; it should not be treated as an independently audited inventory of every global dependency. Source: Japan Cabinet Office, quantum ecosystem measures. www8.cao.go.jp

Analytical recommendation. A public investment proposal should therefore identify the suppliers required to keep the system working and the alternatives available if access is interrupted.

Why scientific strength may not become industrial strength

[P] The European Commission’s Quantum Europe Strategy, adopted on 2 July 2025, identifies difficulty translating scientific strengths into market opportunities and fragmentation across national strategies. Its stated areas include research, infrastructure, industrial development, security and skills. Source: European Commission, Quantum Europe Strategy. Shaping Europe’s digital future

Analytical judgment. For political leaders, this creates a responsibility to connect research with users, production and support. Funding excellent science remains valuable, but the intended public benefit must be described accurately: scientific knowledge, industrial capacity and an operational service are different outputs.

How to understand a funding announcement

Analytical framework.

Financial statement	What it means	What it does not establish
A budget is announced	An authority states an intended financial commitment	That all the money has been legally authorised or spent
Funding is authorised	Expenditure has an approved basis	That the project has received every payment
A contract is awarded	A supplier has an agreed responsibility	That the required performance has been delivered
Money is paid	Funds have been transferred	That the system has passed acceptance
Equipment is installed	Physical delivery has occurred	That users can depend on the intended service
Performance is accepted	The agreed test has been passed	That future maintenance and supply continuity are automatically secured

Analytical recommendation. National comparisons should identify these categories before adding figures or ranking countries. A large announced envelope and a smaller completed deployment cannot be compared as if they measure the same achievement.

0.11 — Decisions, Public Expenditure and Questions for Suppliers

What deserves immediate action

Analytical recommendations.

Decision area	Action now	Evidence required before expanding expenditure
Long-lived confidential information	Identify exposure and establish a cryptographic migration programme	Verified implementation and removal of relevant vulnerable dependencies
Public-service digital identities and software trust	Assess signatures, certificates and update mechanisms	End-to-end tests of the replacement trust arrangements
Timing resilience	Identify services affected by loss or manipulation of external timing	Performance over the required disruption period
New measurement instruments	Commission trials around a clearly stated task	Repeatability, comparison results and support requirements
Quantum computing applications	Fund bounded experiments with competitive existing methods as comparators	A useful result with documented reliability and full workflow cost
Special quantum communications links	Examine the actual mechanism and trust arrangement	A justified benefit for the specific application
Industrial capacity	Map critical dependencies and support justified bottlenecks	Deliverable production or support capability
Workforce	Train operators, engineers, security specialists and procurement staff	Evidence that skills can support the intended programme

How to structure public support

Analytical recommendation. Separate funding into purposes that can be reviewed honestly.

Purpose	What public money buys	Appropriate success measure
Research	Knowledge and reduction of scientific uncertainty	Reproducible results and clearly stated unresolved questions
Prototype development	Evidence that parts can work together	Integrated performance against a defined task
Field evaluation	Evidence under relevant operating conditions	Recorded performance, faults and comparison results
Deployment	A service accepted by an identified user	Availability, accuracy, support and lifetime expenditure
Security transition	Improved protection across relevant services	Verified coverage and controlled exceptions
Industrial development	Production, repair or supply capability	Consistency, lead times and continuity

Questions a minister should ask before signing

Question	Why it changes the decision
What exact public problem will this address?	Establishes whether there is an identified need
What has already been demonstrated?	Separates evidence from intention
Under which conditions did it work?	Reveals whether the evidence fits the proposed use
What is still missing?	Makes the remaining development responsibility visible
What is the best existing alternative?	Prevents comparison with a weak or outdated method
Who owns the result and accepts delivery?	Establishes accountability
What happens if the principal milestone is missed?	Defines the response to delay or failure
What will it cost throughout its service life?	Includes operations and support
Which suppliers are indispensable?	Reveals continuity risks
Can performance be independently examined?	Protects the authority’s ability to verify claims
How will users be trained and equipment repaired?	Tests whether deployment can be sustained
How will the public benefit be measured?	Connects technical expenditure with an accountable outcome

How leaders should communicate the situation

Analytical recommendation. Public statements should describe the capability at its demonstrated level.

Evidence available	Appropriate public wording
A scientific result	“Researchers demonstrated the specified effect under the reported conditions.”
An authorised programme	“The government has funded work to develop and evaluate the capability.”
A future target	“The programme aims to reach this milestone, subject to the stated technical requirements.”
A reported trial	“The authority reports a trial; operational performance remains under evaluation.”
An accepted service	“The system has met the specified acceptance requirements for this defined use.”

A clear statement of limits protects the credibility of the programme. It also allows future progress to be recognised when the next milestone is actually achieved.

0.12 — Key Judgments, Evidence That Would Change Them and Source Register

Key judgments

1. Different quantum technologies require different political decisions.
Analytical judgment. A software security transition, a field instrument and a future computing architecture should have separate owners, budgets and acceptance criteria.

2. Security preparation has a practical basis today.
[P] Final post-quantum standards provide a basis for implementation, while official guidance treats migration as a substantial organisational task. The decision changed is whether to begin discovery, testing and replacement planning now. Sources: NIST standards and NCSC migration guidance. NIST

3. Demonstrated improvement in quantum computing reliability is significant, but it does not establish every promised application.
[D] Protected-memory experiments provide evidence of progress. Analytical judgment. The decision changed is how research milestones should be funded and verified before operational commitments increase. Source: Nature, error correction below threshold. Nature

4. Measurement and timing deserve application-specific assessment.
[D] Established atomic metrology and published field measurements demonstrate concrete functions. Analytical judgment. The decision changed is whether to commission a defined operational comparison rather than wait for quantum computing to mature. Sources: PTB atomic timing and airborne gravimetry. PTB.de

5. Future capability should be governed through evidence and alternative plans.
[R] Roadmaps specify goals rather than guaranteed outcomes. Analytical judgment. The decision changed is whether procurement and national strategies contain independent reviews, staged commitments and responses to missed milestones. Source: IBM roadmap. Hardware and roadmap

What would change this assessment?

Analytical review criteria.

Area	Evidence that would justify a stronger assessment
Quantum computing	An independently assessed application delivering a dependable benefit against a competitive conventional method
Reliability	Protected operations sustained at the scale and duration required by the intended task
Manufacturing	Multiple systems delivered with consistent performance and manageable support requirements
Information protection	Verified migration across the intended services, including identity and software trust
Quantum communications	Evidence of a justified security benefit under a complete, tested trust arrangement
Sensors	Repeatable performance in the intended environment, with a credible cost and maintenance comparison
Defence use	Mission-relevant acceptance evidence, trained operators and established logistics
AI assistance	Measured improvements retained within the integrated physical system
Industrial capability	Demonstrated production, repair and supply continuity at the required scale

Open official record

The assessment remains limited to the public evidence cited. A public roadmap cannot establish undisclosed capability, and the absence of an open demonstration cannot prove that no classified activity exists. Equally, speculation about classified activity cannot be used to justify a procurement claim.

A political authority should therefore require the strongest evidence available within the relevant security setting, while ensuring that public statements remain consistent with what can actually be supported.

Source register

Retrieval date for every entry: 4 October 2026.

Source	Publication or document status	Why it matters in Pillar 0
NIST — First three finalised post-quantum standards	Announcement dated 13 August 2024	Establishes that concrete standards exist
NIST — FIPS 203: ML-KEM	Final standard, 13 August 2024	Defines the key-establishment method
NIST — FIPS 204: ML-DSA	Final standard, 13 August 2024	Defines a digital-signature method
NIST — FIPS 205: SLH-DSA	Final standard, 13 August 2024	Defines a second digital-signature approach
NIST — What Is Post-Quantum Cryptography?	Official explanatory resource	Explains early preparation and retained-data exposure
NIST — Post-quantum cryptography FAQs	Official technical guidance	Distinguishes different cryptographic risks
NIST — Quantum Computing Explained	Official explanatory resource	Corrects the “all answers at once” description
NCSC — PQC migration timelines	Guidance published in March 2025	Establishes planning and migration milestones
NSA — QKD and quantum cryptography	Official agency position; some explanatory passages retain older standardisation context	Identifies limitations relevant to national-security procurement
Nature — Quantum error correction below the surface code threshold	First published 9 December 2024; corrected in April 2026	Demonstrates progress in protected quantum memory
npj Quantum Information — Limitations of noisy quantum devices	Published 28 November 2025	Explains limits under specified noise assumptions
Nature — Reinforcement learning control of quantum error correction	Published 8 July 2026	Provides a measured example of AI-assisted control
Nature — Learning high-accuracy error decoding	Published 20 November 2024	Provides evidence on machine-learning interpretation of errors
IBM — Quantum hardware and roadmap	Current vendor roadmap, including a 2029 target	Provides a future milestone requiring verification
DARPA — Quantum Benchmarking Initiative	Current programme record	Defines evaluation of computational value against cost
Gidney — RSA-2048 factoring resource estimate	Preprint first submitted 21 May 2025	Shows why cryptographic relevance requires a complete resource model
Gidney — Full resource model	Technical text retrieved for assumptions and tables	Details logical operations, protection and physical costs
PTB — Realisation of the SI second	Official metrology resource	Establishes the existing public role of atomic timing
Earth System Science Data — Airborne gravimetry	Published 17 April 2025; measurements from 2023	Provides field evidence and a conventional comparison
Nature — Quantum sensing for gravity cartography	Published February 2022	Establishes a specific underground-structure experiment
Royal Navy — Quantum clock trials	Official announcement, 14 August 2026	Identifies a reported defence timing trial
Nature — Quantum memory nodes in a telecom network	Published May 2024	Establishes a defined quantum networking experiment
European Commission — Quantum Europe Strategy	Adopted 2 July 2025	Identifies industrial translation and coordination concerns
Japan Cabinet Office — Quantum ecosystem measures	Policy document dated 30 May 2025	Identifies component, supply-chain and industrial-development priorities

Final political assessment — analytical judgment. Governments can make useful decisions now by improving information protection, testing instruments against real public needs, supporting verifiable computing milestones and strengthening the capacity to operate and maintain technology. The central responsibility through 2031 is to connect each commitment with a measurable result, while preserving the ability to change course as the evidence develops.

---

Pillar I — Mechanisms, Hardware and Computational Evidence

Evidence cut-off and retrieval date: 4 October 2026. This instalment completes Chapters 1–4 together, following the structure in your supplied prompt and scheme. It develops the technical and computational evidence without reproducing the opening assessment’s migration timetable, national summaries or cryptanalytic resource table.

Classification: [D] demonstrated result, bounded by the experiment; [P] funded, contracted, mandated or standardised programme; [R] roadmap, prospective application or conditional projection; [C] attributed assertion lacking the primary technical evidence required for demonstration status; [H] misleading equivalence, unsupported generalisation or category error. Analytical recommendations below are identified as recommendations rather than presented as demonstrated capabilities.

Chapter 1 — Executive Adjudication

Principal judgment

The appropriate unit of assessment is a verified workload executed under a complete resource and error budget. Qubit numbers, isolated gate fidelities and encoded-state demonstrations each describe part of that capability, but they cannot individually establish useful, sustained computation. The reviewed record contains significant advances in physical processors, logical subroutines and hybrid scientific workflows; the decision problem is determining which missing elements separate each result from its proposed application. This judgment follows from the different experimental scopes of A 98-qubit trapped-ion quantum computer with all-to-all connectivity — Quantinuum and collaborators, Nature, Jun 2026, Experimental Demonstration of Logical Magic State Distillation — author manuscript of the Nature paper, published Jul 2025, and Chemistry Beyond the Scale of Exact Diagonalization on a Quantum-Centric Supercomputer — IBM, RIKEN and collaborators, Science Advances, Jun 2025. Nature

Five actionable findings

Finding	Evidence and bounded interpretation	Decision consequence
1. Buy verified computation rather than a qubit inventory.	[D] The 2026 Helios paper reports an integrated trapped-ion processor with measured component errors and circuit benchmarks. Those measurements support the processor’s documented performance; they do not convert its physical register into an equal-sized fault-tolerant logical register. A 98-qubit trapped-ion quantum computer with all-to-all connectivity — Nature, Jun 2026.	Require the proposed workload, compiled circuit, accepted-result rate, elapsed time, verification method and comparator before approving application procurement.
2. Treat universal fault tolerance as an integration problem.	[D] Neutral-atom logical magic-state distillation demonstrates a necessary resource-generation operation. Its importance lies in improving encoded resource states, rather than establishing an unrestricted production computer. Experimental Demonstration of Logical Magic State Distillation — Nature, Jul 2025; linked author manuscript.	Fund milestones that connect memory, gates, resource-state production, measurement and classical control within one sustained computation.
3. Require classical competition to remain open throughout a pilot.	[D] Subsequent classical work produced fast, converged simulations of the experiment originally presented as evidence of pre-fault-tolerant utility. A benchmark conclusion can change while the quantum hardware remains the same. Fast and converged classical simulations of evidence for the utility of quantum computing before fault tolerance — Caltech author repository, Science Advances, Jan 2024.	Establish an independent classical team and permit updated algorithms; a vendor-selected baseline should not determine the investment case.
4. Assess manufacturing through reproducibility and yield.	[D] Foundry-compatible silicon spin-qubit work reports high-fidelity operations across four characterised devices. This is evidence of repeatable unit-cell performance, not evidence of a large operational processor or production yield at that scale. Industry-compatible silicon spin-qubit unit cells exceeding 99% fidelity — Nature, Sep 2025.	Request distributions across devices, wafers and operating conditions, together with calibration effort, packaging losses and repair procedures.
5. Evaluate AI against a named engineering bottleneck.	[D] Reinforcement learning has been experimentally used to adjust control parameters during error correction. [R] Its extension to larger systems remains a separate scaling proposition. Reinforcement learning control of quantum error correction — Google Quantum AI, Google DeepMind and collaborators, Nature, Jul 2026.	Measure logical performance, drift tolerance, training cost and control latency with and without the intervention; an AI label should not substitute for these results.

The investment distinction

The recommended investment structure separates research option value, engineering capability and application value. A device may justify research funding because it tests a difficult physical mechanism. A control system may justify engineering funding because it improves reliable operation. An application contract requires additional evidence that the complete workflow produces a useful result under the buyer’s constraints. Moving between these categories should require a new acceptance test.

This distinction is particularly important for hybrid chemistry. [D] Quantum samples can contribute to a workflow whose other stages run on a classical supercomputer. The value question then concerns the incremental contribution of those samples, including the classical processing they require. Independent analysis of quantum-selected configuration interaction identifies sampling and compactness limitations that bear directly on this question. Critical Limitations in Quantum-Selected Configuration Interaction Methods — Reinholdt and collaborators, Journal of Chemical Theory and Computation, Jun 2025; linked author version. arxiv.org

Key judgments

The evidence supports differentiated research and engineering commitments. It supports application procurement only within the demonstrated workload boundary and an independently evaluated comparison. A single score combining physical qubits, fidelity, funding and commercial announcements would conceal the distinctions that matter to the decision.

What would change the assessment

A stronger application assessment would require reproducible, sustained execution of a named useful workload, with disclosed logical resources where applicable, complete elapsed time, output accuracy, unsuccessful attempts and contemporary classical competition. A stronger industrial assessment would require repeatable manufacturing and service performance beyond selected laboratory devices.

Open official record

The principal unresolved records are application-level execution logs, accepted-run statistics, calibration interruptions, manufacturing distributions and independently reproduced comparisons. The linked Helios paper, silicon unit-cell paper and chemistry methods critique provide evidence relevant to these questions without supplying every record required for operational acceptance.

Chapter 2 — The Taxonomy

Principal judgment

Quantum terminology becomes decision-useful only when it identifies a physical mechanism, a delivered function and a bounded assurance claim. The same word can otherwise refer to incompatible systems: classical cryptography designed to resist quantum attacks, optical key distribution, entanglement-assisted processing or a sensor exploiting quantum properties. NIST’s key-encapsulation standard and the experimental optical-network computing paper illustrate this difference directly. Module-Lattice-Based Key-Encapsulation Mechanism Standard — NIST, FIPS 203, Aug 2024; Distributed quantum computing across an optical network link — Nature, Feb 2025. CSRC

2.1 Mechanism dictionary and misleading expressions

Expression	Mechanism that must be identified	Defensible classification	Misleading interpretation to reject	Evidence required
“Quantum internet”	Entanglement distribution, quantum memories, interfaces, routing and the accompanying classical control.	[D] for a documented experiment; [R] for a proposed broader network.	[H] A quantum link automatically provides an operational internet, unrestricted quantum computing or instantaneous communication.	Entanglement rate and fidelity, memory lifetime, losses, failure handling and delivered protocol. Distributed quantum computing across an optical network link — Nature, Feb 2025.
“Quantum-safe VPN”	Named key establishment, authentication, signatures and symmetric protection within the actual protocol.	[P] for an applicable standard; deployment status needs separate evidence.	[H] Use of one quantum-resistant primitive proves the entire VPN implementation secure.	Complete cryptographic inventory, protocol composition and implementation assurance. Module-Lattice-Based Key-Encapsulation Mechanism Standard — NIST, Aug 2024.
“Unhackable quantum link”	Usually QKD equipment and key-management infrastructure.	[H] when asserted without qualification.	[H] Protection of key distribution removes endpoint, authentication, equipment or availability vulnerabilities.	Authentication design, implementation testing, relay trust and service continuity. Cryptograph Solutions are Limited Compared to Quantum-Resistant Algorithms — NSA, Oct 2026.
“Quantum encryption”	Specify whether the system distributes keys, encapsulates keys, signs messages or protects payloads.	Unresolved until decomposed.	[H] QKD, post-quantum cryptography and payload encryption are interchangeable.	Name the primitive and its exact role. Module-Lattice-Based Key-Encapsulation Mechanism Standard — NIST, Aug 2024.
“Quantum radar”	Quantum illumination or another explicitly specified sensing protocol.	[D] for the bounded laboratory result.	[H] Laboratory detection advantage establishes deployable long-range detection of operational targets.	Target model, losses, noise, receiver, temperature, range and equal-resource classical comparison. Demonstration of Quantum Advantage in Microwave Quantum Radar — Nature Physics, 2023; linked author version revised Mar 2023.
“AI-powered quantum computer”	Calibration, decoding, control, compilation or another named intervention.	[D] for a measured intervention; [R] for extrapolation.	[H] AI removes the need for physically reliable operations or error correction.	Ablation study, logical outcome, latency and behaviour under drift. Reinforcement learning control of quantum error correction — Nature, Jul 2026.
“Logical qubit”	An encoded information unit under a specified code and protocol.	[D] only within the demonstrated operations and correction conditions.	[H] Every encoded register supports the same gates, protection and computation duration.	Code parameters, syndrome procedure, correction, accepted fraction and logical errors. Logical quantum processor based on reconfigurable atom arrays — Nature, Dec 2023.
“Fault-tolerant”	A circuit or architecture designed to constrain error propagation under stated assumptions.	[D] for tested operations; [R] for an unbuilt complete architecture.	[H] One fault-tolerant subroutine establishes universal sustained computation.	Tested fault model, logical operations, repeated correction and integration. Fault-tolerant control of an error-corrected qubit — Nature, Oct 2021.
“Quantum advantage”	A specified quantum-versus-classical comparison.	[D] for experimental performance, with the comparison’s scope disclosed.	[H] A benchmark separation proves general commercial advantage.	Problem family, accuracy, resources, classical methods and comparison date. Observation of constructive interference at the edge of quantum ergodicity — Nature, Oct 2025.
“Topological qubit”	A specified physical encoding, readout and proposed protection mechanism.	[D] for demonstrated device behaviour; [C] for broader unsupported vendor assertions.	[H] Parity readout alone establishes protected universal computation.	Protection tests, coherent operations, error mechanisms and reproducibility. Interferometric single-shot parity measurement in InAs–Al hybrid devices — Nature, Feb 2025.

2.2 Three stacks with separate acceptance boundaries

The proposed classification below prevents evidence from one stack from being used to certify another.

Stack	Delivered function	Recommended acceptance boundary	Evidence that cannot substitute for acceptance
A — Computing	A sampled output, observable, energy estimate, optimisation result or other defined computation.	Output accuracy, accepted-result rate, complete runtime, resource accounting and classical comparison.	Security standards, a sensor trial or an announced processor capacity.
B — Communications and cryptography	Key establishment, authentication, protected traffic or a specified quantum-network protocol.	Threat model, primitive composition, endpoint assurance, relay assumptions and availability.	A difficult sampling experiment or an encoded logical state.
C — Sensing, timing and metrology	A measured physical quantity or mission-relevant estimate.	Sensitivity, bandwidth, bias, drift, environmental tolerance, duration and system integration.	A computing roadmap or an optical key-distribution demonstration.

[D] The radar experiment is especially instructive: its author record describes a microwave quantum-illumination demonstration performed inside a dilution refrigerator and identifies restricted operating parameters and low-temperature requirements. Its evidentiary value concerns the mechanism and the experimental detection task. [R] Operational radar capability would require additional evidence about propagation, target conditions, receiver integration and environmental operation. Demonstration of Quantum Advantage in Microwave Quantum Radar — Nature Physics, 2023; author version, Mar 2023. arxiv.org

2.3 Two readiness axes

The following is an analytical assessment framework, not an official universal quantum-readiness scale.

Technical state	What it establishes	Institutional state	What it establishes
Laboratory mechanism	The effect exists under documented experimental conditions.	Research activity	An institution is investigating it.
Characterised component	A device has measured interfaces and performance.	Funded programme	Resources have been committed to a specified programme.
Integrated prototype	Several components execute the intended function together.	Validated requirement	A user has defined the need and acceptance criteria.
Field trial	The system has been tested in a relevant operating environment.	Procurement	A buyer has contracted for specified equipment or services.
Repeatable production	Manufacturing and service performance are reproducible.	Accreditation or acceptance	The system has passed the relevant institutional assessment.
Operational qualification	The complete system meets its defined mission envelope.	Doctrine or routine use	It has an established role in operational practice.

A procurement announcement should therefore be recorded on the institutional axis at the level its documents justify. It should not advance the technical axis by implication. Conversely, a technically persuasive experiment can remain a research result without an operational customer or qualification record.

2.4 Metrics that should remain distinct

Metric	Correct interpretation	Additional information needed
Gate fidelity	Closeness of an implemented operation to its target under a specified characterisation method.	Method, uncertainty, parallel operation, drift and leakage.
SPAM performance	State-preparation-and-measurement performance.	Whether preparation and measurement errors are separated.
Logical error per cycle	Encoded performance during the stated correction cycle.	Cycle duration, code, decoder and error correlations.
Postselected fidelity	Quality of the accepted subset.	Acceptance probability and cost of rejected attempts.
Memory lifetime	Preservation of information under the tested memory protocol.	Logical gates and computation duration are separate questions.
Circuit throughput	Completed circuits or outputs over time.	Circuit size, accuracy, calibration and rejected runs.
Advantage ratio	Relative performance for the specified comparison.	Classical methods, hardware, precision and resource equality.

These distinctions have direct experimental consequences. [D] The superconducting magic-state study reports different infidelities for feedforward and postselected procedures, while also measuring their yields; collapsing those results into one fidelity would discard information about usable output production. Encoding a magic state with beyond break-even fidelity — IBM and collaborators, Nature, Jan 2024. Nature

Key judgments

Terminology should identify the mechanism before describing benefit. Readiness should be recorded separately for technical performance and institutional adoption. Metrics should retain their experimental conditions, particularly where postselection, error mitigation or model-based comparison affects interpretation.

What would change the assessment

A term becomes suitable for procurement when its supplier supplies a mechanism-specific specification, a relevant test protocol and an assurance boundary. Greater readiness requires evidence at the next level on the appropriate axis.

Open official record

For security products, request the protocol and assurance documents supporting the advertised property, using NIST FIPS 203 and the NSA assessment of specialised cryptographic solutions as mechanism-specific references. For computing and sensing, request experimental conditions and complete accepted-output records.

Chapter 3 — Quantum Computing: Hardware Truth

Principal judgment

The six hardware families have demonstrated different capabilities and face different integration burdens; the reviewed evidence does not justify a single ranking based on qubit count. A useful comparison must retain the device’s physical register, demonstrated encoding, operation quality, accepted-run conditions and scaling assumptions. The examples below are selected evidence points, not an exhaustive inventory or a claim that these are each platform’s global maximum.

3.1 Comparative hardware evidence

Platform	Demonstrated evidence	Quantitative detail	Logical or correction boundary	Source
Superconducting	[D] Error-suppressed preparation of an encoded magic resource on IBM hardware.	Four data qubits encode two logical qubits; reported logical infidelity 1.87 ± 0.16% with feedforward and 1.23 ± 0.11% with postselection.	Resource-state preparation; an accepted state’s quality and production yield must remain separate.	Encoding a magic state with beyond break-even fidelity — Nature, Jan 2024.
Trapped ion	[D] Integrated Helios processor with transport-based connectivity.	98 physical qubits; mean infidelities: 2.5(1) × 10⁻⁵ for one-qubit gates, 7.9(2) × 10⁻⁴ for two-qubit gates and 3.3(5) × 10⁻⁴ for SPAM.	Physical processor and circuit benchmarks; these counts do not describe 98 error-corrected logical qubits.	A 98-qubit trapped-ion quantum computer with all-to-all connectivity — Nature, Jun 2026.
Neutral atom	[D] Reconfigurable encoded processor executing logical circuits.	Up to 280 physical atoms and 48 logical qubits; circuits include 228 logical two-qubit gates and 48 logical CCZ gates.	The highlighted large logical register uses an [[8,3,2]] encoding; error detection and postselection must be distinguished from unrestricted repeated correction.	Logical quantum processor based on reconfigurable atom arrays — Nature, Dec 2023.
Photonic	[D] Aurora modular photonic architecture and real-time decoding demonstration.	35 chips, 84 squeezers, 36 photon-number-resolving detectors, furnishing 12 physical qubit modes per clock cycle.	A distance-2 repetition code; temporal cluster-state size is not a simultaneous logical register.	Scaling and networking a modular photonic quantum computer — Xanadu, Nature, Jan 2025; PMC full text.
Silicon spin	[D] Foundry-compatible two-qubit unit cells characterised across four devices.	Controlled-Z fidelities range from 99.04(16)% to 99.56(6)%; paper reports 95% confidence for the characterisation error bars.	Physical operations and manufacturing compatibility; no corresponding large logical register is established by these devices.	Industry-compatible silicon spin-qubit unit cells exceeding 99% fidelity — Nature, Sep 2025.
Topological approaches	[D] Single-shot parity measurement in hybrid devices; separate minimal-chain readout experiments.	In the InAs–Al work, optimal readout gives an assignment-error probability of approximately 1%.	The authors explicitly retain ambiguity between the targeted topological interpretation and certain trivial low-energy states.	Interferometric single-shot parity measurement in InAs–Al hybrid devices — Nature, Feb 2025.

Comparison rule: the entries use different operations, characterisation protocols and correction conditions. Their percentages should not be treated as measurements of the same system-level quantity. Nature

3.2 Superconducting hardware: resource-state production and control

[D] The encoded magic-state experiment illustrates a practical trade-off. Feedforward changes later operations according to intermediate measurements, increasing useful-state production relative to a more restrictive postselection procedure. It also introduces additional control-related waiting during which errors can accumulate. The reported advantage therefore involves both output quality and output yield. Encoding a magic state with beyond break-even fidelity — Nature, Jan 2024.

The recommended procurement quantity is consequently accepted resource states per unit time at the required infidelity, including control delays and rejected attempts. A supplier reporting only the best accepted-state fidelity leaves the buyer unable to determine whether an algorithm’s resource demand can be met.

[D] The July 2026 reinforcement-learning experiment addresses another operational issue: maintaining calibration during error correction. It uses error-detection events as a learning signal for continuous control adjustment. The experiment reports 3.5-fold greater logical stability against injected drift. The larger-code scaling evidence described in the paper is numerical rather than a hardware demonstration at those larger sizes. Reinforcement learning control of quantum error correction — Nature, Jul 2026. Nature

Control question	Recommended evidence	Decision relevance
Does calibration interrupt execution?	Time-stamped interruption and recovery logs.	Long computations require continuity.
Does performance persist during drift?	Defined drift tests and naturally occurring drift records.	A short stable interval may not represent sustained operation.
Does learning improve logical outcomes?	Comparison with the same hardware and fixed control.	Physical calibration scores may conceal logical consequences.
What is the learning overhead?	Training samples, compute resources and elapsed time.	Improvement must be assessed with its operating cost.
What happens when adaptation fails?	Detection, rollback and run-rejection procedures.	Failed adaptation must not silently invalidate results.

3.3 Trapped ions: connectivity has a schedule

[D] Helios implements connectivity through physical transport between memory and operating regions. Its paper describes parallel operations, shared optical resources and real-time compilation of dynamic programmes. The entangling-gate operation itself takes approximately 70 microseconds, but a complete programme also involves transport, cooling and other operations. Gate duration alone therefore cannot establish application runtime. A 98-qubit trapped-ion quantum computer with all-to-all connectivity — Nature, Jun 2026.

The architectural question is how effectively movement, cooling and gates overlap under an actual compiled workload. Connectivity may reduce the number of routing gates required by an algorithm, while transport introduces its own scheduling demands. The relevant comparison should measure both effects within the same programme.

[D] Earlier trapped-ion work demonstrated fault-tolerant control of one Bacon–Shor encoded qubit using 13 physical ions. It reported average logical preparation-and-measurement error of 0.6% and Clifford-gate error of 0.3% after offline correction. Those results establish bounded logical operations; the offline qualifier matters when assessing real-time integration. Fault-tolerant control of an error-corrected qubit — Egan and collaborators, Nature, Oct 2021. Nature

Architecture issue	Recommended test
Transport-related performance	Error and duration distributions for the programme’s actual movement operations.
Parallel optical control	Gate characterisation under simultaneous execution.
Cooling overhead	Full compiled schedule including cooling and waiting.
Measurement and feedforward	Latency distribution and its effect on encoded performance.
Modular expansion	Measured interface performance and complete distributed operations.

3.4 Neutral atoms: logical registers and non-Clifford resources

[D] The reconfigurable-array experiment establishes that many encoded qubits can participate in structured logical circuits. Its large-register demonstration combines encoding, logical operations and error-detection-based selection. The retained runs should therefore be evaluated with their acceptance fraction rather than interpreted as an unconditional corrected computation. Logical quantum processor based on reconfigurable atom arrays — Nature, Dec 2023.

[D] Subsequent logical magic-state distillation demonstrated improved output-state fidelity relative to input logical states using distance-3 and distance-5 colour codes. This is a distinct capability: producing resources needed for universal encoded operations. It strengthens the technical pathway without establishing the throughput, integration and sustained runtime required by an arbitrary large algorithm. Experimental Demonstration of Logical Magic State Distillation — Nature, Jul 2025; author manuscript. arxiv.org

Stage	Function	Evidence required before advancing
Encoded-state preparation	Creates the logical input.	Preparation error and acceptance statistics.
Repeated syndrome extraction	Supplies information about accumulating errors.	Performance over the required number of cycles.
Logical entangling operations	Connects encoded registers.	Errors during gates, movement and measurement.
Resource-state preparation	Supplies non-Clifford resources.	Input quality and production rate.
Distillation	Improves resource-state quality.	Output error, accepted yield and consumed resources.
Integrated computation	Uses these elements together.	Complete workload execution with a total error budget.

This proposed stage structure prevents the presence of one component from being used to certify the others. The next decisive experiment would connect the components under the same control and operating conditions while retaining the complete accepted-run record.

3.5 Photonics: temporal scale, losses and useful outcomes

[D] Aurora generated a cluster state described as containing 86.4 billion modes through temporal operation. The paper simultaneously identifies 12 physical qubit modes per clock cycle and calls the machine a scale model with insufficient component performance for the intended full architecture. The cumulative mode count therefore describes the generated stream, rather than billions of simultaneously available, protected logical qubits. Scaling and networking a modular photonic quantum computer — Nature, Jan 2025. PMC

A useful photonic comparison must report losses alongside fidelity. [C — author preprint] The April 2024 manufacturable-platform manuscript reports high-fidelity component operations, including two-qubit fusion and chip-to-chip transfer, while explicitly excluding loss from the quoted interconnect fidelity. Those numbers support the attributed component claim, not an independently certified operational computer. A manufacturable platform for photonic quantum computing — Alexander and collaborators, author preprint, Apr 2024. arxiv.org

Quantity	Recommended reporting condition
Source quality	State quality and useful-event probability.
Transmission	Loss through the complete path, including interfaces.
Fusion or entangling operation	Conditional fidelity and success probability.
Detector performance	Efficiency, errors, timing and recovery characteristics.
Switching and delay	Loss, latency and reliability during the actual schedule.
Logical processing	Accepted encoded outputs per unit time at the required error.

For procurement, a conditional fidelity should always be accompanied by the probability and cost of obtaining the condition. Otherwise a component can appear excellent while delivering too few useful events for the proposed architecture.

3.6 Silicon: three different scaling questions

Silicon evidence should distinguish foundry-compatible electrostatic devices, precision donor processors and mobile-spin interconnection. Their results address different engineering constraints.

Approach	Demonstrated result	Quantitative boundary	Interpretation
Foundry-compatible unit cells	[D] Characterised high-fidelity two-qubit operations on four devices.	Controlled-Z range 99.04(16)%–99.56(6)%; SPAM exceeds 99.9% on three devices, with 99.33(9)% on the fourth.	Reproducibility evidence at unit-cell scale. Industry-compatible silicon spin-qubit unit cells exceeding 99% fidelity — Nature, Sep 2025.
Donor-based processor	[D] An 11-qubit processor using coupled registers.	Reported operation fidelities 99.10%–99.99%, Bell-state fidelity up to 99.5% and an eight-nuclear-spin GHZ state.	A physical processor and entanglement result, rather than 11 protected logical qubits. An 11-qubit atom processor in silicon — Nature, Dec 2025.
Mobile spins	[D] Two-qubit logic combined with movement and teleportation.	Approximately 99% two-qubit gate fidelity; 87% conditional, postselected teleportation fidelity over 320 nanometres.	Movement and conditional transfer have separate quality measures. Two-qubit logic and teleportation with mobile spin qubits in silicon — Nature, May 2026.

The recommended industrial assessment should therefore ask three separate questions: can the unit cell be fabricated reproducibly; can many unit cells be controlled together; and can the complete array execute the required encoded operations? Compatibility with an established manufacturing process materially informs the first question, but does not settle the latter two.

The mobile-spin result also illustrates why a distance measurement needs its physical scale and conditions. Nanometre-scale movement within a device is relevant to processor layout. It should not be described as evidence of a deployable communications network.

3.7 Topological approaches: measurement, protection and vendor claims

[D] The InAs–Al parity paper demonstrates an important measurement capability. Its authors explicitly state that this measurement by itself does not uniquely distinguish Majorana zero modes in a topological phase from certain fine-tuned low-energy Andreev states in a trivial phase. That interpretive boundary should remain attached to the result. Interferometric single-shot parity measurement in InAs–Al hybrid devices — Nature, Feb 2025.

[D] The 2026 minimal-Kitaev-chain paper reports real-time parity readout with switching lifetimes exceeding one millisecond. It identifies the two-site modes as offering limited protection compared with longer chains. The result advances readout and device control while leaving protection and scalable computation as separate evidentiary questions. Single-shot parity readout of a minimal Kitaev chain — Nature, Feb 2026. Nature

[C] Microsoft’s June 2026 Majorana 2 announcement describes improved device reliability and longer-lived states. [R] Its timetable for a scalable future computer remains a company objective. The announcement should be recorded as a first-party claim unless primary technical evidence establishes the relevant operations and protection. Introducing Majorana 2 — Microsoft, Jun 2026. news.microsoft.com

Claim level	Recommended discriminating evidence
Parity readout	Single-shot errors, lifetimes and measurement back-action.
Topological interpretation	Tests that constrain competing explanations.
Coherent encoded control	Reproducible preparation and operations on the encoding.
Protection	Error behaviour under controlled perturbations and scaling.
Logical processing	A specified logical gate set and error accounting.
Universal sustained computation	Integrated encoded execution at the proposed workload scale.

This framework permits recognition of real experimental progress while avoiding an unsupported promotion from a readout result to an operational computer.

3.8 Error correction: overhead depends on architecture and assumptions

[R — model-based architectural projection] The bivariate-bicycle-code study estimates preservation of 12 logical qubits using 288 physical qubits, under a stated physical error rate of 0.1%, for nearly one million syndrome cycles. Its comparable surface-code estimate requires nearly 3,000 physical qubits. The result is theoretical and numerical, rather than a hardware demonstration of that memory. High-threshold and low-overhead fault-tolerant quantum memory — Bravyi and collaborators, Nature, Mar 2024. Nature

The comparison shows why a generic “physical qubits per logical qubit” conversion is inadequate. The projected saving depends on the code, noise assumptions, connectivity and correction circuit. Its system value must also account for logical gates and resources beyond memory.

Resource-estimation field	Recommended disclosure
Error-correcting code	Code family, parameters and layout.
Physical noise	Gate, measurement, preparation, idle and correlated errors.
Connectivity	Required interactions and their implementation.
Syndrome cycle	Duration and constituent operations.
Decoder	Algorithm, latency, resources and failure behaviour.
Logical operations	Gate-specific errors and durations.
Non-Clifford resources	Production quality, rate and factory footprint.
Complete workload	Logical register, depth, repetitions and success target.

3.9 Manufacturing and supply-chain acceptance

[C — supplier-reported delivery] Oxford Instruments reported installation of three dilution refrigerators at the UK National Quantum Computing Centre in December 2024. This establishes an attributed enabling-equipment delivery; three refrigerators should not be recorded as three operational quantum computers. Oxford Instruments NanoScience installs dilution refrigerators in the NQCC’s purpose-built research labs — Oxford Instruments, Dec 2024. Oxford Instruments

The proposed diligence matrix below identifies records to collect rather than asserting market shares, national independence or undisclosed supplier concentration.

Dependency	Records to request	Acceptance question
Cryogenic equipment	Cooling capacity, load curves, service intervals and recovery time.	Does performance persist with the complete installed system?
Lasers and optics	Stability, lifetime, alignment effort, spares and replacement qualification.	Can control performance be reproduced after maintenance?
Silicon and fabrication	Material specification, process variation, device distributions and wafer evidence.	How representative are the characterised devices?
Photonic packaging	Interface loss, assembly yield and environmental testing.	Does packaging retain component performance?
Classical electronics	Channel capacity, timing, latency, firmware and failover.	Can the controller support the required correction cycle?
Calibration software	Procedures, access rights, portability and recovery documentation.	Can the buyer operate and diagnose the system?
Detectors and readout	Efficiency, errors, throughput and operating requirements.	Is useful-result production limited by measurement?
Service and replacement	Repair times, qualified parts and requalification tests.	Can capability be restored within the buyer’s requirement?

Key judgments

Hardware assessment requires a complete architecture, not a modality label. Logical subroutines are substantial technical achievements, but their integration and throughput determine application feasibility. Manufacturing evidence should describe distributions and reproducibility; supply-chain evidence should describe operational dependencies and recoverability.

What would change the assessment

The strongest upgrade would be repeated, sustained encoded computation with a universal operation set, measured resource-state throughput, real-time correction and demonstrated error performance across the required duration. Industrial readiness would additionally require comparable performance across production units and maintenance cycles.

Open official record

The records most likely to alter the assessment are full error distributions, parallel-operation tests, logical execution logs, accepted yields, decoder latency, fabrication distributions and service evidence. The low-overhead memory study identifies architectural opportunities; the silicon reproducibility study supplies bounded manufacturing evidence; the minimal-chain readout paper defines a narrower demonstrated topological-device milestone.

Chapter 4 — What Quantum Computers Can and Cannot Do in 2026

Principal judgment

The reviewed record supports specified experiments and selected hybrid scientific workflows; it does not establish general superiority across chemistry, optimisation, database search or cryptanalysis. A defensible application claim must identify the mathematical task, its effective size, required accuracy, complete workflow and contemporary classical competition.

4.1 Application evidence matrix

Application	Evidence class and demonstrated scope	Quantitative anchor	What the evidence does not establish	Appropriate evaluation
Electronic structure	[D] Quantum sampling within a hybrid workflow using Heron and Fugaku.	Circuits up to 77 qubits and 10,570 gates.	General superiority over the strongest approximate classical chemistry methods.	Compare energy accuracy and full workflow resources. Chemistry Beyond the Scale of Exact Diagonalization on a Quantum-Centric Supercomputer — Science Advances, Jun 2025.
Many-body simulation	[D] Krylov methods demonstrated on specified spin-model sectors.	Up to 56 sites, including restricted particle-number sectors.	Arbitrary simulation of the full 56-qubit state space.	Report effective sector and projected-subspace dimensions. Krylov diagonalization of large many-body Hamiltonians on a quantum processor — Nature Communications, Jun 2025.
Quantum dynamics	[D] Higher-order correlation measurements and Hamiltonian-learning example.	Reported 65-qubit circuits in a specified experimental regime.	General commercial chemistry, optimisation or materials advantage.	Evaluate the scientific observable and current simulation methods. Observation of constructive interference at the edge of quantum ergodicity — Nature, Oct 2025.
Approximate optimisation	[D] Annealing comparison on a defined spin-glass ensemble.	More than 1,300 error-suppressed logical variables, with a specified classical comparator and approximation criterion.	Universal advantage for logistics, finance or every NP-hard problem.	Preserve instance distribution, embedding and solution quality. Scaling Advantage in Approximate Optimization with Quantum Annealing — Physical Review Letters, Apr 2025.
Sampling	[D] Execution of defined random-circuit benchmarks.	Platform-specific register and circuit parameters.	Automatic usefulness of the sampled distribution to a customer.	Define the downstream use and independently test comparison claims. A 98-qubit trapped-ion quantum computer with all-to-all connectivity — Nature, Jun 2026.
Search	[D] Small distributed Grover demonstration; [R] useful-scale execution.	Reported distributed-search success of approximately 71%.	A speedup for an ordinary database once oracle construction and data access are included.	Account for the complete oracle and workflow. Distributed quantum computing across an optical network link — Nature, Feb 2025.
Factoring and discrete logarithms	Established algorithmic basis; [R] relevant-scale physical execution.	Resources depend on the named cryptographic target and architecture.	A physical-qubit headline proves an operational attack.	Require complete logical circuit, correction, runtime and success accounting. Polynomial-Time Algorithms for Prime Factorization and Discrete Logarithms on a Quantum Computer — Shor, author manuscript, Aug 1995.

4.2 Chemistry: what “beyond exact diagonalisation” means

[D] The Heron–Fugaku study prepares quantum samples and uses distributed classical processing to obtain ground-state energy upper bounds and sparse wavefunction approximations for nitrogen and iron–sulphur systems. Its maximum circuit size is a useful scale marker, but the relevant computational object is the complete hybrid workflow. Chemistry Beyond the Scale of Exact Diagonalization on a Quantum-Centric Supercomputer — Science Advances, Jun 2025.

The phrase “beyond exact diagonalisation” identifies a particular comparison boundary. It should not be expanded into “beyond classical chemistry,” because approximate methods are part of the relevant competition. The application judgment should ask whether the quantum contribution improves a useful accuracy–cost trade-off against those methods.

[D — published numerical analysis] Reinholdt and collaborators identify a tension in quantum-selected configuration interaction: repeated sampling can revisit already selected configurations, while avoiding that problem can produce less compact expansions than classical selection heuristics. Their analysis challenges the practical advantage of the method in the examined cases without invalidating every possible future hybrid chemistry approach. Critical Limitations in Quantum-Selected Configuration Interaction Methods — Journal of Chemical Theory and Computation, Jun 2025. arxiv.org

Chemistry evaluation field	Recommended specification	Why it matters
Molecular system	Geometry, charge and electronic state.	Defines the physical question.
Model	Basis set, active space and Hamiltonian.	Defines the computational approximation.
Target quantity	Energy, difference, barrier or other observable.	Determines whether the result is useful.
Accuracy	Numerical tolerance and uncertainty.	Prevents comparison at unequal quality.
Quantum stage	State preparation, circuit and accepted samples.	Identifies the processor’s contribution.
Classical stage	Selection, recovery, diagonalisation and other processing.	Captures hybrid cost.
Comparator	Appropriate contemporary classical methods.	Tests incremental value.
Total execution	Calibration, sampling, retries and postprocessing.	Supports a practical time comparison.
Validation	Independent reference or defensible error bound.	Prevents an internally consistent answer from being mistaken for an accurate one.

Recommended decision rule: retain chemistry pilots when they produce new scientific information or demonstrate a credible route to a better accuracy–cost trade-off. Do not justify production acquisition solely through active-space size.

4.3 Materials and many-body simulation: count the effective problem

[D] The Krylov study demonstrates calculations in specified particle-number sectors, including cases with 56 sites and one particle, 44 sites and three particles, and 42 sites and five particles. These restrictions materially affect the mathematical size. Krylov diagonalization of large many-body Hamiltonians on a quantum processor — Nature Communications, Jun 2025. Nature Communications

These calculations do not dismiss the experiment. They identify the correct object against which classical difficulty should be assessed. The evaluation should additionally report the Krylov subspace actually constructed, measured observables and accuracy.

A materials application requires another bridge: the simulated Hamiltonian and observable must correspond to a useful materials question. A technically difficult spin-model experiment can have scientific value without yet establishing an industrial materials-design capability.

4.4 Quantum dynamics: difficult observables need bounded claims

[D] The October 2025 study measures second-order out-of-time-order correlators through repeated time-reversal protocols on superconducting hardware. It links the measured interference structure to classical simulation difficulty and includes a Hamiltonian-learning example. This is stronger evidence than a generic claim that “the quantum computer performed a large calculation,” because it identifies the observable and physical mechanism. Observation of constructive interference at the edge of quantum ergodicity — Google Quantum AI and collaborators, Nature, Oct 2025. Nature

The associated application assessment should remain tied to that observable. Extending it to drug discovery, aircraft design or general optimisation would require separate algorithms, error budgets and comparisons. The existence of a hard-to-simulate experiment does not supply those missing demonstrations.

4.5 Optimisation: ensemble, approximation and embedding

[D] The annealing study reports a scaling comparison for approximate optimisation on a defined spin-glass problem family, using quantum annealing correction and a named classical method. Its “logical” variables are error-suppressed annealing variables; they should not be equated with universal fault-tolerant circuit-model qubits. Scaling Advantage in Approximate Optimization with Quantum Annealing — Physical Review Letters, Apr 2025. Phys. Rev. Lett.

Question	Recommended acceptance requirement
Which problems were tested?	Full instance-generation procedure and held-out instances.
What solution quality was required?	Objective gap, feasibility and success probability.
How was the problem represented?	Variables, interactions, penalties and precision.
What did embedding consume?	Physical resources, preprocessing and discarded instances.
Which classical methods competed?	Appropriate tuned methods, with disclosed resources.
What time was measured?	Complete time to the required valid solution.
Does the result transfer?	New evidence on the buyer’s actual problem distribution.

For a logistics or financial pilot, the recommended test should include all operational constraints and the existing production solver. A favourable result on another ensemble should motivate that test, rather than predetermine its outcome.

4.6 Sampling: computational separation and application value

[D] Random-circuit sampling provides evidence about execution of complex quantum circuits and comparison with classical simulation. Its application value depends on an additional question: what useful output does the sampled distribution enable? The Helios paper’s circuit evidence should be evaluated within its stated benchmark before extending it to customer workloads. A 98-qubit trapped-ion quantum computer with all-to-all connectivity — Nature, Jun 2026.

The recommended assessment separates:

Question	Record needed
Was the intended distribution generated?	Verification protocol, uncertainty and device data.
What classical comparison supports the claim?	Methods, resources, precision and comparison date.
What useful task consumes the samples?	Named downstream algorithm and measurable benefit.

The first two can establish an important experimental result. The third establishes the basis for an application investment.

4.7 Why classical competition must remain current

[D] The 2023 IBM utility paper studied a 127-qubit processor and compared error-mitigated results with specified classical approaches. Subsequent published classical work provided fast, converged simulations for the experiment. The original quantum measurements remain experimental evidence, while the interpretation of their comparative difficulty changes. Evidence for the utility of quantum computing before fault tolerance — Nature, Jun 2023; Fast and converged classical simulations of evidence for the utility of quantum computing before fault tolerance — Science Advances, Jan 2024; Caltech repository. Nature

Recommended benchmark governance should therefore freeze the problem definition and output tolerance, while allowing the classical methods to improve. A procurement case that depends on prohibiting improved competition is not a durable application case.

4.8 Search: the oracle is part of the computation

The established Grover algorithm gives a quadratic query improvement for its defined search model. [R] Whether that produces a practical advantage depends on constructing and executing the oracle, supplying the data and accounting for repetitions and errors. A fast quantum mechanical algorithm for database search — Grover, author manuscript, May 1996. arxiv.org

[D] The distributed ion experiment demonstrates a remote controlled-Z gate with fidelity 86.2 ± 0.9% and a small Grover search with approximately 71% success. These results establish a distributed-processing experiment. They do not establish acceleration of a large ordinary database. Distributed quantum computing across an optical network link — Nature, Feb 2025.

Search proposal field	Recommended disclosure
Search space	Number and representation of candidates.
Predicate	Exact test defining an acceptable answer.
Oracle	Compiled circuit, depth and ancillary resources.
Data access	Loading, updates and access assumptions.
Repetitions	Success amplification and failed attempts.
Classical comparison	Complete incumbent workflow.

4.9 Factoring and discrete logarithms: relevance requires complete resources

Shor’s algorithm supplies the algorithmic basis for quantum factoring and discrete logarithms. [R] Relevant-scale physical execution requires a sufficiently reliable architecture and complete resource accounting; the mathematical algorithm alone does not establish a fielded attack. Polynomial-Time Algorithms for Prime Factorization and Discrete Logarithms on a Quantum Computer — Shor, Aug 1995 author manuscript. arxiv.org

The cryptanalytic estimates from the opening assessment are not repeated here. Chapter 5 will assess the named schemes and complete attack-resource assumptions. For this chapter, the application acceptance boundary is:

Required field	Why it is indispensable
Named cryptographic target	Defines the mathematical problem and parameter size.
Logical register	Separates algorithm data from physical inventory.
T/Toffoli or equivalent resources	Captures non-Clifford demand.
Circuit depth and dependencies	Constrains execution time.
Code and distance	Defines protection assumptions.
Physical overhead	Includes data, ancillas, factories and routing.
Cycle and decoder timing	Determines feasible sustained execution.
Success probability	Accounts for failure and repetition.
Complete runtime	Connects the architecture to operational relevance.

4.10 Quantitative illustration: errors accumulate

The following is an illustrative calculation, not measured hardware data or a forecast. Assume independent operations, identical failure probability \(p\), no correction and a requirement that all \(G\) operations succeed:\[ P_{\text{no failure}}=(1-p)^G. \]

Assumed failure probability per operation	Operations	Probability of no failure
1%	100	36.6%
1%	1,000	0.00432%
0.1%	1,000	36.8%
0.1%	10,000	0.00452%
0.01%	10,000	36.8%
0.01%	100,000	0.00454%

The calculation shows why a high isolated fidelity does not automatically support a deep successful computation. Real devices require richer error models, including correlations, leakage and unequal operations. Error correction changes the analysis; its purpose is to make logical failure compatible with the total workload. Architecture-dependent memory estimates provide a concrete example of that more complete reasoning. High-threshold and low-overhead fault-tolerant quantum memory — Nature, Mar 2024.

4.11 Application claims that should fail the current evidence gate

These are assessment rules for unsupported proposals, rather than statements that the applications can never become feasible.

Proposed claim	Classification when unsupported	Evidence needed for reconsideration
“The qubit count proves useful chemistry advantage.”	[H]	Full molecular model, target accuracy and classical comparison. Chemistry workflow; methods critique.
“A spin-glass result proves faster operational logistics.”	[H]	Tests on representative logistics instances with all constraints and preprocessing. Annealing study.
“A difficult sampling benchmark proves commercial value.”	[H]	A downstream task whose benefit depends on those samples. Helios benchmark paper.
“An encoded register is a universal fault-tolerant computer.”	[H]	Integrated universal operations, repeated correction, throughput and runtime. Logical processor; logical distillation.
“AI control removes hardware reliability requirements.”	[H]	Sustained logical evidence under the actual physical operating conditions. Reinforcement-learning control.
“Parity measurement establishes scalable protected computation.”	[H]	Discriminating physical evidence, protection tests and logical execution. Parity-measurement paper.
“A processor announcement establishes a cryptanalytic capability.”	[H]	Named target, complete logical resources and demonstrated or fully specified execution architecture. Algorithmic basis.

Key judgments

The most defensible computing assessments concern named scientific observables, defined optimisation ensembles and documented hybrid workflows. Problem size should describe the effective mathematical task, rather than only the physical register. Application advantage requires a complete, current comparison, and its validity should remain open to improved classical methods.

What would change the assessment

The decisive evidence would be independent reproduction of a useful workload at the buyer’s required accuracy, with complete elapsed time, accepted-run statistics, classical processing and contemporary comparators. For fault-tolerant applications, it would additionally require the relevant logical operations and resource production to remain reliable throughout execution.

Open official record

The missing records with greatest decision value are full workflow timings, rejected-run costs, effective problem dimensions, independent comparator results and complete logical execution data. The hybrid chemistry study, Krylov study, annealing comparison and published classical re-evaluation provide the starting evidence for those assessments.

---

Pillar II — Security Migration and Operational Applications

Chapter 5 — Cryptography

5.1. The security decision concerns an entire trust architecture

[P] The principal deployable response to the quantum threat is now anchored in published cryptographic standards: ML-KEM for key encapsulation, ML-DSA for digital signatures and SLH-DSA for stateless hash-based signatures. Their publication establishes interoperable algorithm specifications; it does not certify every product implementing them, migrate an organisation’s certificate hierarchy or establish resistance to implementation attacks.

Sources: NIST, FIPS 203: Module-Lattice-Based Key-Encapsulation Mechanism Standard, FIPS 204: Module-Lattice-Based Digital Signature Standard, and FIPS 205: Stateless Hash-Based Digital Signature Standard, August 2024. CSRC

The recommended assessment unit is therefore the security function and its dependencies. For each function, the owner should identify the initiating endpoint, terminating endpoint, intermediate trust boundaries, algorithm, implementation, replacement authority and expected lifetime.

Security function	Question the migration programme should answer	Required completion evidence
Session establishment	Can both endpoints establish a session using the authorised replacement mechanism?	Negotiation traces, interoperability results and explicit downgrade tests
Endpoint authentication	Does the session authenticate the intended peer through a suitable trust chain?	Certificate-path tests, revocation behaviour and identity-binding verification
Bulk confidentiality	Are payload keys generated, stored, rotated and destroyed appropriately?	Key-management design and operational audit
Firmware acceptance	Can an installed device verify new signatures throughout its remaining service life?	Tests on deployed bootloaders, recovery paths and update packages
Software distribution	Can every build, signing, distribution and installation stage process the new scheme?	End-to-end release rehearsal
Archival authenticity	Can a future verifier establish what was signed, when and under which trust state?	Documented preservation, timestamp and renewal procedures
Recovery and continuity	What happens when a peer, appliance or certificate is incompatible?	Tested recovery procedures and controlled exception handling
Supplier assurance	Who maintains the implementation after initial integration?	Versioned support commitments and replacement procedures

Recommended interpretation: migration is complete only when the relevant security service works across its actual operational boundary. An algorithm appearing in a product specification should count as a component milestone, not as evidence that the surrounding service has completed its transition.

5.2. ML-KEM parameters: exact sizes and what they mean

[P] FIPS 203 specifies the following encoded sizes. These are cryptographic objects, not complete TLS handshakes, certificate chains, VPN negotiations or application messages.

Parameter set	NIST security category	Encapsulation key	Decapsulation key	Ciphertext	Shared secret
ML-KEM-512	1	800 bytes	1,632 bytes	768 bytes	32 bytes
ML-KEM-768	3	1,184 bytes	2,400 bytes	1,088 bytes	32 bytes
ML-KEM-1024	5	1,568 bytes	3,168 bytes	1,568 bytes	32 bytes

[P] NIST identifies ML-KEM-768 as its general default recommendation in the standard. That recommendation should not be substituted for a specialised national-security profile or a jurisdiction-specific requirement.

Source: NIST, FIPS 203, Table 3 and parameter-selection guidance, August 2024. nvlpubs.nist.gov

For transport planning, the following arithmetic is useful, provided its scope remains explicit:

Illustrative object exchange	Calculation	Encoded bytes
ML-KEM-512 public key plus ciphertext	800 + 768	1,568
ML-KEM-768 public key plus ciphertext	1,184 + 1,088	2,272
ML-KEM-1024 public key plus ciphertext	1,568 + 1,568	3,136

These totals are derived from FIPS 203, not measured network traffic. A real protocol may transmit these objects in different messages, reuse keys, add a classical component, include certificates or introduce additional framing.

The recommended performance assessment should distinguish five quantities:

1. Encoded object size, determined by the selected algorithm and parameter set.
2. Protocol message size, determined by how the objects are packaged.
3. Computation time, measured on the actual processor and implementation.
4. Network completion time, measured under relevant loss, congestion and connection patterns.
5. Service impact, measured as successful transactions, failed negotiations and recovery time.

A larger cryptographic object does not establish a proportional increase in transaction latency. Conversely, acceptable primitive-level benchmarks do not establish acceptable behaviour on a constrained appliance or a lossy communications path.

5.3. Signature parameters: the migration has a different bottleneck

[P] FIPS 204 specifies three ML-DSA parameter sets:

Parameter set	NIST security category	Public key	Private key	Signature
ML-DSA-44	2	1,312 bytes	2,560 bytes	2,420 bytes
ML-DSA-65	3	1,952 bytes	4,032 bytes	3,309 bytes
ML-DSA-87	5	2,592 bytes	4,896 bytes	4,627 bytes

Source: NIST, FIPS 204, Table 2, August 2024. nvlpubs.nist.gov

[P] FIPS 205 provides SHA2-based and SHAKE-based variants with the following public-key and signature sizes:

SLH-DSA parameter suffix	NIST security category	Public key	Signature
128s	1	32 bytes	7,856 bytes
128f	1	32 bytes	17,088 bytes
192s	3	48 bytes	16,224 bytes
192f	3	48 bytes	35,664 bytes
256s	5	64 bytes	29,792 bytes
256f	5	64 bytes	49,856 bytes

The listed sizes apply to both hash-family variants for each suffix. Small public keys should not be confused with small signed objects.

Source: NIST, FIPS 205, parameter table, August 2024. nvlpubs.nist.gov

Recommended assessment: treat signature migration as a separate workstream from session establishment. Its acceptance environment includes certificate parsers, update-package formats, verification memory, hardware roots of trust and the installed equipment estate.

Signature-bearing system	Proposed test	Decision the test should support
Certificate hierarchy	Validate complete paths with realistic chain lengths	Whether the existing identity infrastructure can support the chosen profile
Embedded firmware	Verify representative packages on the oldest supported devices	Whether software updates suffice or hardware replacement is necessary
Secure boot	Exercise acceptance, rejection and recovery paths	Whether migration can preserve the device’s trust boundary
High-volume signing service	Measure throughput and tail latency under peak demand	Whether signing infrastructure needs additional capacity
Offline verifier	Test updates without continuous network access	Whether disconnected operations can sustain the replacement trust model
Long-lived signed record	Rehearse future verification and trust renewal	Whether archival procedures preserve evidentiary value

For an acquisition authority, the difficult question is often whether the installed verifier can be changed safely. Funding a new signing server does not resolve an immutable verifier in equipment expected to remain in service.

5.4. Algorithm diversity and the status of HQC

[P] NIST selected HQC in March 2025 for subsequent standardisation as a backup key-encapsulation mechanism using a different mathematical foundation from ML-KEM. [R] The selection announcement described a prospective final standard in 2027. Selection, a draft and a final standard are separate milestones.

The source does not justify describing HQC as a replacement already required across deployed systems, or postponing migration to the completed standards while waiting for it.

Source: NIST, NIST Selects HQC as Fifth Algorithm for Post-Quantum Encryption, March 2025. Here, the announcement’s “encryption” terminology refers specifically to a key-encapsulation mechanism. NIST

Recommended control: maintain algorithm diversity through replaceable implementations, documented interfaces and tested transition procedures. Adding multiple algorithms without a coherent negotiation and authentication design should not count as improved assurance.

A useful procurement requirement is consequently more demanding than “supports PQC.” It should specify:

• The exact algorithm, parameter set and protocol profile.
• The implementation version and applicable validation evidence.
• The mechanism for updating the implementation.
• The supported peer configurations.
• The treatment of incompatible peers and downgrade attempts.
• The operational owner responsible for replacement if an algorithm or implementation is retired.

5.5. Hybrid migration: assurance depends on composition

[P] ANSSI’s follow-up migration guidance supports hybrid approaches and discusses composition requirements. For key establishment, combining mechanisms requires an appropriate construction; for combined signatures, accepting either signature is not equivalent to requiring both. The intended assurance depends on the combiner and verification rule.

Source: ANSSI, Avis sur la migration vers la cryptographie post-quantique — follow-up guidance, 2023 follow-up, official portal publication January 2024. messervices.cyber.gouv.fr

The recommended assurance review should answer the following questions before approving a hybrid deployment:

Review area	Required question
Key combination	What is the security argument for the combiner?
Authentication	Which mechanism authenticates the peer and negotiated configuration?
Transcript binding	Are identities, algorithms and negotiation results bound into the authenticated exchange?
Downgrade handling	Can an active intermediary induce a weaker configuration?
Signature acceptance	Must every required signature verify, or can one alone authorise acceptance?
Error handling	Do malformed inputs expose secrets, distinguish internal failures or trigger unsafe fallback?
Legacy compatibility	Which peers remain outside the intended assurance model?
Operational evidence	Can administrators prove which mechanism protected a particular session or signed object?

[H] “Hybrid” should be rejected as an assurance conclusion when the seller cannot identify the construction and its acceptance rules. The label describes a combination; it does not supply a security proof.

5.6. Confidentiality lifetime and signature lifetime require different clocks

[P] The multinational ANSSI–BSI–NLNCSA–Swedish position paper explicitly recognises the store-now-decrypt-later problem: intercepted material can remain sensitive beyond the period during which its original public-key protection is expected to be adequate. This is a reason to assess retained information now; it is not evidence that an adversary currently possesses a cryptographically relevant quantum computer.

Source: ANSSI, BSI, NLNCSA and Swedish NCSA, Position Paper over Quantum Key Distribution, January 2024. aivd.nl

For planning, define:\[ L=\text{remaining confidentiality lifetime},\qquad M=\text{migration duration},\qquad Q=\text{assumed time until a relevant adversarial capability}. \]

The inequality\[ L+M\geq Q \]

is a scenario-screening rule, not a prediction. It asks whether a chosen scenario leaves enough time to complete migration before newly exposed information outlives the assumed protection horizon.

For illustration only, if an organisation requires ten further years of confidentiality and assumes three years to migrate, its exposure assessment spans thirteen years. This arithmetic establishes no date for the arrival of a quantum computer.

Signature risk should be screened differently. The owner should ask how long equipment will continue accepting a particular signing authority, whether the verifier can be updated, and what procedures preserve the authenticity of historical records.

Planning clock	Exposure being assessed	Recommended response
Confidentiality lifetime	Future recovery of information captured during the vulnerable period	Prioritise exposed, long-lived secrets and their key-establishment paths
Equipment service life	Future acceptance of unauthorised software or commands	Assess verifier replacement and firmware trust transitions
Certificate lifetime	Continued reliance on a retiring trust hierarchy	Plan issuance, validation and revocation changes
Archival verification lifetime	Future inability to establish the authenticity of a historical object	Define preservation and renewal procedures
Migration duration	Time needed to replace dependencies and incompatible equipment	Begin discovery before selecting a universal completion date

5.7. A resource estimate is not a demonstrated cryptanalytic service

[R] A published 2017 logical-circuit study estimated the resources below for an elliptic-curve discrete-logarithm computation against NIST P-256. It is a historical model of a particular circuit construction, not a current minimum and not an attack executed on quantum hardware.

Source: Roetteler, Naehrig, Svore and Lauter, Quantum resource estimates for computing elliptic curve discrete logarithms, ASIACRYPT 2017, Table 2. arxiv.org

---

---

[H] Interpreting the table as evidence that a processor containing 2,330 physical qubits can recover a P-256 private key is a category error. The physical architecture, logical error budget, non-Clifford implementation and execution schedule are unresolved in that translation.

Recommended reporting rule: any newer attack estimate should replace the complete resource row, including its assumptions. A reduced logical-qubit count should not be reported as a reduced operational attack cost unless the associated gate count, error correction, physical resources and runtime have also been assessed.

5.8. Symmetric cryptography should not inherit the RSA/ECC risk model

[R] NIST’s PQC FAQ cautions against translating Grover’s theoretical query reduction directly into a practical attack estimate. It discusses sequential execution costs and the limited benefit of parallelisation, and permits continued use of AES-128, AES-192 and AES-256 subject to applicable guidance.

Source: NIST, Post-Quantum Cryptography FAQs, maintained page, retrieved October 2026. CSRC

[H] “Quantum computers halve every key length” is an inadequate security statement. It suppresses the attack model, reversible implementation, error-correction cost and elapsed time.

Recommended control: select symmetric parameters through the applicable security profile, confidentiality horizon and implementation environment. For new systems protecting long-lived secrets, AES-256 can be considered within that profile; it should not be presented as a substitute for migrating vulnerable public-key establishment or signature functions.

Proposed assessment	Evidence to require
Adequacy of a symmetric parameter set	Applicable policy and a named attack model
Quality of key management	Generation, storage, rotation, access and destruction procedures
Adequacy of session establishment	Evidence that payload keys are not exposed through an obsolete establishment mechanism
Integrity protection	Correct authenticated-encryption or message-authentication implementation
Long-term storage protection	Protection of wrapping keys, backups and recovery copies

5.9. QKD: physical key distribution and system assurance remain separate questions

[P] NSA’s published position states:

“NSA does not recommend the usage of quantum key distribution and quantum cryptography for securing the transmission of data in National Security Systems (NSS)”

The position retains a condition: the identified limitations would need to be overcome. NSA identifies missing source authentication, specialised infrastructure, trusted-relay exposure, implementation-validation difficulties and denial-of-service risk. This is a scoped position on national-security systems, not a denial that QKD experiments produce keys.

Source: NSA, Post-Quantum Cybersecurity Resources — QKD and quantum-cryptography guidance, maintained institutional guidance, retrieved October 2026. nsa.gov

[P] ANSSI’s QKD guidance allows consideration of a limited niche involving fixed optical links and defence in depth, while favouring PQC for broad modern-system migration. It warns against allowing QKD expenditure to displace more immediate cybersecurity work.

Source: ANSSI, Should Quantum Key Distribution be Used for Secure Communications?, May 2020. MesServicesCyber

[P] The January 2024 joint European position prioritises PQC and suitable symmetric-key approaches, and identifies practical assurance work still required for QKD protocols and devices. Its discussion includes finite-key security, physical attacks and evaluation methods. It also recognises existing ISO/IEC evaluation standards and BSI-supported protection-profile work; standards activity should therefore be distinguished from certification of a particular installed network.

Source: ANSSI, BSI, NLNCSA and Swedish NCSA, Joint position paper, official Dutch publication, January 2024. aivd.nl

The recommended QKD acquisition review should examine the following boundaries:

Assurance boundary	Required evidence before acceptance
Peer identity	How the classical channel is authenticated
Device model	Which source, detector and calibration assumptions underpin the proof
Finite-key operation	Security parameters for the actual block sizes and operating conditions
Intermediate nodes	Which locations must remain trusted and physically protected
Payload protection	Whether keys feed AES or another specified mechanism
Availability	Behaviour under fibre interruption, optical disturbance and key exhaustion
Maintenance	How firmware, calibration and defective components are replaced
Endpoint security	How plaintext and keys are protected after delivery
Accreditation	The authority, evaluation scope and approved configuration

[H] A claim of an “unhackable link” should be retired when it substitutes a protocol-level theoretical property for evidence covering these boundaries.

5.10. QSDC and quantum networks: measured results, bounded interpretations

[D] A 2022 quantum-secure direct communication experiment reported the following results using time-bin and phase states:

Experimental configuration	Reported loss	Reported secure communication rate
30 km commercial fibre	6 dB	22.4 kbit/s
100 km ultralow-loss fibre	15.8 dB	0.54 bit/s

The protocol carries information through a quantum communication construction rather than merely distributing keys. These results establish operation under the paper’s conditions; they do not establish a general-purpose replacement for high-throughput communications.

Source: Long and colleagues, Realization of quantum secure direct communication over 100 km fiber with time-bin and phase quantum states, Light: Science & Applications, April 2022. PMC

[D] A separate 2024 experiment entangled diamond-based quantum-memory nodes through a 35 km deployed telecom fibre loop, reporting nuclear-spin Bell-state fidelity of 0.69(7) after error detection. Fibre-path length should not be recast as straight-line separation between the nodes.

Source: Knaut and colleagues, Entanglement of nanophotonic quantum memory nodes in a telecom network, Nature, May 2024. Nature

Recommended interpretation: assess three different deliverables separately:

Deliverable	Appropriate acceptance question
QKD key service	Does the installed system deliver usable keys with the required assurance and availability?
QSDC communication service	Does the protocol provide the required protected payload rate under its stated assumptions?
Memory-based quantum network	Can nodes establish, retain and use entanglement across the intended architecture?

[H] Evidence for one deliverable does not establish the other two. In particular, a long trusted-node QKD route is not evidence of a memory-based quantum repeater network.

5.11. Migration governance: the instrument and its scope matter

[P] The June 2026 US executive order introduces differentiated requirements and prospective implementation actions:

Provision	Specified action or target	Scope distinction
Section 4(b)	OMB guidance requiring PQC key establishment by 31 December 2030	Federal HVAs and high-impact systems; excludes NSS
Section 4(b)	PQC digital signatures by 31 December 2031	Same specified system categories
Section 4(c)	NIST migration pilot completed by 31 December 2027	Appropriate subset of NIST systems
Section 5(c)	NSA migration reporting within 180 days and annually thereafter	Separate national-security-system reporting path
Section 5(d)	Public cryptographic-bill-of-materials guidance within 270 days	CISA-led guidance action
Section 6(c)	Proposed FAR rule within 180 days, addressing covered-contractor compliance by 2030	Direction to propose a rule, not evidence of an already effective contract clause

Source: White House, Executive Order 14412: Securing the Nation Against Advanced Cryptographic Attacks, June 2026. The White House

[P] The European Commission’s coordinated roadmap provides a framework for member-state transition planning. It should be read as a coordination instrument, with implementation responsibilities assessed through the relevant national authorities.

Source: European Commission, Una tabella di marcia coordinata per l’attuazione della transizione alla crittografia post-quantistica, June 2025, page updated September 2026. Plasmare il futuro digitale dell’Europa

[P] NCSC’s migration guidance describes discovery and planning as substantial work, potentially taking two to three years in large organisations. That supports beginning dependency discovery early; it does not establish an identical project duration for every owner.

Source: UK NCSC, Timelines for migration to post-quantum cryptography, March 2025. National Cyber Security Centre

Recommended governance rule: keep an instrument register recording jurisdiction, system category, issuing authority, binding status, implementation milestones and permitted exceptions. A deadline without its scope should not enter an acquisition specification.

5.12. Asset-class decision table

The following is a recommended decision framework, not a claim that every organisation has the same secrecy periods or legal obligations. Lifetimes must be supplied by the information and system owners.

Asset class	Lifetime to establish	Migration priority	Recommended control	QKD consideration
Intelligence identities and relationships	Period during which disclosure can endanger people or operations	Highest where exposed traffic carries enduring secrets	Approved PQC or symmetric arrangements; reduce unnecessary retained copies	Only within an explicitly authorised threat model
Genomic and longitudinal health records	Remaining period of personal and institutional sensitivity	High where records or transfers have long retention	PQC establishment, suitable symmetric protection and storage-key governance	Requires a specific fixed-link justification
Strategic research and industrial designs	Remaining commercial or strategic value	High for valuable material exposed across organisational boundaries	PQC transfer paths and protected archival keys	Possible additional control on a bounded link
Firmware roots of trust	Remaining equipment life	High for devices with difficult or immutable verifier replacement	Signature migration or hardware replacement plan	Does not replace firmware-signature work
Payment infrastructure	Transaction security and infrastructure service life	High because transition must preserve continuity	PQC signatures and establishment with interoperability testing	Does not replace signature migration
Critical-infrastructure administration	Service life and consequences of unauthorised control	High where cryptographic dependencies support remote administration	Approved migration, segmentation and controlled recovery	Assess only as a supplementary channel
Routine short-lived sessions	Actual sensitivity beyond session completion	Prioritise through exposure and policy	Supported PQC protocol implementation	Requires evidence of additional value
Fixed inter-site links	Information lifetime and accepted endpoint/node trust	Case-specific	PQC and authenticated symmetric payload protection	Consider only after assurance and availability review
Offline archives	Retention and future verification requirements	Separate confidentiality and authenticity assessments	Key rewrapping, preservation and trust-renewal procedures	Does not resolve offline archival verification

5.13. AI-assisted cryptanalysis: assess the named scheme

[C] NIST’s maintained PQC page reports a July 2026 announcement that an AI model helped identify a vulnerability in the candidate signature scheme HAWK. [P] It records the candidate’s subsequent withdrawal. NIST explicitly states that this finding does not affect its finalised ML-KEM and ML-DSA standards.

Source: NIST, Post-quantum cryptography — current standards and HAWK notice, notice concerning July 2026, retrieved October 2026. NIST

Recommended interpretation: demand the attacked scheme, parameter set, vulnerability class, reproducibility and attack cost. An algorithmic weakness, a coding error and a side-channel attack should produce different corrective actions.

[H] “AI has broken PQC” is an unsupported generalisation when the evidence concerns one named candidate or implementation. The appropriate institutional response is to review the affected dependency and preserve replacement capability.

Chapter 5 — Key judgments

• [P] Completed standards provide an actionable migration foundation, while deployment assurance remains a separate obligation. The decision changed is whether to begin integration work now. See NIST FIPS 203.
• [P] Agency positions support broad PQC migration and impose substantial qualifications on QKD use. The decision changed is whether a proposed QKD acquisition can satisfy the owner’s actual assurance requirements. See NSA’s scoped position.
• [R] Logical-circuit estimates establish conditional computational requirements, not a functioning adversarial service. The decision changed is how to report cryptanalytic risk without equating physical-qubit counts with attack capability. See the P-256 resource study.

What would change the assessment

A reproducible attack against a finalised scheme at relevant parameters; an implementation vulnerability affecting a widely deployed product; a complete physical attack estimate with validated error-correction assumptions; or a QKD deployment demonstrating assurance and continuity adequate for the relevant authority.

Open official record

The important unresolved questions concern inventory completeness, installed-verifier replacement, implementation validation, exception management and independently audited migration completion. The instruments cited above do not establish that their covered organisations have already completed those tasks.

Chapter 6 — Civilian Uses That Are Real

6.1. Civilian evidence should be evaluated at the level of the named service

Recommended assessment: distinguish scientific performance, integration into an existing workflow and evidence that the workflow produces a useful operational outcome. A sector partnership should enter the register as a partnership; a hardware experiment as an experiment; and an accredited service only when its authorisation and operating scope are documented.

Civilian application	Named evidence	Technical readiness assessment	Institutional readiness assessment
Payment-system PQC	BIS Project Leap phase 2	Integrated pilot in an operating payment environment	Central-bank and infrastructure-owner participation
Metropolitan QKD service	Toshiba–BT trial involving HSBC	Named service trial; technical assurance requires separate evidence	Commercial operator and named customer
Pharmaceutical computation	IBM–Moderna mRNA study	Hardware research experiment	Research collaboration; no clinical-service conclusion
Utility communications	Published trusted-node QKD experiment	Limited field demonstration on utility fibre	Research and utility integration
Airborne gravity surveying	GIRAFE/iMAR Iceland–Greenland campaign	Field experiment with classical comparison	Scientific survey programme
Biomedical sensing	Wearable optically pumped MEG	Integrated human research prototype	Research use; accreditation not established by the paper
Magnetic navigation	April 2025 author preprint	Author-reported airborne and ground trials	Development activity; operational authorisation not established
Time metrology	PTB realisation and dissemination of the second	Established atomic-metrology function	Statutory national-metrology responsibility

The supporting records and limitations are examined below. These classifications should not be read as official technology-readiness-level assignments.

6.2. Finance: migration testing has a clearer operational purpose than a generic quantum-computing pilot

[P] Project Leap phase 2 involved the BIS Innovation Hub, Banca d’Italia, Banque de France, Deutsche Bundesbank and infrastructure participants. [C] Its December 2025 report describes testing PQC signatures in an operational payment system while sending liquidity transfers, and reports integration and performance differences requiring attention.

The evidence is significant because the experiment addresses a concrete security function inside payment infrastructure. It does not establish system-wide production migration or quantum-computing advantage in financial optimisation.

Source: BIS Innovation Hub, Project Leap phase 2: quantum-proofing payment systems, December 2025. “Quantum-proofing” here means migration to post-quantum cryptography. bis.org

Recommended next-stage financial assessment:

Test area	Measurement to request	Acceptance purpose
Transaction completion	Success and failure rates under representative load	Establish continuity of the payment service
Signature verification	Median and tail latency on each receiving platform	Detect heterogeneous implementation bottlenecks
Message handling	Actual signed-message sizes and parser behaviour	Identify incompatible interfaces
Peak demand	Throughput during stress periods	Establish capacity requirements
Recovery	Behaviour after partial upgrade or signing-service failure	Establish controlled continuity
Participant interoperability	Results across institutions and implementations	Avoid a pilot that works only within one configuration
Auditability	Ability to reconstruct algorithm and key use	Preserve security accountability

The recommended financial decision is to expand bounded migration pilots into representative operational testing. Claims about portfolio optimisation, pricing or settlement acceleration should require their own computational evidence and should not inherit credibility from a successful cryptographic trial.

6.3. Telecom: service availability, security assurance and quantum networking are different deliverables

[C] Toshiba’s Japanese-language announcement identifies HSBC as a participant in the Toshiba–BT metropolitan QKD commercial trial service in London. The announcement describes a named operator, customer and fibre-based service arrangement. It does not independently establish every security or availability property of the installed service.

Source: Toshiba, 東芝とBTによる量子暗号通信の商用メトロネットワークのトライアルサービスに金融大手HSBCが参画, July 2023. The mechanism is QKD supplying cryptographic keys. 東芝デジタルソリューションズ

Recommended operator evidence package:

Service dimension	Required documentation
Physical route	Actual optical path, attenuation, permitted switching and maintenance conditions
Key delivery	Delivered secret-key rate, buffer behaviour and exhaustion policy
Customer isolation	Separation of key material and management interfaces
Endpoint trust	Protection of key-management systems and customer encryptors
Intermediate trust	Named trusted nodes and their protection requirements
Availability	Scheduled maintenance, failure modes and recovery results
Security evaluation	Applicable protocol proof and installed-device assessment
Commercial support	Replacement times, spares, software support and liability allocation

The recommended purchasing question is whether the additional key service improves the customer’s defined assurance case at an acceptable operational cost. It should not be reduced to whether the operator can use the word “quantum” in a service description.

6.4. Pharma: hardware experiments should retain their actual problem size

[D] The IBM–Moderna mRNA secondary-structure study, recorded as a QCE 2025 conference paper, reports IBM hardware experiments using up to 156 qubits and 950 nonlocal gates for sequences up to 60 nucleotides. Its workflow combines quantum sampling with classical training or post-processing.

[C] The publication record describes noiseless tensor-network simulations up to 358 qubits, while the May 2025 preprint states 354 qubits. These are separately versioned simulation statements, not a larger hardware experiment.

Sources: IBM Research, Towards secondary structure prediction of longer mRNA sequences using a quantum-centric optimization scheme, August 2025; and May 2025 author preprint. IBM Research

Evidence item	Classification	Permitted interpretation
Quantum-processor experiment	[D]	The stated hybrid workflow ran at the reported hardware scale
Noiseless tensor-network calculation	[C], reported simulation	A classical simulation explored a larger representation
Biological usefulness	[R], further validation required	Must be assessed against experimentally relevant structure and function
Clinical benefit	[R], not established here	Requires evidence connecting the workflow to a therapeutic outcome
General drug-discovery acceleration	[H] if inferred from this experiment alone	Exceeds the demonstrated task and problem size

Recommended continuation: require competitive classical baselines, full elapsed workflow time, sampling costs, reproducibility and biological validation. The research question is whether the quantum component improves a relevant endpoint once the entire hybrid process is counted.

Proposed benchmark	Why it matters
Best available classical method for the same formulation	Establishes the actual comparison
Alternative biological formulations	Tests whether the chosen optimisation objective is useful
Total wall-clock time	Includes training, queueing, execution and post-processing
Solution quality distribution	Prevents selection of only favourable runs
Scaling with sequence length	Tests whether the demonstrated regime extends usefully
Laboratory biological validation	Connects computational output to the intended scientific purpose

6.5. Energy: the utility network is a deployment environment, not evidence of grid optimisation

[D] A published trusted-node QKD study integrated three QKD systems on real electrical-utility fibre and examined interoperability and key-rate behaviour. It is evidence for field integration of communication-security components.

It does not establish quantum-computing advantage for dispatch, state estimation or power-system optimisation.

Source: ORNL, Trusted Node QKD at an Electrical Utility, IEEE Access, 2021. ORNL

Recommended energy-sector separation:

Proposed activity	Security or operational endpoint to assess
QKD-assisted utility communications	Assured key delivery and acceptable availability
PQC migration of administrative access	Authorised remote access across the installed estate
Quantum optimisation experiment	Improved solution quality or time against competitive classical methods
Quantum materials calculation	Validated prediction relevant to a specified material
Quantum sensing near infrastructure	Reliable measurement under the actual environmental conditions

For a utility, a communication-security experiment should advance only after examining outage behaviour, maintenance access, trusted sites and compatibility with operational equipment. A computational experiment should advance through a separate performance and usefulness review.

6.6. Gravity surveying: field data support a bounded application

[D] The Iceland–Greenland airborne campaign used the GIRAFE cold-atom gravimeter alongside an iMAR classical strapdown instrument, with flights conducted in June–July 2023. The 2025 publication reports accuracy around 1–2 mGal, depending on flight conditions, for both systems. The campaign also used GNSS receivers.

Source: Jensen and colleagues, Airborne gravimetry with quantum technology: observations from Iceland and Greenland, Earth System Science Data, April 2025. essd.copernicus.org

Reported feature	Assessment
Airborne cold-atom instrument	[D] Field operation under survey conditions
Classical instrument in the campaign	[D] Relevant comparative evidence
Approximately 1–2 mGal accuracy	[D] Campaign-specific result, affected by conditions
GNSS-supported survey	[D] Navigation support present
Autonomous navigation without GNSS	[H] Not established by this campaign

For scale, 1 mGal equals \(10^{-5}\,\mathrm{m\,s^{-2}}\). This is a unit conversion, not an additional experimental result.

Recommended interpretation: evaluate where the instruments’ differing error characteristics improve a combined survey product. Require repeatability, calibration burden, spatial resolution, operating cost and comparison across representative flight conditions before asserting a commercial advantage.

6.7. Biomedical sensing: a human research instrument has been demonstrated

[D] A 2018 wearable magnetoencephalography experiment combined optically pumped quantum sensors with background-field control and measured brain activity while participants moved naturally. It established a human research prototype with capabilities different from a fixed superconducting-sensor arrangement.

Source: Boto and colleagues, Moving magnetoencephalography towards real-world applications with a wearable system, Nature, March 2018. Nature

Recommended clinical assessment: distinguish the ability to record a signal from the ability to support a validated diagnostic or treatment decision.

Clinical-development gate	Required evidence
Measurement repeatability	Consistent results across sessions and operators
Patient suitability	Performance across the intended population
Environmental tolerance	Defined shielding and field-control requirements
Diagnostic relevance	Validation against the intended clinical endpoint
Workflow integration	Setup, acquisition, interpretation and cleaning procedures
Regulatory status	Authorisation for the specified use and configuration
Supportability	Calibration, maintenance and fault-detection arrangements

[H] A demonstrated wearable MEG research system should not automatically be described as an accredited routine clinical service.

6.8. Civil PNT: magnetic-navigation reports merit testing, not universal performance claims

[C] An April 2025 author preprint reports airborne and ground magnetic-navigation trials using quantum magnetometers, denoising and map matching. It reports a best final airborne positioning result of 22 m, flights up to 19,000 feet, and improvements relative to the specified velocity-aided inertial comparator.

These remain author-reported results in the retrieved preprint. They do not establish certification for civil aviation or performance across every route and platform.

Source: Muradoglu and colleagues, Quantum-assured magnetic navigation achieves positioning accuracy better than a strategic-grade INS in airborne and ground-based field trials, April 2025 preprint. arxiv.org

Recommended PNT acceptance framework:

Dimension	Proposed measurement
Accuracy	Position-error distribution across the complete route
Integrity	Ability to detect and report misleading position estimates
Continuity	Duration and frequency of lost or unreliable fixes
Map dependence	Sensitivity to coverage, resolution and map uncertainty
Platform interference	Performance under representative payload and configuration changes
Route diversity	Results over terrain with different magnetic characteristics
Reference truth	Independent reconstruction of the trial trajectory
Recovery	Behaviour after prolonged ambiguity or sensor interruption

The fusion and map-matching contribution should be measured separately from the sensor contribution where feasible. This is the appropriate place to test algorithmic acceleration; it does not remove the need to validate the physical measurement or reference map.

6.9. Metrology: established atomic timing and deployable advanced clocks have different readiness

[P] PTB documents its national responsibility for realising and disseminating the SI second through its time-metrology work. That established institutional function should be distinguished from the readiness of a particular transportable optical or cold-atom clock.

Source: PTB, Realisation of the SI second, maintained institutional page, retrieved October 2026. PTB.de

Recommended assessment: require the complete uncertainty and dissemination chain. A clock’s internal performance is only one part of the service delivered to a telecommunications node, financial timestamping system or remote instrument.

Metrology layer	Evidence to request
Frequency reference	Calibration and uncertainty budget
Clock output	Stability and accuracy over relevant averaging periods
Time transfer	Distribution-path uncertainty
Customer interface	Delivered timing performance
Environmental operation	Temperature, vibration and power sensitivity
Maintenance	Recalibration and intervention requirements
Service continuity	Holdover and recovery behaviour

Chapter 6 — Key judgments

• [P]/[C] Named financial and telecom pilots show institutional engagement with concrete security services. They do not establish migration of an entire sector. See Project Leap phase 2.
• [D] Selected sensing and surveying applications have physical field or human-system evidence. Their useful operating envelopes must remain attached to the result. See the airborne gravity campaign.
• [D] Pharmaceutical computation has identifiable hardware experiments, but the cited study does not establish a clinical outcome. See the QCE 2025 publication record.

What would change the assessment

Independent operational replication; representative baseline comparisons; published service-level performance; documented accreditation; and evidence that the quantum component improves the intended outcome after integration and maintenance costs are included.

Open official record

The major missing items are often outside the scientific paper: sustained availability, total ownership cost, installed-base compatibility, authorisation scope and measured outcomes under ordinary operating conditions.

Chapter 7 — Military and Intelligence Uses That Are Real

7.1. Mission evidence must connect the component to an operational decision

Recommended assessment: organise defence applications around the mission effect, then examine the sensor or cryptographic mechanism. Each programme should state what the operator can do differently, under which conditions and with what confidence.

The technical and institutional axes must remain separate:

Technical evidence	Institutional evidence needed for a stronger operational conclusion
Laboratory sensitivity	Written mission requirement
Integrated prototype	Representative platform and environmental testing
Limited field trial	Operator evaluation and acceptance criteria
Repeated mission trial	Acquisition decision and support arrangements
Series equipment	Training, calibration, spares and configuration control
Accredited deployment	Authorised operating scope and continuing assurance

This is a proposed evidentiary progression, not a claim that every programme must follow an identical procurement sequence.

7.2. Timing: current naval evidence includes integration trials

[C] The Royal Navy’s August 2026 report describes trials involving cold-atom clocks, Saab Giraffe 1X radar equipment, testing at Portsdown Hill and associated industry sites, and rebroadcast activity involving XV Patrick Blackett. It describes simulated GNSS interference and investigation of timestamp alignment.

The report does not publish a complete holdover specification, establish fleet deployment or demonstrate immunity to every form of timing disruption.

Source: Royal Navy, Royal Navy’s technology experts work with industry to test quantum clock in milestone trials, August 2026. royalnavy.mod.uk

[C] A separate June 2025 report describes AQlock trials aboard HMS Puncher. It establishes a named maritime trial, rather than a publicly specified operational timing service.

Source: Royal Navy, Quantum navigation technology takes next step after trial on Royal Navy P2000, June 2025. royalnavy.mod.uk

[C] Dstl’s February 2026 report describes extended unattended operation in conditions intended to be more representative of deployment. [R] It identifies further trials as a prospective development step.

Source: Dstl, Trial speeds up next generation of atomic clocks, February 2026. GOV.UK

Named evidence	Technical readiness judgment	Institutional readiness judgment	Limitation
Royal Navy radar-clock trial	Limited integration trial	Service-owned evaluation	Public performance specification incomplete
HMS Puncher clock trial	Limited maritime field trial	Named naval platform and partners	No fleet-wide acceptance conclusion
Dstl unattended-operation work	Component and device evaluation	Defence research ownership	Operational endurance specification not established

7.3. Distributed radar and SIGINT: delivered timing is the relevant metric

[R] The cited clock trials support further evaluation of timing services for distributed sensing. They do not, by themselves, establish the accuracy of a complete radar or signals-intelligence network.

Recommended acceptance tests:

Timing function	Proposed requirement to measure
Local frequency stability	Performance over mission-relevant averaging intervals
Holdover	Time error accumulated after external-reference loss
Node synchronisation	Relative timing error between deployed nodes
Distribution	Delay variation and asymmetry in the transfer path
Timestamp generation	Error at the actual sensor or receiver interface
Recovery	Behaviour when the external reference returns
Fault indication	Whether operators receive timely warning of degraded assurance
Sustainment	Calibration interval and required technical intervention

A useful physical conversion is

For a one-way propagation interpretation, one nanosecond corresponds to approximately 0.30 m. A monostatic round-trip radar interpretation introduces the appropriate factor of two. These are arithmetic relationships, not performance specifications for the naval trials.

Recommended decision: define the required error at the measurement endpoint before selecting a clock. The acquisition should demonstrate that the distribution architecture preserves the necessary performance under representative disruption.

7.4. Contested-spectrum PNT: robust integration is a funded research problem

[P] DARPA’s Robust Quantum Sensors — RoQS programme explicitly addresses the difficulty of operating quantum sensors under motion, vibration and electromagnetic disturbance. Its programme instrument includes solicitation DARPA-PS-25-11.

[R] The phase-one announcement describes prospective testing with a government-provided helicopter and platform-integration studies, including ground vehicles, submarines, satellites and uncrewed aircraft. Planned testing should not be reported as a completed mission demonstration.

Sources: DARPA, RoQS: Robust Quantum Sensors, maintained programme page; and RoQS launches first phase, August 2025. DARPA

Recommended contested-PNT review:

Navigation dependency	Proposed stress test
Inertial sensor	Representative motion, vibration and thermal transitions
Magnetic sensor	Platform-generated fields and changing configuration
Reference map	Missing, inaccurate and spatially ambiguous data
Fusion software	Sensor disagreement and misleading inputs
Timing source	External-reference loss and recovery
Position solution	Long routes with independent ground truth
Operator interface	Clear indication of degraded confidence
Support system	Field recalibration and replacement procedures

[H] “GNSS-independent” should not be accepted as proof of immunity to navigation deception, environmental interference or estimation failure. The acquisition authority should identify which external dependencies have been removed and which remain.

7.5. ASW and magnetic anomaly detection: a candidate mission requires its own target evidence

[R] RoQS identifies submarine platform-integration studies as a prospective area. That is evidence of programme interest in platform suitability; it does not establish an anti-submarine detection capability or a detection range.

Source: DARPA, RoQS launches first phase, August 2025. DARPA

Recommended ASW evidence package:

Mission question	Required trial information
What was detected?	Target class, configuration and relevant signature
At what geometry?	Sensor–target relationship and trajectory
Against what background?	Environmental and platform interference
With what detection probability?	Results at a specified false-alarm rate
With what localisation quality?	Position uncertainty and track continuity
Against which comparator?	Relevant conventional sensor and processing chain
Under what disclosure limits?	Publicly releasable scope of the conclusion
With what sustainment burden?	Calibration, platform compensation and maintenance

[H] A laboratory magnetometer sensitivity cannot be converted directly into an ASW range. Sensitivity is an instrument property; detection and classification require a target signature, background model and decision threshold.

Recommended decision: fund blinded, representative target trials before assigning mission capability. Where target details cannot be released, public reporting should still distinguish a research milestone from an accepted operational system.

7.6. Tunnel and void detection: civil proof of measurement is not military target identification

[D] The Birmingham gravity-cartography experiment demonstrated outdoor detection of a two-metre tunnel, reporting 20 E statistical uncertainty and 0.5 m spatial resolution across an 8.5 m survey line. The published result also includes substantial uncertainty in inferred depth.

Source: Stray and colleagues, Quantum sensing for gravity cartography, Nature, February 2022. University of Birmingham

[R] This supports further evaluation for subsurface survey missions. It does not establish reliable military identification of arbitrary underground structures.

Recommended military trial design:

Trial element	Purpose
Blinded target inventory	Prevent prior knowledge from determining interpretation
Multiple depths and geometries	Test the usable operating envelope
Geological variation	Measure background ambiguity
Non-target anomalies	Establish false-alarm performance
Survey-time constraint	Test whether the method fits the mission
Competing instruments	Establish comparative value
Independent excavation or reference survey	Confirm the inferred structure
Operator replication	Test dependence on specialist interpretation

The acceptance endpoint should be a specified detection or classification outcome with uncertainty. A successful measurement of a gravity anomaly should not automatically count as identification of its cause.

7.7. RF reception: a programme target must not become a claimed deployed capability

[P] DARPA’s Quantum Apertures programme investigated Rydberg-atom RF receivers. Its current reference page identifies the programme as complete, but does not publish evidence that every stated performance objective was achieved.

[R] The 2021 team-selection announcement described targets including reception over 10 MHz to 40 GHz or more, directionality, sensitivity and compact packaging. Those numbers are programme objectives, not a verified specification for a fielded SIGINT receiver.

Sources: DARPA, QA: Quantum Apertures, completed-programme reference page; and DARPA Selects Research Teams to Enable Quantum Shift in Spectrum Sensing, 2021. DARPA

Recommended RF evaluation:

Performance dimension	Required distinction
Frequency coverage	Tunable coverage versus instantaneous bandwidth
Sensitivity	Performance for the stated modulation and integration time
Dynamic range	Behaviour with strong nearby signals
Direction finding	Angular performance of the complete receiver architecture
Waveform recovery	Successful demodulation of representative signals
Retuning	Time needed to change operating frequency
Packaging	Complete optical, control, power and thermal system
Field operation	Temperature, vibration and electromagnetic tolerance

[H] A statement that a receiver can be tuned across a broad range should not be restated as simultaneous interception of that entire range.

7.8. Quantum radar: prospective defence work is not an operational radar record

[P] Canada’s Quantum 2030 implementation plan identifies defence research missions including quantum radar, quantum lidar, algorithms and networking. [R] It states a TRL 7 by 2030 ambition. This is a programme objective, not a completed capability.

Source: Canadian Department of National Defence, Quantum S&T Strategy Implementation Plan, March 2023. Canada.ca

Recommended radar assessment: carry forward the mechanism analysis from Pillar I, then require evidence at the full-system boundary. Relevant tests should specify transmitted energy, bandwidth, target, range, background, integration time, detection probability and false alarms.

Claimed result	Evidence required
Improved sensitivity	Matched-resource comparison with a suitable classical receiver
Improved target detection	Representative target and clutter trial
Resistance to countermeasures	Named countermeasure and measured outcome
Operational range	Full link budget and field evidence
Compact deployable equipment	Complete system size, weight, power and environmental results
Mission readiness	Acceptance authority, operating scope and support arrangements

[H] A defence research mission labelled “quantum radar” does not establish long-range detection of stealth platforms. Such a conclusion requires its own system and target evidence.

7.9. Communications: migration authority and network experimentation serve different purposes

[P] NSA’s current resource page identifies CNSS Policy 15, released in March 2025, and the CNSA Suite 2.0 materials as the relevant national-security cryptographic instruments. National-security owners should use those applicable profiles rather than assume that general-purpose NIST defaults satisfy their requirements.

Source: NSA, Post-Quantum Cybersecurity Resources, maintained page, retrieved October 2026. nsa.gov

[P] NATO’s quantum strategy identifies transition toward quantum-resistant cryptography as an alliance concern. [R] Its desired outcomes and possible future quantum-communications contributions should not be read as evidence that allied systems have already completed migration.

Source: NATO, Summary of NATO’s Quantum Technologies Strategy, January 2024. NATO Official text

Recommended national-security migration controls:

Control area	Evidence to require
Governing profile	Applicable authority and approved algorithms
Equipment inventory	Cryptographic dependencies by device and service
Tactical constraints	Representative bandwidth, loss and disconnection tests
Coalition interoperability	Authorised peer configurations and transition arrangements
Trust anchors	Verifier replacement and certificate-hierarchy plan
Key management	Approved provisioning, storage and destruction procedures
Configuration assurance	Evidence that installed equipment uses the approved profile
Exceptions	Named owner, residual risk and retirement date
Sustainment	Support for long-lived and difficult-to-access equipment

[R] Early memory-based quantum-network experiments remain relevant to future network research, but the Chapter 5 demonstration does not establish an accredited military communications service. Its evidence should remain attached to the physical experiment, not multiplied into a mission claim.

7.10. Intelligence exposure: retained secrets and operational trust require separate priorities

Recommended intelligence assessment: divide the exposure register into three classes:

Exposure class	Assessment question	Proposed priority
Retained confidentiality	What intercepted information would remain valuable years later?	Protect enduring identities, relationships and strategic information
Live operational authentication	Which future messages, software or authorities must remain trustworthy?	Migrate signatures, identities and verification paths
Archived evidentiary authenticity	How will historical records retain verifiable provenance?	Preserve trust state and define renewal procedures

The recommended collection-risk review should identify which communications are plausibly obtainable, how long their contents remain consequential, and whether the vulnerable cryptographic dependency lies in session establishment, storage or an upstream key-wrapping layer.

The response should then match the exposure. Re-encrypting a local archive cannot recall copies already intercepted. Replacing a network establishment mechanism does not update an offline firmware verifier. Reducing unnecessary retention can lower exposure, but it does not establish secure authentication.

These are planning distinctions, not claims about a particular intelligence service’s holdings or computing capability.

7.11. Computing: a defence owner does not establish mission advantage

[P] Canada’s implementation plan includes a quantum-algorithms mission. [R] Its prospective application objectives require later technical and operational validation; the programme’s existence does not demonstrate a useful military optimisation service.

Source: Canadian Department of National Defence, Quantum S&T Strategy Implementation Plan, March 2023. Canada.ca

Recommended computational evidence package:

Field	Required content
Mission problem	Exact decision or calculation being supported
Instance size	Full problem dimensions and constraints
Encoding	Reduction and data-preparation costs
Hardware	Processor, qubit type and execution conditions
Resources	Circuit depth, samples and applicable error correction
Comparator	Competitive classical algorithm on relevant hardware
Total time	Complete workflow elapsed time
Output quality	Objective value, accuracy or useful decision metric
Robustness	Behaviour across representative instances
Reproducibility	Data, code and sufficient methodological detail

[H] A small scheduling demonstration, a synthetic optimisation instance or a defence-funded processor should not be described as a mission advantage without this comparison.

The recommended defence investment should pursue well-defined research questions and preserve the ability to stop work when improved classical methods remove the hypothesised advantage.

7.12. Military readiness register

This register consolidates the assessment without treating research ownership as procurement or doctrine.

Application	Best evidence examined in this block	Technical readiness judgment	Institutional readiness judgment	Current claim boundary
Naval advanced clocks	[C] Named sea and integration trials	Limited field and integration trials	Service-owned evaluation	No fleet-wide timing specification established
Distributed radar timing	[C] Radar-clock trial	Integration trial	Named defence and industry participants	No complete operational-network result established
Robust quantum sensing	[P] RoQS	Research and prospective platform testing	Funded programme and solicitation	Planned test is not a completed trial
ASW-related sensing	[R] Candidate mission inference	Mission performance not established	Platform-study interest	No justified detection-range claim
Tunnel detection	[D] Civil outdoor experiment	Demonstrated survey experiment	Military acceptance not established by cited evidence	No general target-identification capability
Rydberg RF reception	[P] Quantum Apertures	Programme research; field specification unresolved	Completed research programme	Objectives are not an accredited receiver specification
Quantum radar	[P]/[R] Canadian programme and target	Operational mission result not established here	Defence research ownership	No operational long-range radar conclusion
NSS cryptographic migration	[P] Applicable policy instruments	Implementation depends on system	Formal security authority	Policy does not establish completed estate migration
Experimental quantum networks	[D] Physical experiment discussed in Chapter 5	Experimental network integration	Military service ownership not established	No accredited military network conclusion
Military quantum algorithms	[P]/[R] Named research mission	Application-specific validation required	Defence research ownership	No general mission advantage established

7.13. The next funding gate should require evidence that resolves a mission uncertainty

Recommended acquisition discipline:

Funding gate	Evidence to request	Decision enabled
Component continuation	Repeatable performance under specified conditions	Whether the physical mechanism warrants integration
Platform integration	Representative motion, noise and environmental testing	Whether the device is suitable for the platform
Mission experiment	Blinded targets and relevant comparators	Whether the system changes an operational outcome
Pre-procurement evaluation	Reliability, maintainability and reproducibility	Whether acquisition is justified
Operational acceptance	Authorised configuration, training and logistics	Whether the service can be relied upon
Expansion	Repeated performance across platforms and locations	Whether initial success generalises

For timing, the gate should resolve delivered holdover and synchronisation performance. For navigation, it should resolve position integrity across representative routes. For subsurface sensing, it should resolve detection and false alarms. For RF reception, it should resolve waveform recovery and dynamic range. For cryptography, it should resolve approved interoperability and installed configuration.

A programme should receive additional funding because the proposed experiment can settle a material uncertainty, rather than because its terminology fits a strategic theme.

Chapter 7 — Key judgments

• [C] The strongest current naval evidence examined here concerns named timing trials and integration work. It supports continued evaluation, with operational specifications still required. See the August 2026 Royal Navy trial report.
• [P] Robustness under platform conditions is an explicit funded research objective. The decision changed is whether to require representative testing before procurement. See DARPA RoQS.
• [P]/[R] Defence strategies and target dates establish institutional intent. They do not establish ASW range, operational quantum radar or computational mission advantage. See Canada’s implementation plan.
• [P] National-security cryptographic migration has identified policy authorities and should proceed through the applicable profiles and assurance processes. See NSA’s resource guidance.

What would change the assessment

Published representative mission trials; independent replication; detection results at specified false-alarm rates; delivered timing and navigation specifications; accredited communications configurations; series-production evidence; or computational results that outperform competitive classical methods on relevant mission instances.

Open official record

The decisive unresolved questions are performance under operational disturbance, sustained availability, maintenance burden, acceptance authority and transition from research ownership to procurement and doctrine. Public evidence examined here supports several concrete experiments and programmes; it does not support assigning them a common operational maturity.

---

Pillar III — Industrial Power, Conditional Outlook and Decisions

Chapter 8 — The Industrial and Geopolitical Layer

8.1. Industrial position should be assessed through control, reproducibility and continuity

The recommended unit of industrial assessment is the capability that a country, institution or supplier can reproduce, maintain and improve under realistic constraints. A national laboratory, an installed processor and a domestic company represent different forms of participation; none alone establishes control of the complete production chain.

For a cabinet office, the consequential question is whether the national programme can retain access to the technologies and people required for its chosen applications when commercial conditions, ownership, export permissions or technical requirements change. For an investor or acquisition authority, the corresponding question is whether the supplier can sustain the promised service beyond its initial demonstration.

Assessment axis	Evidence to collect	Conclusion the evidence can support
Hardware design authority	Ownership or licensed control of architecture and design files	Ability to modify the system
Fabrication access	Qualified production process, contractual access and alternative capacity	Ability to produce additional devices
Component continuity	Supplier map, replacement paths and maintenance arrangements	Ability to sustain installed equipment
Integration competence	Repeated assembly, calibration and commissioning results	Ability to reproduce a working system
Software and control authority	Access to firmware, control interfaces, decoders and update procedures	Ability to maintain performance and security
Standards participation	Contributions, evaluation facilities and implementation experience	Ability to influence and implement interoperable requirements
Workforce continuity	Relevant skills, retention and succession arrangements	Ability to operate and improve the capability
Financial durability	Funds available against milestones and continuing operating costs	Ability to complete the programme and support its outputs
Mission ownership	Named user, requirement and acceptance authority	Ability to convert research into an adopted service

Recommended interpretation: industrial sovereignty should be reported as a set of controlled dependencies. The phrase “sovereign quantum computer” should trigger an examination of design authority, production access, operating rights and supportability, rather than close that examination.

8.2. Public funding figures require an accounting classification before comparison

The financial register should distinguish at least six categories:

Financial category	Meaning for the assessment	Error to avoid
Authorised envelope	Amount approved or permitted for a programme	Treating the entire envelope as expenditure
Budget request	Funding sought for a fiscal period	Treating a request as enacted funding
Budget authority or allocation	Funding made available under the relevant process	Treating it as money already used
Contracted commitment	Obligation under a specified agreement	Treating every commitment as an immediate cash payment
Released funds	Resources transferred to the implementing body	Treating transfer as completed technical work
Reported expenditure or utilisation	Resources recorded as spent or used	Treating spending as proof of capability

[P] The US National Quantum Initiative’s FY2025 supplement explicitly separates historical expenditure, FY2024 enacted budget authority and the FY2025 request. Its reporting structure demonstrates why the categories should remain visible in an international comparison.

Source: US National Science and Technology Council, National Quantum Initiative Supplement to the President’s FY2025 Budget, December 2024. quantum.gov

8.3. Selected financial instruments: amounts, periods and limits

The following table records selected public instruments, not exhaustive national spending. The amounts are deliberately retained in their original currencies; converting them into dollars would not resolve differences in time horizon, scope or accounting status.

Jurisdiction	Amount and period in the cited record	Accounting interpretation	Evidence classification and source
United States	US$1,036 million, FY2023; US$1,006 million, FY2024; US$998 million, FY2025	Respectively reported expenditure, enacted budget authority and requested authority	[P] Historical federal QIS series, not a verified FY2026 total. FY2025 supplement, December 2024
United Kingdom	£2.5 billion, ten years from 2024	Strategy funding commitment across a multiyear programme	[P] Commitment does not establish expenditure to date. National Quantum Strategy, March 2023
European Union	€1 billion for the Quantum Technologies Flagship	Long-term EU research initiative	[P] Programme scope differs from aggregate member-state spending. Commission quantum policy record
France	Additional €1 billion, announced for 2026–2030	New investment announcement within the national strategy	[C] Announcement is not an independently audited disbursement record. DGE update, May 2026
Germany	€2.18 billion federal funding, plus €282 million from scientific organisations, described in the 2023 parliamentary discussion	Financing associated with the action concept	[C] Parliamentary description; not a consolidated 2026 expenditure statement. Bundestag committee record, May 2023
India	₹6,003.65 crore, eight-year National Quantum Mission	Approved mission outlay	[P] Envelope must be separated from released and utilised amounts. Parliamentary response, February 2026, Hindi
Israel	NIS100 million, July 2026 infrastructure call	Named national R&D infrastructure initiative	[P] Call for proposals, not evidence that the proposed facility is already operational. Israel Innovation Authority, July 2026, Hebrew
Australia	A$940 million, joint Commonwealth–Queensland investment announced in April 2024	Project-specific support for PsiQuantum in Brisbane	[P] Financial support; delivery of the intended computer remains a separate technical milestone. Government progress report, 2024
Canada	C$360 million, seven years beginning in 2021–2022	Incremental funding supporting the national strategy	[P] Does not encompass every related federal or provincial programme. National Quantum Strategy, French official version

The cited records support the amounts and their classifications, but do not support adding the rows into a global expenditure total. quantum.gov

For China, Japan and Italy, the policy records examined below establish strategic direction or programme architecture. They do not supply a directly comparable, consolidated quantum-only expenditure series for this table. That is a limitation of the comparison, rather than a finding that the countries spend nothing.

8.4. India provides a useful example of financial implementation visibility

[P] A March 2026 parliamentary response distinguishes sanctioned, released and utilised funds for selected implementing institutions:

Institution	Sanctioned through March 2031	Released	Utilised	Utilised ÷ released
IISER Pune	₹30.61 crore	₹18.14 crore	₹4.71 crore	25.96%
IIT Bombay	₹558.76 crore	₹114.89 crore	₹22.31 crore	19.42%
TIFR Mumbai	₹71.03 crore	₹36.15 crore	₹3.63 crore	10.04%

The percentages are this report’s arithmetic, calculated from the published amounts. The rows are not the complete mission budget.

Source: Government of India, Parliament question: women’s participation in quantum research and entrepreneurship, including fellowships, March 2026. Press Information Bureau

Recommended interpretation: these ratios should trigger an implementation review, rather than an automatic performance judgment. Funds sanctioned through 2031 cannot be compared with expenditure recorded in 2026 as if all technical deliverables were already due.

The review should examine procurement timing, facility construction, hiring, grant-release conditions and the relationship between expenditure and completed milestones. Financial visibility is valuable precisely because it permits those questions to be asked without substituting a headline envelope for evidence of implementation.

8.5. United States: evaluate the connection between research, verification and production

[P] The NQI budget supplement describes activities across federal agencies and records work on supply-chain analysis, enabling technologies and workforce development. Its scope is wider than the activities specifically authorised under the original NQI legislation.

Source: NSTC, FY2025 National Quantum Initiative supplement, December 2024. quantum.gov

[P] DARPA’s Quantum Benchmarking Initiative provides a separate verification mechanism. It seeks to determine whether an approach can achieve utility-scale operation by 2033, defining utility in terms of computational value exceeding cost. Its stages move from a system concept, through an R&D plan, to government verification and validation.

Source: DARPA, Quantum Benchmarking Initiative, maintained programme record, retrieved October 2026. DARPA

Analytical judgment: the important US policy distinction is between financing an approach and validating its complete economics. An acquisition authority should preserve that separation even when the supplier has a substantial private balance sheet or extensive scientific publications.

Recommended national assessment: examine whether verification findings can influence funding decisions, whether component suppliers have sustainable production paths, and whether trained personnel can move between research, manufacturing and operational support.

8.6. United Kingdom: mission targets are useful only when intermediate evidence remains visible

[P] The UK strategy commits funding to research hubs, accelerators and challenge-led innovation. [R] Its mission framework includes a 2035 computing objective of one trillion operations, with intermediate computing milestones in 2028 and 2032.

Sources: UK government, National Quantum Strategy, March 2023; and National Quantum Strategy Missions, November 2023. GOV.UK

Analytical judgment: a mission framework can connect scientific work with users, but its value depends on how “operations” are defined and whether intermediate systems meet complete workload requirements.

For the 2026–2031 assessment, the 2035 objective belongs outside the reporting window. The nearer milestones should be evaluated through gate definitions, supported logical register size, success probability and elapsed execution time.

Recommended national assessment: retain separate scorecards for mission progress, manufacturing capability, user adoption and standards implementation. A successful field trial should advance the relevant sensing mission without being counted as progress toward fault-tolerant computation.

8.7. European Union: coordination and industrialisation are explicit policy objectives

[P] The July 2025 Quantum Europe Strategy identifies research, infrastructure, ecosystem development, space and dual-use applications, and skills as interconnected areas of action. It also identifies fragmentation and the conversion of innovation into market opportunities as policy problems.

Source: European Commission, Quantum Europe Strategy, July 2025. Shaping Europe’s digital future

[P] The Commission’s maintained policy record describes a €100 million EuroHPC investment associated with six selected quantum-computing sites, financed equally by the EU and participating countries. [R] The same record describes the Quantum Act as a forthcoming initiative scheduled for adoption in 2026; that wording does not establish an already enacted and effective Act.

Source: European Commission, Quantum policy, infrastructure and Quantum Act record, maintained page, retrieved October 2026. Shaping Europe’s digital future

Analytical judgment: EU-level infrastructure can provide shared access and common testing environments. Its industrial contribution should be judged through utilisation, qualified supply chains, retained design authority and the ability to support equipment across national boundaries.

Recommended comparison rule: do not add EU funding to member-state envelopes unless co-financing and overlapping allocations have been reconciled. The same project can appear in several policy narratives without representing several independent investments.

8.8. France: a larger target changes the verification burden

[C] The May 2026 DGE update reports €1 billion mobilised by the state since 2021, €1.8 billion in cumulative public and private commitments, and €610 million invested through France 2030. These categories should not be added together.

[P] The update identifies PROQCIMA as a programme intended to develop a fault-tolerant computer for state requirements. [R] It raises the stated 2032 objective to 1,024 logical qubits, compared with an initial 128.

Source: French Directorate-General for Enterprise, France 2030 : la France accélère sa stratégie quantique pour renforcer sa souveraineté technologique, May 2026. Direction générale des Entreprises

Analytical judgment: the target revision increases the importance of intermediate acceptance criteria. It is a change in ambition, not evidence that the larger system exists, and should not be described as schedule slippage without evidence of a missed milestone.

Recommended national assessment: request the logical-operation specification, production architecture and milestone verification process. Because the revised endpoint lies in 2032, progress within 2026–2031 should be evaluated through the components and integrated prototypes required to reach it.

8.9. Germany: the 2026 transition should be evaluated through usable systems

[P] Germany’s 2023 action concept established a policy framework through 2026. [R] It included a target for a system with at least 100 individually controllable qubits, scalable to 500, alongside sensing and communications objectives.

Source: German federal government, Handlungskonzept Quantentechnologien, April 2023. quantensysteme.info

[P] The 2026 federal research and innovation report emphasises moving quantum computing, sensing and communications from demonstrators toward usable systems.

Source: German federal government, Bundesbericht Forschung und Innovation 2026 — research and innovation policy guidelines, 2026. BuFI

Analytical judgment: the relevant industrial test is whether demonstrators become reproducible products with defined operating envelopes and support arrangements. A physical-qubit target should remain separate from logical-computing performance.

Recommended national assessment: examine the continuity between the expiring action concept and its successor measures, especially manufacturing qualification, user requirements and evaluation of systems intended for industrial environments.

8.10. China: strategic priority is visible; a comparable expenditure total remains unresolved

[P] The published Fifteenth Five-Year Plan places quantum technologies among strategic frontier fields and identifies quantum technology within the development of future industries.

Source: 中华人民共和国国民经济和社会发展第十五个五年规划纲要, official Chinese publication of the 2026–2030 plan, March 2026. zyshgzb.gov.cn

Analytical judgment: this establishes policy priority and a national planning context. It does not establish a consolidated quantum-only budget, the technical performance of a processor or the operational security of a communications network.

The recommended comparison should therefore retain two separate records: one for policy instruments and disclosed expenditure, and another for reproducible technical evidence. Institutional claims of indigenous production should remain [C] until supported by an auditable production record or appropriate technical evidence.

Recommended national assessment: examine named institutions, procurement, fabrication routes and qualified components. Do not infer a quantum budget by assigning an unspecified portion of broader science, semiconductor or strategic-industry expenditure to the sector.

8.11. Japan: its own policy record identifies international dependencies

[P] Japan’s May 2025 ecosystem measures explicitly address component supply chains, industrialisation and international cooperation. [C] The document identifies import dependencies for some high-end lasers and single-photon detectors and discusses supply risks involving materials and advanced measurement equipment.

Source: Japanese Cabinet Office, 量子エコシステム構築に向けた推進方策, May 2025. www8.cao.go.jp

[P] The Cabinet Office’s current record also lists 2026 cooperation instruments involving the Netherlands, Australia and India.

Source: Japanese Cabinet Office, 量子技術イノベーション — strategy and international-cooperation register, maintained page, retrieved October 2026. 内閣府

Analytical judgment: the Japanese record supports an assessment based on strategic autonomy combined with reliable international access. Domestic integration does not require every component to be domestically manufactured, but dependence should be visible and manageable.

Recommended national assessment: identify which component technologies are intended to become nationally indispensable, which remain internationally sourced, and how qualification or replacement would proceed if a supply path were interrupted.

8.12. India: institutional architecture and physical-qubit targets should remain distinct

[P] India’s National Quantum Mission has four thematic hubs covering computing, communication, sensing and metrology, and materials and devices. The official Hindi parliamentary record describes their governance and implementation arrangements.

[R] The mission specifies physical-qubit development bands of 20–50, 50–100, and 50–1,000 at its three-, five- and eight-year stages.

Source: Government of India, राष्ट्रीय क्वांटम मिशन — parliamentary response, February 2026. Press Information Bureau

Analytical judgment: the hub structure provides identifiable institutional owners. Its physical-qubit milestones do not establish a future logical register or cryptanalytic capability.

Recommended national assessment: connect hub financing to commissioned facilities, qualified production processes, trained personnel and independently evaluated devices. The financial implementation table above offers a starting point, but technical milestones require their own evidence.

8.13. Israel: multi-platform infrastructure is an integration instrument

[P] The Israel Innovation Authority’s July 2026 call proposes national R&D infrastructure integrating at least three quantum-processing technologies. It describes services spanning benchmarking, integration, control, error correction, software, applications and workforce development.

Source: Israel Innovation Authority, רשות החדשנות מקימה תשתית מו״פ לאומית למחשוב קוונטי בהשקעה של 100 מיליון שקלים, July 2026. רשות החדשנות

Analytical judgment: a multi-platform facility can improve comparative testing and reduce premature commitment to one architecture. Its value should be measured through the quality of evaluation, access, integration and technology transfer.

The call does not establish that every platform will be domestically fabricated or that the infrastructure already delivers the proposed services.

Recommended national assessment: require a clear distinction between ownership of equipment, access to control interfaces, rights to experimental data and the ability to modify or replace the underlying platform.

8.14. Australia: project concentration makes delivery verification consequential

[P] The Australian government’s progress report identifies the joint PsiQuantum investment and support for other activities, including Silicon Quantum Computing and feasibility projects.

Source: Australian Department of Industry, Science and Resources, State of Australian Quantum report — progress, 2024. Department of Industry Science and Resources

Analytical judgment: substantial project-specific support creates a concentrated exposure to that project’s technical and contractual execution. The assessment should separately examine construction progress, manufacturing qualification, system acceptance, continuing operating costs and the domestic capabilities retained through the arrangement.

Recommended national assessment: define what remains useful if the central computing milestone arrives late or changes technically. Relevant retained assets may include qualified processes, integration knowledge, facilities and personnel, but their existence and transferability must be documented rather than assumed.

8.15. Canada: research, talent and commercialisation have identified programme roles

[P] Canada’s national strategy organises support around research, talent and commercialisation, with missions covering computing, communications and sensing. Its funding mechanisms include support for research collaboration and operational prototype testing.

Source: Innovation, Science and Economic Development Canada, Stratégie quantique nationale du Canada, official French strategy record. Stratégie quantique nationale

Analytical judgment: the commercialisation test is whether an enterprise can repeatedly deliver and support a product at an acceptable cost. Scientific participation and prototype funding should remain visible as inputs, rather than be recorded as completed industrial output.

Recommended national assessment: examine production access, market concentration, skilled-worker retention and the transition from prototype evaluation into recurring procurement.

8.16. Italy: coordination should be connected to funded implementation

[P] Italy adopted its national quantum-technology strategy through the Interministerial Committee for Digital Transition in July 2025. The government record describes coordination between the digital-transformation and university-research policy structures.

Source: Italian Department for Digital Transformation, Tecnologie quantistiche: una Strategia per l’Italia, July 2025. innovazione.gov.it

Analytical judgment: the central evaluation question is how coordination becomes a funded and accountable implementation chain. An assessment should identify programme owners, facilities, industrial contributions, user access and acceptance criteria.

Recommended national assessment: distinguish capabilities developed domestically, capabilities accessed through European infrastructure and capabilities bought from international suppliers. Each can contribute to national resilience, but they confer different forms of control.

8.17. Sovereignty should be tested at the dependency level

The following is a recommended dependency audit. It identifies potential bottlenecks to investigate; it does not assert a verified shortage or export restriction for every item.

Dependency	Exposure to examine	Required continuity evidence
Dilution refrigeration	Cooling capacity, installation and specialist maintenance	Qualified alternatives, spares and service arrangements
Precision lasers	Required noise, stability and wavelength performance	Replacement qualification and long-term support
Isotopically tailored materials	Purity, reproducibility and production access	Specification, batch acceptance and alternate source
Photonic components	Loss, detector performance and packaging	Process qualification and reproducible assembly
Control electronics	Timing, noise, scaling and firmware access	Supported interfaces and replaceable implementation
Fabrication equipment	Access to suitable processes and tolerances	Production agreements and transfer feasibility
Vacuum and packaging	Reliability and reproducible integration	Environmental tests and manufacturing documentation
Decoder infrastructure	Throughput, latency and control-system integration	Performance at the intended logical scale
Calibration knowledge	Dependence on individual specialists	Documented procedures and trained replacement staff
Software infrastructure	Licensing, portability and update continuity	Data export, reproducible workflows and transition rights

[P] Japan’s ecosystem measures expressly include refrigerators, electronics, optical systems, detectors and control systems among component areas for policy attention.

Source: Cabinet Office, Quantum ecosystem measures, component and supply-chain sections, May 2025. www8.cao.go.jp

A practical sovereignty test should ask three questions for each dependency: who can alter it, who can reproduce it, and how long replacement takes. The answers should determine resilience measures and procurement rights.

8.18. Export controls: the legal choke point must be identified precisely

[P] The September 2024 US rule introduced quantum-computing controls including:

Classification	Scope described in the 2024 instrument
4A906	Specified quantum computers and related assemblies or components
4D906	Specified software for development or production of covered components
4E906	Specified technology for development, production or relevant software use

The rule combines controllable physical-qubit count with gate-error criteria and includes technical definitions and notes. It does not establish that every product marketed as quantum falls within the same classification.

Source: US Department of Commerce, Commerce Control List additions and revisions: advanced technologies consistent with international-partner controls, September 2024. federalregister.gov

[P] BIS also described exclusions and a general-licence mechanism concerning certain deemed exports and reexports, with reporting conditions.

Source: BIS, Department of Commerce Implements Controls on Quantum Computing and Other Advanced Technologies Alongside International Partners, September 2024. bis.gov

Recommended assessment: maintain a classification record at item and technology level, including the relevant transaction, recipient, destination and current consolidated provisions. The BIS Interactive Commerce Control List provides the current-reference entry point; the historical rule explains the introduction of the controls.

8.19. Standards influence and talent should be evaluated through outputs

A recommended standards scorecard should examine technical contributions, reference implementations, test methods, evaluation laboratories and adoption support. Committee membership alone should not be treated as demonstrated influence over a final specification.

A recommended workforce scorecard should distinguish scientific research from the engineering and operational skills needed to reproduce systems:

Workforce category	Capability to measure
Device scientists	Understanding and improvement of the physical mechanism
Fabrication engineers	Repeatable process and yield
Integration engineers	Assembly, packaging and commissioning
Control specialists	Calibration, feedback and performance maintenance
Software and decoder engineers	Reliable execution and classical-system integration
Validation specialists	Independent testing and uncertainty analysis
Field-service personnel	Repair, recalibration and sustained operation
Application specialists	Connection between the technology and user requirements

Analytical judgment: the industrial programme should fund these roles according to its intended outputs. A research workforce can establish scientific strength while leaving manufacturing or service continuity unresolved.

Chapter 8 — Key judgments

[P] National programmes are identifiable, but their financial instruments differ materially. The decision changed is how to compare investment without equating requests, commitments and expenditure. The US budget supplement and Indian implementation response provide useful accounting distinctions.

[R] Larger logical-qubit targets increase the verification burden; they do not resolve it. France’s revised PROQCIMA objective should be assessed through intermediate operational specifications.

Analytical judgment: the most useful sovereignty measures concern controlled dependencies, reproducible production and sustained access. Funding an installed system should be accompanied by evidence of the rights and capabilities retained.

What would change the assessment

Audited expenditure reconciled across programmes; qualified production and yield records; documented component substitution; completed acceptance tests; and evidence that industrial outputs remain supportable beyond their initial demonstrations.

Open official record

Comparable national expenditure, complete dependency maps, production yields and maintenance costs remain unevenly disclosed. These gaps prevent a defensible single ranking of national quantum power.

Chapter 9 — Five-Year Outlook, 2026–2031, Under AI Acceleration

9.1. Scenarios should branch on measurable conditions

The recommended outlook uses three branches:

• Constrained: component progress continues, but integrated reliability, manufacturing or operational adoption remains limiting.
• Base: several approaches establish useful integrated systems within defined operating envelopes, while fault-tolerant computing remains workload-limited.
• Accelerated: coordinated improvements in physical performance, error correction, classical control and production permit substantially deeper logical computation and wider adoption.

These are [R] analytical scenarios, not assigned probabilities. Their purpose is to connect decisions with evidence that can confirm or reject a branch.

Branch variable	Evidence needed to advance the assessment
Error correction	Repeated suppression under relevant operations and conditions
Logical register	Simultaneously usable logical qubits executing the required operations
Decoder	Sustained throughput, decision latency and integration into feedback
Manufacturing	Repeatable yield and commissioned-system consistency
Cryptanalysis	Updated complete resource estimates or reproducible attacks against named schemes
Application value	End-to-end comparison against a competitive conventional alternative
Institutional adoption	Acceptance, support, ownership and operational authorisation

9.2. Public roadmaps provide targets with different scopes

[R] The following targets are prospective and should retain their original definitions:

Roadmap or programme	Stated target	Relationship to 2026–2031
IBM Starling	2029, 200 logical qubits, 100 million quantum gates	Inside the window; performance remains prospective
Quantinuum Apollo	Updated company record identifies 2029 launch	Inside the window; complete application resources require verification
French PROQCIMA	1,024 logical qubits by 2032	Endpoint lies beyond the window
UK computing mission	2032 billion-operation intermediate milestone; 2035 trillion-operation objective	Both endpoints lie beyond the window
DARPA QBI	Assess utility-scale feasibility by 2033	Verification horizon extends beyond the window

Sources: IBM, Hardware and roadmap, maintained record; Quantinuum, Stage B announcement and roadmap update, November 2025; DGE, PROQCIMA update, May 2026; UK government, Quantum Missions, November 2023; DARPA, QBI. Hardware and roadmap

[R] Quantinuum’s September 2024 announcement used a “by 2030” framing and described circuits with millions of gates. The subsequent 2029 statement is an earlier target, rather than evidence of delay.

Source: Quantinuum, Accelerated roadmap announcement, September 2024. quantinuum.com

Recommended reporting rule: retain dated versions and record changes in metric, scope and delivery date. A roadmap revision should be classified accurately as acceleration, delay, increased ambition or redefinition.

9.3. AI-assisted decoding has demonstrated value, with an execution constraint

[D] The 2024 AlphaQubit paper evaluated neural decoding on experimental Sycamore surface-code data at distances three and five. It reported improved decoding accuracy relative to the evaluated comparators. Larger-distance results used simulation.

The paper distinguishes throughput from final-decision latency and states that its design had not been optimised for inference speed. Consequently, better retrospective decoding does not alone establish the feedback performance required for a long algorithm.

Source: Bausch and colleagues, Learning high-accuracy error decoding for quantum processors, Nature, November 2024. Nature

Recommended next-stage test: require the decoder to meet both a throughput requirement and a decision deadline under the intended code, noise regime and number of logical blocks. Average throughput should not conceal a persistent backlog or unacceptable latency tail.

9.4. A 2026 control result supports acceleration through calibration

[D] A July 2026 study applied reinforcement-learning fine-tuning to Google’s Willow processor after conventional calibration. It reported approximately 20% additional logical-error suppression in repeated distance-five surface- and colour-code runs. Its distance-seven surface-code result reached a logical error rate of $7.72(9) \times 10^{-4}$ using AlphaQubit2.

The paper distinguishes experimental repeated-shot control from simulated steering during a long logical computation. It also identifies fast drift that its current steering cannot follow.

Source: Sivak and colleagues, Reinforcement learning control of quantum error correction, Nature, July 2026. Nature

Analytical judgment: this is evidence for a specific acceleration mechanism—using error-detection information to improve control. It is not evidence that learning removes irreducible physical noise or supplies an entire fault-tolerant architecture.

9.5. AI should enter the outlook through identifiable bottlenecks

AI contribution	Mechanism to assess	Evidence classification	Ceiling that remains
QEC decoding	Infer corrections from syndrome history and analogue information	[D] Selected decoding results; [R] broader deployment	Latency, scale and physical-noise behaviour
Calibration and control	Search and adapt control parameters using measured outcomes	[D] Selected 2026 hardware result	Irreducible noise, fast drift and unsafe exploration
Materials and device search	Prioritise candidate structures or fabrication parameters	[R] Proposed acceleration pathway here	Fabrication, measurement and reproducibility
PQC cryptanalysis and implementation testing	Search for weaknesses in a named construction or implementation	[R] Scheme-specific continuation	Proof, attack cost and independent reproduction
Sensor fusion	Combine sensor streams, maps and uncertainty estimates	[R] Application-specific development	Measurement quality, map ambiguity and integrity

The recommended programme should measure the improvement at the bottleneck and then measure the complete system. An improvement in one stage can be useful without determining the total rate of progress.

9.6. Long computations require an explicit failure budget

These rows require a defined opportunity count. It may include logical operations, memory rounds, state preparation and other events. A published memory-error rate cannot be substituted directly for a complete logical-operation failure model.

[R] Surface-code resource analysis explicitly connects fault-tolerant operations with physical space and execution requirements.

Source: Fowler and colleagues, Surface codes: Towards practical large-scale quantum computation, published in Physical Review A, 2012. arxiv.org

Recommended interpretation: ask whether the processor meets the failure budget of the intended computation, rather than whether its logical qubits are better than its physical qubits under one experiment.

9.7. Decoder capacity requires two separate inequalities

Decoder indicator	Proposed management use
Sustained processing rate	Establish whether backlog remains bounded
Median latency	Describe ordinary operation
High-percentile latency	Test deadline reliability
Maximum tested block count	Establish demonstrated scale
Response to noise changes	Test robustness outside the training distribution
Resource consumption	Assess classical compute, power and cooling
Recovery behaviour	Establish what happens after interruption or overload

Recommended interpretation: a fast average decoder can still fail an operational requirement if a small fraction of decisions arrive too late. The scenario should advance only when both conditions are met at the intended scale.

9.8. Manufacturing yield can become a system-level constraint

The following is an illustrative manufacturing model, not an estimate of any supplier’s yield.

Recommended industrial test: request the actual acceptance model. The decisive data are qualified module yield, repair success, calibration time and the distribution of performance across commissioned systems.

Production indicator	Question it resolves
Device yield	How frequently fabrication produces an acceptable device
Module yield	How frequently assembly meets the specification
Commissioning time	How much specialist effort is needed to reach operation
Inter-system variation	Whether one demonstration generalises to production
Repairability	Whether failed parts can be replaced economically
Long-duration availability	Whether acceptable initial performance persists

9.9. Constrained scenario: progress continues without broad integrated capability

The numerical thresholds below are [R] planning tests selected for this report. They are not vendor promises or consensus forecasts.

Year-band	Error correction and logical register	Decoder condition	Manufacturing condition	Cryptanalytic condition	Resulting assessment
2026–2027	Ten simultaneously usable logical qubits with sustained universal operations remain difficult to reproduce across systems	Accuracy gains do not consistently meet feedback deadlines	Commissioning remains heavily specialist-dependent	Complete estimates continue to rely on unvalidated integration assumptions	Fund components and bounded experiments
2028–2029	Registers expand, but a 100-logical-qubit deep workload remains limited by cumulative error or operation cost	Backlog, latency or adaptation limits usable depth	Yield and calibration impede expansion	No demonstrated named cryptographic target; estimates remain conditional	Restrict adoption claims to measured workloads
2030–2031	A reproducible 200-logical-qubit system at the required depth remains unestablished	Classical processing remains an architectural constraint	Support costs limit replication	No complete operational attack chain established	Broad cryptanalytic capability is not assumed within this window

[R] Scenario consequence: sensing and security migration can still produce substantial institutional outputs because their delivery paths are separate from deep logical computation. Computing investment should remain tied to specific unresolved integration problems.

9.10. Base scenario: selected systems become useful within defined limits

Year-band	Error correction and logical register	Decoder condition	Manufacturing condition	Cryptanalytic condition	Resulting assessment
2026–2027	At least ten usable logical qubits support reproducible, bounded universal-operation experiments	Feedback succeeds for the demonstrated code and scale	Repeat commissioning produces similar performance	Resource models improve but retain physical assumptions	Expand well-defined research workloads
2028–2029	Systems in the approximately 200-logical-qubit class reach independently verified workload specifications	Sustained throughput and latency meet the relevant control budget	Qualified modules and replacement procedures emerge	Roadmap resources remain distinct from attack resources	Evaluate application value end to end
2030–2031	At least one deep logical application demonstrates reproducible value at its stated register size	Decoding scales without unacceptable classical overhead	Useful systems can be replicated and maintained	A named attack still requires its complete architecture and execution schedule	Adopt selected services; retain cryptanalytic uncertainty

[R] Scenario consequence: this branch permits early fault-tolerant services without assuming general-purpose economic advantage. The defining evidence is a completed workload with a meaningful conventional comparison.

9.11. Accelerated scenario: several bottlenecks improve together

Year-band	Error correction and logical register	Decoder condition	Manufacturing condition	Cryptanalytic condition	Resulting assessment
2026–2027	At least 100 logical qubits execute an appropriate universal-operation workload with documented success probability	AI and conventional decoding meet tested feedback deadlines	Repeated modules retain the demonstrated performance	Algorithmic improvements reduce complete resource requirements	Reassess the pace of integration
2028–2029	At least 1,000 usable logical qubits are available for deep computation, including necessary ancillary resources	Decoder capacity scales across simultaneous blocks	Production, interconnect and calibration meet system requirements	Complete named-scheme models become consistent with a credible architecture	Accelerate verification and security reviews
2030–2031	At least 2,000 usable logical qubits support the required deep workload; factory and memory resources are separately accounted for	Sustained operation meets throughput and latency requirements	Large systems remain stable over the required execution duration	Cryptanalytic capability becomes credible only if every target-specific resource condition is satisfied	Treat earlier arrival as a contingency requiring direct evidence

[R] Scenario consequence: this is a demanding conjunction of assumptions. A larger logical register alone does not activate the branch. Error budgets, non-Clifford production, interconnects, control and sustained operation must improve together.

These thresholds are designed to test institutional preparedness against faster progress. They should not be published as expected delivery dates.

9.12. Cryptographic relevance is a target-specific resource test

The previously established RSA and ECC estimates should be used as workload tests, without repeating their introductory resource tables here.

[R] Gidney’s RSA-2048 analysis explicitly includes distinct compute, hot-storage and cold-storage regions, state-production resources and physical timing assumptions. It also leaves detailed cold-storage workload analysis as future work.

Source: Gidney, How to factor 2048 bit RSA integers with less than a million noisy qubits, author preprint first submitted May 2025, retrieved version inspected October 2026. arxiv.org

A roadmap should pass the following mapping before receiving a cryptanalytic interpretation:

Attack-resource field	Required roadmap correspondence
Named scheme and parameter	Exact target represented by the estimate
Logical algorithm register	Simultaneously available algorithm qubits
Ancillary resources	Workspace, routing, preparation and verification
T/Toffoli requirement	Demonstrated production and consumption schedule
Error correction	Applicable code, distances and noise assumptions
Physical overhead	Complete allocation rather than one encoding ratio
Runtime	Timing, repetition and control delays
Success probability	Combined algorithmic and physical failure treatment
Sustained operation	Availability across the required execution period

[R] Assessment: the selected roadmaps examined here announce important prospective capabilities, but do not establish a verified operational attack against a named standard-sized cryptographic target by 2031. That finding does not prove such an attack is impossible within the window; it means its arrival cannot be treated as demonstrated or scheduled fact.

9.13. Migration deadlines can precede the adversarial computer

[P] Cryptographic transition instruments impose institutional work independently of a demonstrated adversarial quantum computer. The applicable US and UK records establish migration planning and implementation responsibilities.

Sources: White House, Executive Order 14412, June 2026; UK NCSC, Timelines for migration to post-quantum cryptography, March 2025. www.whitehouse.gov

Analytical judgment: the ordering is coherent because an owner must discover dependencies, replace incompatible equipment, preserve service continuity and protect information whose confidentiality extends beyond the transition period.

The recommended scenario response should therefore avoid tying migration funding to a forecast of a particular machine’s arrival. Slower computing progress may alter the assessed urgency of some exposures, but does not erase applicable requirements or make unmaintained cryptographic dependencies acceptable.

9.14. Branch-switching indicators

Observation	Recommended assessment change
Logical-error suppression stalls under representative operations	Move the computing outlook toward constrained
Decoder meets accuracy but misses control deadlines	Hold the scale assessment
Several commissioned systems reproduce performance	Strengthen the manufacturing branch
Deep application result survives a stronger classical comparison	Strengthen the useful-computing branch
Complete attack estimate becomes materially cheaper	Reassess cryptanalytic contingency
Widely deployed PQC implementation has a reproducible vulnerability	Initiate scheme- or implementation-specific remediation
Field sensor demonstrates mission value across environments	Advance that application independently
Programme funding is released but milestone delivery remains unclear	Review execution before expanding commitments

Chapter 9 — Key judgments

[D] AI has demonstrated improvements in decoding and control under specified conditions. The decision changed is whether to fund those bottlenecks with explicit latency, robustness and integration tests. See AlphaQubit and the July 2026 control study.

[R] Roadmaps inside the window describe different metrics and levels of integration. The decision changed is whether a target can be mapped onto a complete workload, rather than compared through headline qubit counts.

Analytical judgment: the base planning position should permit useful selected systems while retaining “not within this window” for a verified cryptanalytic service. Preparedness for the accelerated branch should come through migration and monitoring, without treating the branch as a forecast.

What would change the assessment

Independent deep-circuit validation; scalable low-latency control; reproducible manufacturing; complete named-target attack resources; or a useful application result that persists against improved conventional alternatives.

Open official record

The unresolved variables include simultaneous usable logical capacity, complete operation costs, decoder scaling, production yield and sustained availability. Assigning precise probabilities to the scenarios would exceed the evidence examined here.

Chapter 10 — Implications for Decision-Makers

10.1. Action matrix

The matrix is prescriptive:

• ACT NOW: undertake authorised implementation or readiness work.
• FUND AND WATCH: support a bounded experiment that resolves a material uncertainty.
• DO NOT BUY: withhold capability procurement until the required evidence exists.
• RETIRE THE CLAIM: remove an unsupported assertion from planning or communication.

Application or claim	Civil government	Defence	Critical infrastructure	Enterprise
PQC dependency discovery	ACT NOW: establish owners and inventory	ACT NOW: use applicable national-security profiles	ACT NOW: identify operational and supplier dependencies	ACT NOW: map services and installed verifiers
PQC integration	ACT NOW: test priority services	ACT NOW: test authorised interoperability	ACT NOW: preserve continuity and recovery	ACT NOW: coordinate platform and supplier transitions
Advanced timing	FUND AND WATCH: test delivered service	FUND AND WATCH: representative holdover and synchronisation	FUND AND WATCH: endpoint timing and continuity	FUND AND WATCH: only for defined requirements
Magnetic or inertial navigation	FUND AND WATCH: integrity and authorisation	FUND AND WATCH: representative contested conditions	FUND AND WATCH: route-specific trials	DO NOT BUY: universal performance claims
Gravity surveying	FUND AND WATCH: benchmark the survey product	FUND AND WATCH: blinded target trials	FUND AND WATCH: site-specific comparison	FUND AND WATCH: complete ownership-cost test
Biomedical quantum sensing	FUND AND WATCH: clinical endpoint and authorisation	FUND AND WATCH: defined medical requirement	DO NOT BUY: unsupported diagnostic benefit	DO NOT BUY: research result sold as clinical service
Fixed-link QKD	FUND AND WATCH: bounded assurance case	DO NOT BUY: absent applicable approval	FUND AND WATCH: availability and trusted-node review	FUND AND WATCH: incremental value against alternatives
Memory-based quantum networks	FUND AND WATCH: research infrastructure	FUND AND WATCH: defined experimental requirement	DO NOT BUY: claimed operational protection without evidence	DO NOT BUY: production-service claims from experiments
Rydberg RF receivers	FUND AND WATCH: complete receiver performance	FUND AND WATCH: waveform and platform trials	FUND AND WATCH: specific monitoring need	DO NOT BUY: range substituted for instantaneous bandwidth
Deep logical computing	FUND AND WATCH: verifiable workloads	FUND AND WATCH: mission comparator	FUND AND WATCH: bounded useful problem	FUND AND WATCH: total workflow value
Operational quantum radar	DO NOT BUY: absent system evidence	DO NOT BUY: absent representative acceptance	DO NOT BUY: unsupported range claims	DO NOT BUY: laboratory analogy sold as service
“Unhackable quantum link”	RETIRE THE CLAIM	RETIRE THE CLAIM	RETIRE THE CLAIM	RETIRE THE CLAIM
“AI solves fault tolerance”	RETIRE THE CLAIM	RETIRE THE CLAIM	RETIRE THE CLAIM	RETIRE THE CLAIM
Physical qubits equated with cryptanalytic capacity	RETIRE THE CLAIM	RETIRE THE CLAIM	RETIRE THE CLAIM	RETIRE THE CLAIM

[P] The standards and agency guidance provide the foundation for the security rows; [D]/[C] the experiments and trial reports in Chapters 6–7 provide the application boundaries.

Reference points: NIST FIPS 203, NSA guidance, DARPA RoQS, and DARPA QBI. csrc.nist.gov

10.2. A decision should identify the owner, the uncertainty and the exit condition

The recommended decision memorandum should contain six fields:

Field	Required content
Decision owner	Person or body authorised to accept the outcome
Intended outcome	Security function, scientific result or mission effect
Present evidence	Demonstration, instrument, roadmap or claim
Uncertainty to resolve	Question the next expenditure will answer
Acceptance condition	Measurable evidence required to continue or adopt
Exit condition	Circumstance requiring redesign, pause or termination

This structure permits ambitious work while preserving accountability. A programme can continue despite uncertainty when its next experiment is designed to resolve that uncertainty and the decision owner understands the remaining exposure.

10.3. Proposed first-year implementation sequence

The following periods are recommended management intervals, not statutory deadlines.

Interval	Work to complete	Reviewable output
First 90 days	Identify decision owners, priority exposures and active quantum commitments	Portfolio register with evidence tags and applicable authorities
Months 3–6	Define migration dependencies, pilot questions and supplier information requirements	Approved test plans and dependency maps
Months 6–9	Execute representative tests and inspect financial implementation	Results linked to acceptance criteria
Months 9–12	Decide expansion, redesign or termination	Updated investment and acquisition decisions

For cryptography, the sequence should be aligned with applicable mandates. For research, its purpose is to prevent indefinite continuation without a measurable question.

10.4. Expenditure should purchase evidence or an accepted service

Expenditure type	Recommended contracting focus
Research grant	Reproducible result and uncertainty resolved
Prototype contract	Integrated performance under stated conditions
Infrastructure investment	Access, utilisation, support and retained capability
Capability procurement	Accepted operating envelope and service continuity
Supplier-development support	Qualified production and replacement capacity
Migration expenditure	Verified transition of a defined security function

Recommended financial discipline: record capital expenditure, operating cost, specialist staffing and replacement obligations separately. A low initial equipment price should not determine the choice when calibration, support or infrastructure dominate the service.

10.5. Proposed portfolio indicators

Indicator	Numerator or measurement	Denominator or reference
Cryptographic inventory coverage	Assessed priority systems	Defined priority-system population
Migration completion	Services passing approved transition tests	Services scheduled for transition
Exception exposure	Unresolved exceptions weighted by consequence	Complete exception register
Demonstration reproducibility	Successful independent repetitions	Planned relevant repetitions
Manufacturing consistency	Commissioned systems passing specification	Commissioned systems tested
Operational availability	Time delivering the required service	Required service time
Evidence conversion	Experiments resolving their stated uncertainty	Experiments completed
Adoption conversion	Systems accepted by a named owner	Systems proposed for adoption

These indicators are intended to measure implementation. Publication counts, partnerships and announced investment can remain contextual measures, but should not replace them.

10.6. Capability contracts should preserve evaluation and transition rights

The recommended contract should define access to the data needed to verify performance, the accepted configuration, changes requiring re-evaluation and the rights needed to transition if the supplier or architecture changes.

Contract provision	Purpose
Performance-data access	Permit independent assessment
Configuration identification	Establish what was tested and accepted
Change notification	Prevent unsupported extrapolation after modification
Support commitment	Preserve service continuity
Calibration and maintenance documentation	Reduce dependence on undocumented expertise
Interoperability specification	Define supported peers and interfaces
Data portability	Preserve institutional work and evidence
Failure and recovery procedure	Establish a controlled response
Replacement or transition provision	Preserve options if the service becomes unsuitable

10.7. Public reporting should preserve epistemic status

A recommended public statement should name the experiment or instrument, describe what it establishes and retain the relevant limit. A technical milestone can be communicated strongly without upgrading it into an operational capability.

[H] Statements that turn a programme objective into a completed deployment, or convert a component result into a whole-system guarantee, should be removed from official reporting.

Chapter 10 — Key judgments

Recommended decision: act on cryptographic migration through the relevant authorities, fund bounded application trials, and make capability procurement depend on representative acceptance evidence.

Recommended investment rule: the next expenditure should purchase an accepted service, a reproducible production capability or an answer to a material uncertainty.

Recommended reporting rule: preserve the distinction between delivered capability and funded ambition in every decision memorandum.

What would change the assessment

Accepted service-level results, repeated production performance, clarified operating costs or evidence that a programme’s intended advantage no longer survives comparison with conventional alternatives.

Open official record

Portfolio-level spending and programme announcements are often easier to obtain than acceptance results and continuing costs. Acquisition authorities should make those latter records part of the delivery requirement.

Chapter 11 — Annexes

11.1. Glossary and terminology controls

The definitions below establish the report’s usage. They are not capability claims.

Term	Meaning in this report	Terminology to reject or qualify
Physical qubit	Physical degree of freedom used to encode quantum information	“Qubit” without identifying the physical or logical level
Logical qubit	Encoded information protected under a specified error-correction construction	Equating one logical qubit with a fixed universal physical overhead
Error-corrected memory	Protected storage of a quantum state under stated conditions	Describing storage alone as universal computation
Fault-tolerant operation	Operation designed to limit error propagation under stated assumptions	Assuming a complete machine from one protected gate
Fault-tolerant computer	Integrated system meeting the error budget of its computations	Undefined “fully error-free computer”
Code distance	Parameter associated with the protection of a specified code	Comparing distances across different architectures without context
Decoder	Classical process inferring an appropriate response from error information	Equating decoding accuracy with feedback suitability
Syndrome	Information obtained from checks used in error correction	Treating it as a direct measurement of the protected logical state
T/Toffoli resources	Non-Clifford computational resources counted under a stated model	Comparing them directly with an undefined total gate count
Magic-state production	Preparation and processing supporting relevant logical operations	Omitting its space, time and error costs
Quantum advantage	Better performance on a stated task against a specified comparator	Unqualified “faster than every classical computer”
Useful advantage	Advantage relevant to an intended outcome after full workflow costs	Inferring economic value from a benchmark alone
QKD	Quantum key distribution under a specified protocol and device model	“Unhackable communications”
QSDC	Quantum-secure direct communication under a stated construction	Treating it as interchangeable with QKD
PQC	Classical cryptographic constructions designed for quantum resistance	Describing PQC as requiring quantum hardware
KEM	Key-encapsulation mechanism	Calling a KEM the complete payload-encryption service
Digital signature	Mechanism supporting authenticity and integrity under a trust model	Equating signature protection with confidentiality
Trusted node	Intermediate node whose security forms part of the assurance assumptions	Claiming unconditional end-to-end protection without addressing it
Quantum repeater	Architecture for extending quantum-network functionality	Calling every trusted QKD relay a repeater in this sense
Quantum sensing	Measurement using an identified quantum mechanism	Assuming every sensor has the same maturity or operational advantage
Holdover	Timing performance after loss of an external reference	“GNSS independence” without a duration and error specification
Quantum radar	Label requiring an identified sensing and receiver mechanism	Assuming long-range target capability from the label
Crypto-agility	Controlled ability to replace cryptographic dependencies	Unrestricted algorithm negotiation
Sovereignty	Specified control over design, production, operation or continuity	Treating installation location as complete technological control

Foundational references: NIST FIPS 203, FIPS 204, FIPS 205, and Fowler et al., Surface-code architecture and operations. csrc.nist.gov

11.2. Claim-correction log

The statements below are assertions to test and correct; the table does not quantify their prevalence.

Assertion	Classification	Precise correction	Evidence reference
“A processor with thousands of physical qubits can recover a standard ECC key.”	[H]	Requires a complete logical and physical attack implementation, not a physical-qubit count	P-256 resource study
“A published PQC standard means the organisation has migrated.”	[H]	Publication specifies the algorithm; migration requires implementation and system verification	FIPS 203
“QKD removes the need for authentication.”	[H]	Authentication remains part of the assurance architecture	NSA QKD guidance
“A long QKD route is a quantum-repeater network.”	[H]	Trusted relaying and memory-based entanglement networking are distinct architectures	Joint European QKD position
“A noiseless simulation is a larger quantum-hardware demonstration.”	[H]	Preserve the distinction between simulated representation and executed hardware	mRNA conference record
“An airborne gravimeter survey demonstrates navigation without GNSS.”	[H]	The cited survey used GNSS support and measured gravity-survey performance	Iceland–Greenland campaign
“A naval clock trial establishes fleet-wide operational capability.”	[H]	A trial report requires subsequent performance specification and acceptance evidence	Royal Navy August 2026 trials
“AI has solved fault tolerance.”	[H]	Selected decoding and control improvements do not establish the complete architecture	2026 reinforcement-learning study
“The largest announced programme budget identifies the leading nation.”	[H]	Compare accounting status, period, outputs and controlled dependencies	US financial reporting categories
“A 2029 fault-tolerant roadmap is a scheduled cryptanalytic service.”	[H]	Match the roadmap to the named attack’s complete resources and sustained execution	IBM roadmap and RSA resource model

11.3. Uncertainty register

Uncertainty	Consequence for decisions	Evidence that would reduce it	Interim treatment
Complete logical-operation error budget	Limits credible deep workload size	Representative integrated operations	Retain workload-specific qualification
Simultaneous usable logical capacity	Limits roadmap interpretation	Execution with the claimed active register	Separate memory and computation
Non-Clifford resource production	Limits runtime and physical allocation	Sustained production with accepted error	Require explicit factory accounting
Decoder tail latency	Can constrain feedback	Relevant latency distribution at scale	Hold scale claims pending testing
Noise outside the evaluated model	Can invalidate resource projections	Representative correlated-error tests	Preserve model assumptions
Production yield	Determines replication cost and pace	Qualified multi-device and module records	Treat one device as insufficient
Calibration and maintenance burden	Determines sustainable availability	Long-duration service records	Include specialist costs
Critical-component continuity	Affects industrial resilience	Qualified alternative supply paths	Maintain dependency register
Comparative application value	Affects adoption and investment	Strong conventional benchmark	Fund bounded comparisons
Target-specific cryptanalytic resources	Affects contingency planning	Complete reproducible estimate or attack	Avoid a single arrival date
PQC implementation assurance	Affects present cybersecurity	Validation and reproducible testing	Maintain update and replacement paths
National financial comparability	Affects geopolitical rankings	Reconciled fiscal and programme series	Do not publish an aggregate ranking
Operational accreditation	Affects procurement authority	Named acceptance and approved scope	Retain trial classification
Financial execution	Affects programme completion	Released and utilised funds linked to milestones	Review implementation separately

11.4. Readiness recording template

Every application entry should preserve both axes:

Field	Required entry
Application	Specific service or mission
Mechanism	Computing, QKD, QSDC, PQC, symmetric cryptography or identified sensor
Evidence tag	D, P, R, C or H
Date and source	Publication and retrieval dates
Platform	Named hardware or infrastructure
Metric	Measured performance or explicit programme target
Conditions	Environment, comparator and assumptions
Technical readiness	Laboratory phenomenon; component; integrated prototype; limited field trial; series or accredited deployment
Institutional readiness	No owner; research programme; requirement; procurement; doctrine and logistics
Acceptance authority	Named body or unresolved status
Limitation	What the evidence does not establish
Next decision	Continue, test, adopt, redesign or stop

This template prevents technical results and institutional status from being collapsed into a single maturity score.

11.5. Source register — industrial programmes, scenarios and policy instruments

All records below were retrieved on 4 October 2026. Dates describe the source or instrument, rather than the date on which the claimed future capability will exist.

ID	Source	Date / status	Principal use
I-01	NSTC — NQI FY2025 budget supplement	December 2024	Federal financial categories and programme scope
I-02	UK — National Quantum Strategy	March 2023	Funding commitment and national framework
I-03	UK — National Quantum Strategy Missions	November 2023	Prospective mission milestones
I-04	European Commission — Quantum Europe Strategy	July 2025	Industrial coordination and policy objectives
I-05	European Commission — Quantum policy record	Maintained page	Flagship, EuroHPC and forthcoming Act
I-06	France DGE — national strategy acceleration	May 2026	Financial announcement and PROQCIMA revision
I-07	Germany — Handlungskonzept Quantentechnologien	April 2023	Framework and prospective milestones
I-08	Bundestag — action-concept committee discussion	May 2023	Reported financing categories
I-09	Germany — federal research and innovation report	2026	Transition toward usable systems
I-10	China — Fifteenth Five-Year Plan	March 2026	Quantum policy priority
I-11	Japan — quantum ecosystem measures	May 2025	Industrialisation and dependencies
I-12	Japan — strategy and cooperation register	Maintained page	Current policy and cooperation instruments
I-13	India — National Quantum Mission, Hindi parliamentary response	February 2026	Mission governance and targets
I-14	India — institution-level financial response	March 2026	Sanctioned, released and utilised funds
I-15	Israel Innovation Authority — national infrastructure call	July 2026	Multi-platform R&D infrastructure
I-16	Australia — State of Australian Quantum, progress	2024	Project support and programme activities
I-17	Canada — national strategy, French version	Official strategy record	Funding, talent and commercialisation
I-18	Italy — national quantum strategy adoption	July 2025	National coordination framework
I-19	US Federal Register — advanced-technology controls	September 2024	Introduction of specified quantum controls
I-20	BIS — control implementation announcement	September 2024	Deemed-export and reporting context
I-21	BIS — Interactive Commerce Control List	Maintained reference	Consolidated classification entry point
S-01	IBM — hardware and roadmap	Maintained record	Starling target
S-02	Quantinuum — accelerated roadmap	September 2024	Original Apollo framing
S-03	Quantinuum — Stage B announcement	November 2025	Updated roadmap statement
S-04	DARPA — Quantum Benchmarking Initiative	Maintained programme	Verification and value-versus-cost test
S-05	Bausch et al. — high-accuracy decoding	November 2024	AI decoder evidence and speed limits
S-06	Sivak et al. — reinforcement-learning control	July 2026	Hardware control improvement
S-07	Fowler et al. — surface-code computation	Published 2012	Fault-tolerant architecture
S-08	Gidney — RSA-2048 resource analysis	Preprint first submitted May 2025	Complete resource-model inspection

11.6. Source register — security and application evidence carried forward from Pillar II

This register consolidates the principal security and application records from the supplied block. Their original experimental conditions and numerical results remain in Chapters 5–7.

ID	Source	Date / status	Evidence role
C-01	NIST FIPS 203 — ML-KEM	August 2024	Final standard
C-02	NIST FIPS 204 — ML-DSA	August 2024	Final standard
C-03	NIST FIPS 205 — SLH-DSA	August 2024	Final standard
C-04	NIST — HQC selection	March 2025	Selection and prospective standardisation
C-05	NIST — PQC current record	Maintained page	Standards and candidate status
C-06	NIST — PQC FAQs	Maintained page	Symmetric-cryptography assessment
C-07	NSA — post-quantum resources	Maintained guidance	NSS instruments and QKD position
C-08	ANSSI — QKD guidance	May 2020	Scoped communications guidance
C-09	ANSSI — PQC follow-up guidance	2023 follow-up	Hybrid migration
C-10	Joint European QKD position	January 2024	Assurance and migration priorities
C-11	White House — Executive Order 14412	June 2026	Scoped transition instrument
C-12	European Commission — coordinated PQC roadmap	June 2025	Member-state coordination
C-13	NCSC — migration timelines	March 2025	Discovery and transition planning
C-14	Roetteler et al. — ECC resources	Published 2017	Logical-circuit resource model
C-15	Long et al. — QSDC fibre experiment	April 2022	Physical communication experiment
C-16	Knaut et al. — memory nodes in telecom fibre	May 2024	Quantum-network experiment
A-01	BIS — Project Leap phase 2	December 2025	Payment-system migration pilot
A-02	Toshiba — HSBC metropolitan QKD trial	July 2023	Named commercial-trial announcement
A-03	IBM Research — mRNA conference paper	August 2025	Hardware and simulation distinction
A-04	mRNA author preprint	May 2025	Version-specific research record
A-05	ORNL — trusted-node QKD at a utility	Published 2021	Field integration
A-06	Jensen et al. — airborne gravimetry	April 2025	Survey field evidence
A-07	Boto et al. — wearable MEG	March 2018	Human research prototype
A-08	Magnetic-navigation author preprint	April 2025	Author-reported field results
A-09	PTB — realisation of the SI second	Maintained record	National time-metrology function
M-01	Royal Navy — radar-clock trials	August 2026	Institutional trial report
M-02	Royal Navy — P2000 clock trial	June 2025	Named maritime trial
M-03	Dstl — atomic-clock development	February 2026	Unattended-operation report
M-04	DARPA — RoQS	Maintained programme	Robust sensor development
M-05	DARPA — RoQS first phase	August 2025	Prospective platform testing
M-06	Stray et al. — gravity cartography	February 2022	Outdoor subsurface experiment
M-07	DARPA — Quantum Apertures	Completed-programme reference	RF-receiver research
M-08	DARPA — spectrum-sensing team selection	2021	Programme targets
M-09	NATO — quantum strategy summary	January 2024	Alliance objectives
M-10	Canada DND — Quantum S&T implementation plan	March 2023	Defence research missions and targets

11.7. Quantitative provenance rules

Number type	Required treatment
Measured performance	Date, platform, conditions, uncertainty and comparator
Programme budget	Currency, period, accounting status and scope
Roadmap target	Named owner, publication version and prospective status
Resource estimate	Target scheme, architecture, code, overhead and runtime assumptions
Derived arithmetic	Formula and input values
Illustrative scenario	Explicit assumptions and analytical purpose
National comparison	Reconciled scope and avoidance of double counting
Operational claim	Trial, contract, acceptance or doctrine evidence

The manufacturing-yield calculation, utilisation percentages and failure-budget examples in this block are identified as derived or illustrative. They should not enter an empirical database without those labels.

11.8. Closing adjudication

Analytical judgment: the strategic task for 2026–2031 is to protect enduring information, convert bounded experiments into accepted services, and retain the industrial capabilities needed to reproduce those services. Progress in computing, communications and sensing should be evaluated through their separate technical and institutional requirements.

[R] Public roadmaps permit serious consideration of early fault-tolerant computing within the window, but do not establish a verified cryptanalytic service by its end. [P] Security-transition instruments already provide responsibilities that can be acted upon. [D] Selected experiments establish real technical progress, while leaving application-specific adoption questions to be resolved.

The resulting decision standard is demanding but usable: finance the work that resolves a consequential uncertainty, procure the service that passes its acceptance conditions, and preserve the distinction between what has been demonstrated and what remains an ambition.

---

Copyright of debuglies.com – Even partial reproduction of the contents is not permitted without prior authorization Reproduction reserved