What the three companies signed between 4 and 8 September 2026

Rigetti Computing and D-Wave Quantum both date their CHIPS and Science Act award agreements with the U.S. Department of Commerce to 4 September 2026. Rigetti entered the accompanying securities agreement on 8 September, with the share issuance expected the same day; D-Wave's filing states that it will enter its securities agreement and report issuance afterward. Both filings disclose a minority, non-controlling equity stake for Commerce as a condition of the award. Quantinuum announced finalization of its own $100 million CHIPS research award on 8 September; that release confirms the award and its manufacturing partners and discloses no equity, voting or issuance terms. All three sit inside a federal program of about $2 billion across nine companies announced in May 2026.

Why the instrument matters more than the amount

A grant closes when its milestones are audited. Stock does not close. Rigetti's agreement generally bars Commerce from voting the shares except on specified class-rights and business-combination matters, and D-Wave describes similarly narrow rights, so the department's leverage sits elsewhere: in transfer restrictions, registration rights, repurchase mechanics, intellectual property licenses, march-in rights and award remedies. The August 2025 Intel conversion is the ancestor of the mechanism and an imperfect precedent, because that stake is expressly passive, with no board seat and no governance or information rights.

Three public payments, three visibility regimes

Appropriated research money reports to Congress and federal equity reports through securities filings. A third payment is far harder to see: Cook County's Class 8 MICRO classification would cut the assessment rate on Chicago's South Works quantum campus from 25 percent to 10 percent for thirty years, an estimated $175 million reduction, subject to city and county review, with no community benefits agreement signed. This post reads the three instruments together, sets them beside the public auditing the science is already doing on itself, and asks what a public shareholder in quantum computing would need in order to know whether the position is working.

A 1999 result and a 2026 withdrawal, from the same company

NEC researchers at the Fundamental Research Laboratories in Tsukuba demonstrated the first coherent control of a superconducting qubit in a paper Nature received on 26 January 1999 and published on 29 April. That single-Cooper-pair box is the direct ancestor of the transmon circuits used at Google, IBM and Fujitsu. At the end of March 2026, NEC closed its own superconducting hardware program. Japanese outlets reported the shutdown between 4 and 7 September 2026, and NEC's press office declined to go beyond a line about continuing technical assessment.

What the 2022 Cabinet Office submission showed, and what NEC announced

On 6 December 2022 NEC filed a slide deck with Japan's Cabinet Office working group on practical quantum applications. Page 10 charts a four-qubit basic unit of superconducting parametrons scaling to 100 qubits and then past 1,000, for the quantum annealer. The hardware NEC went on to announce was an eight-qubit parametron annealing machine built with the National Institute of Advanced Industrial Science and Technology, offered to Tohoku University over the internet from June 2023. No public announcement of a 100-qubit NEC machine has been found.

Two European decisions from the same period point the other way

NEC judged the payback horizon too long. Days before the reporting appeared, the Novo Nordisk Foundation confirmed a 5,300-square-meter quantum chip plant for Copenhagen, and a European Commission official told a Parliament audience that the forthcoming EU Quantum Act "is not about budget." Three institutions, one question about how long to wait for quantum hardware to pay, three different answers.

Four documents issued between 4 August and 3 September 2026, read together in the week of 29 August to 4 September, each name a requirement their own sources cannot yet test

On 3 September 2026 the G7 Cybersecurity Working Group wrote public procurement into the transition to post-quantum cryptography, the same day a deployed-silicon preprint reported a class of ML-DSA defects that known-answer tests do not reach. On 2 September a United States congressional commission listed what the country lacks before an implantable brain-computer interface can reach a patient, with active reimbursement coverage on the list. On 27 and 28 August a quantum navigation supplier named the defense organizations it works with while a submission to the independent AUKUS Public Inquiry described quantum magnetic sensing as a threat to submarine stealth. And the Ninth Circuit's ruling of 4 August, attributing computer access to the user when an AI agent acts, met a research preview of 27 August in which agents operate laboratory pipettes.

The shape the four share, and where its cost lands

In each case the sources themselves identify a gap between the requirement and the capacity to meet or verify it: a conformance suite that a preprint shows cannot see a defect class, an approval in one jurisdiction with no coverage decision in another, adjacent sensing capabilities evaluated as a purchase by one office and as a threat by another, and an access ruling built on a record of screenshots where the laboratory example has none. The cost of the gap falls on whoever signs the document the requirement calls for. This long read sets the four side by side and draws one question per pillar for the people who will be asked to sign, with the founder's two War on the Rocks essays, on testing deployed systems and on a single accountable office at the National Security Council, holding the two ends of the answer for quantum governance.

Five research teams posted fault-tolerance preprints within two days, each on a different component of a protected computation

Between 2 and 3 September 2026 five preprints on fault-tolerant quantum computation appeared on arXiv from five research teams: two on computing with quantum low-density parity-check codes, and three on the encoding, the inter-module links and the decoding such a computation would need. Rahul Sahay, David Long and Vedika Khemani built a framework that puts logical Pauli, Clifford and non-Clifford gates on qLDPC codes in one homological language and used it to find constant-depth implementations of the full Clifford group on toric-code blocks and addressable CCZ gates in three dimensions. Tang, Duan and Wu showed how to run the nonlocal gates such codes need on a two-dimensional crystal of 512 ions without moving any of them, and simulated a logical error rate of one in a trillion. Hillmann, Eisert and Arzani lifted low-density lattice codes into the bosonic GKP setting. Schmidt and five co-authors halved the distillation distance needed for fault tolerance across modules joined by noisy Bell pairs. Liu, Zeng, Wu and Lao recovered discarded decoder samples to close up to 83 percent of the gap between practical and optimal decoding.

Where the gate count meets the migration calendar

A high-rate code stores many logical qubits in few physical ones; the open question has been what a gate on such a code costs. That cost is one of the quantities inside any estimate of when a cryptographically relevant attack on RSA or elliptic-curve keys becomes feasible. The Oratomic-led estimate of 30 March 2026 by Madelyn Cain and co-authors, 10,000 neutral-atom qubits for Shor's algorithm, rested in its authors' words on high-rate codes and efficient logical instruction sets, and Cloudflare cited that estimate among the developments behind its 7 April roadmap, which targets completion of its network migration in 2029. Mauritz Kop and Joseph Federici wrote in July that falling estimates make quantum readiness a present coordination problem with lead times of a decade or more, and these five papers are the week's work on the assumptions inside such estimates.

The question to put to any resource estimate

This post reads the five papers together and draws one diligence rule from them: a claim about how many physical qubits break an RSA or elliptic-curve key is only as good as the logical gate set it assumes, and that set is now being worked out paper by paper. Which of the five results a given estimate relies on is the question a committee can ask, and a reader of this site can check.

A twenty-qubit code carried two logical qubits, computed on them, and handed them to another code

On 2 September 2026 a Quantinuum team posted arXiv:2609.03194, reporting three experiments on the company's 98-qubit Helios trapped-ion processor with a [[20,2,6]] C4-Helix code: twenty physical qubits holding two logical qubits at distance six. Repeated error correction over twenty rounds gave a logical error of 4.6 × 10−5 per logical qubit per cycle, down from 2.1 × 10−4 for the smaller [[10,2,3]] code on the same machine. Randomized benchmarking of the complete Clifford group on the two logical qubits, with active correction between gates, gave 2.8 × 10−4 per two-qubit logical Clifford against 1.2 × 10−3 for unencoded ions. A chain-map CNOT into a 25-qubit distance-five surface code prepared a three-qubit GHZ state at a fidelity lower bound of 99.925 percent against 99.54 percent physical. None of the three figures relies on discarding failed runs; the Clifford and GHZ results beat their stated physical baselines, and the memory result improved on the smaller logical code.

What the authors say is still missing

The paper's discussion section states that no universal Clifford+T computation was implemented, that the non-Clifford resource states must be imported through the interface just demonstrated, and that the measured error rates sit around 10−4 where the early fault-tolerant regime the authors target begins at 10−6. Their simulations put that regime within reach once physical two-qubit infidelity improves by about an order of magnitude from the 7.9 × 10−4 reported in the Helios technical paper of 7 November 2025. Decoding for the memory experiment was done offline, and the confidence interval on the memory figure runs from 2.0 × 10−5 to 1.08 × 10−4.

The four numbers a proposal should carry

This post reads the result against the processor's own baseline and draws one diligence rule from it: a logical error rate is meaningful only beside the physical rate on the same machine, the physical qubits consumed per logical qubit, and the set of operations the code can perform while protected. This paper supplies all four and names what it has not done, which is the form a checkable error-correction claim takes.

An author of the 2025 advantage paper published the classical answer

On 12 March 2025 D-Wave published quench dynamics of spin glasses run on an Advantage2 annealing processor in Science, and estimated in its announcement that the same simulation would have taken the Frontier supercomputer at Oak Ridge nearly a million years and more electricity than the world uses in a year. On 1 September 2026 Roeland Wiersema, the fifth of the sixty-two authors on that paper and now at the Flatiron Institute's Center for Computational Quantum Physics, published a classical simulation covering all four of the experiment's problem graphs.

What the new paper reports, and on which instances

Using time-dependent variational Monte Carlo with a path-factorized correlator state, Wiersema reports final two-spin correlation errors on par with the processor for the sizes where converged reference data exists, and a relative two-spin correlation error of about 7.6 percent against the processor's data on a 72-spin biclique instance, a topology tensor-network methods handle least well. He notes that no other variational method is known to produce a correct state at that scale. The technical content is three numerical repairs: parallel tempering for slow Markov chains, blurred sampling for high-variance estimators, and an importance-weighted adaptive integrator.

The price of the result, stated by its author

The paper prices itself. The simulations took hundreds of GPU hours where the processor returns the same correlations in seconds, the largest instances of the original experiment stay out of reach, and an entire instance class is left untouched. Wiersema writes that the findings sharpen the question of quantum advantage without settling it, and his acknowledgments thank two D-Wave scientists for discussions during the work. We read the sequence from the March 2024 preprint through two classical preprints in March 2025, a D-Wave-led evaluation in August 2025 and two papers in 2026, and what it leaves for anyone judging a vendor's performance claim.

What DAS Photonics qualified in Valencia on 1 September 2026

ESA announced that DAS Photonics has completed qualification of the Engineering Qualification Model of the Faint Pulse Source, the optical source that generates the pulses QKDSat uses to distribute keys from orbit. The unit was qualified to European Cooperation for Space Standardisation requirements at the ESA-VSC testing facility in Valencia and handed to Redwire Europe. ESA describes it as the most advanced qualified faint pulse source with embedded security mechanisms in Europe for a satellite-based quantum key distribution system. QKDSat runs under the ARTES programme, with Honeywell Aerospace, Redwire Europe and DAS Photonics named as partners.

What Britain's NCSC says about QKD, and for whom

The National Cyber Security Centre's paper on quantum networking technologies, published on 5 August 2025 at version 1.0, states that the NCSC will not support the use of QKD for government or military applications, and that using a QKD system should not count towards assessments of data-in-transit security under its Cyber Assessment Framework. For other sectors it recommends that QKD should not be solely relied upon for generating and distributing keys. The objection is that QKD supplies a shared secret and no way to know who is at the other end. The NCSC's post-quantum migration timeline, published 20 March 2025, runs on standardised algorithms: discovery by 2028, highest-priority systems by 2031, everything by 2035.

Where the qualification chain and the accreditation chain part

Madrid's MadQCI network links 30 locations over more than 700 kilometres of fibre, including hospitals in the Vithas group and cryptography work with Banco Santander. Component-level security criteria for QKD modules exist. A published acceptance case for a complete satellite system carrying government traffic does not.

What Prof. Mauritz Kop told The Prode on August 31, 2026

The Prode, the independent interview platform founded by journalist Ritwij Raj, released its conversation with Prof. Mauritz Kop on the emerging quantum race and its implications for defense and national security. Kop, founder of the Stanford Center for Responsible Quantum Technology and of Quentir, answers six questions now facing the US national-security and technology-policy communities — from the timeline for a cryptographically relevant quantum computer to the shape of government-industry collaboration. The full video is embedded in this article.

Q-Day math and the deadlines that already exist

Nobody can date Q-Day, and Kop declines to invent one. What changed this March is the engineering target: two papers lowered the resource estimates for machines running Shor's algorithm — one neutral-atom architecture at 10,000 reconfigurable qubits, a Google-led estimate under 500,000 physical qubits for the P-256 curve. His planning instrument is Mosca's theorem, and his planning horizon is public: NIST's 2024 standards, NSA's January 2027 acquisition requirement and 2035 completion date, the June 2026 executive order with 2030 and 2031 milestones, and the UK and EU tracks toward 2035.

Harvest now, decrypt later — and the data that cannot be reissued

Adversaries can store ciphertext today and read it once hardware matures. Kop extends the harvest-now-decrypt-later problem to its sharpest case: a password can be changed, a genome cannot, and it carries information about your relatives as well as yourself. His prescription is anticipatory data stewardship — if future decryption is foreseeable, it belongs in today's duty of care.

Sensing, deterrence by denial, and the golden triangle

On quantum sensing, Kop ranks the public evidence: military value appears first in positioning, navigation, and timing when GPS is jammed, while ocean-transparency claims outrun public evidence. The deterrence logic he draws is denial — guaranteed navigation and timing, quantum-secure command and control, and communications that stay up under attack. And for the valley of death between research grant and purchase, his answer is the golden triangle of academia, policy, and industry: a real mission, a first paying customer, and a test an independent team can challenge — the program he carries forward from Stanford to CIGI, the US Air Force Academy, and Quentir.

What the Ministry of Health and Welfare selected on 31 August 2026

Korea's health ministry, working through the Korea Health Industry Development Institute and its K-Health Future Promotion Team, selected a consortium led by Kyung Hee University's Son Seok-kyun, a physicist, for RFP5 of the Korean ARPA-H project: quantum sensing-based ultra-high-sensitivity early diagnosis technology. The award is KRW 8 billion across four and a half years. RFP5 was one of nine challenges opened on 1 June 2026 for applications closing 1 July. The design fuses a biocompatible molecular quantum nanosensor with a cardiac organoid organ-on-a-chip, and the entry indication is anticancer-drug-induced cardiotoxicity.

Why the funder's identity changes what the instrument has to prove

What Korea's health ministry announced is a clinical endpoint: damage a cardio-oncology clinic currently detects through echocardiography and blood biomarkers, after the muscle has already been affected. The consortium proposes to read intracellular oxidative stress, local temperature and metabolic activity upstream of that point. Six institutions hold six named pieces of the work, from spin-Hamiltonian signal interpretation at Chosun University to MEMS and microfluidic integration at KETI and NV-center cross-validation at Korea University.

Korea has now funded both halves of quantum medicine, through two ministries

On 19 August this blog covered a quantum drug-design programme funded by Korea's science ministry. This award comes from the health ministry, and it funds the measurement half of the same problem. Two ministries, two verification cultures, one national portfolio. The assessment routes for a device with no predecessor already exist, through FDA De Novo classification and European MDR/IVDR conformity assessment. What no authority has yet issued is quantum-specific assessment guidance, or a precedent decision on a diagnostic claim resting on a quantum-sensed intracellular measurement.

What Bharti, Haug and Tanggara posted on arXiv on 26 August 2026

A forty-page preprint, Fault-tolerant quantum computation cannot be achieved with constant spacetime overhead (arXiv:2608.26272, submitted 26 August 2026 at 18:00:25 UTC), proves a lower bound on the resources quantum error correction requires. Kishor Bharti, Tobias Haug and Andrew Tanggara show that for the simplest task in the field, holding quantum information steady in a memory, the minimum number of physical storage locations scales as Θ(S(K + log(S/ε))) for width K, duration S and target error ε. The second term is an additive reliability cost that no protocol removes inside this model. The result is filed under quantum physics and information theory, and it has not yet been peer reviewed.

The condition the authors attach to it in the same paper

Relative overhead scales as Θ(1 + log(S/ε)/K), so it stays bounded once the width K reaches Ω(log(S/ε)), and the logarithmic term becomes negligible only in the stronger regime K = ω(log(S/ε)). The absolute reliability cost is never removed at any width. The authors state the consequence directly. Constant overhead can be possible for sufficiently wide computations. They add that no single constant bounds the spacetime overhead uniformly over all widths. Standard implementations of Shor's algorithm amortize the cost. Grover search sits near the crossover.

Why the qualification matters more than the headline

A reader who takes only the title away will conclude that a hard limit has been placed under every claim about when a cryptographically relevant quantum computer arrives. The paper does not support that reading, and its memory bound is proved under a deliberately generous erasure model. This read separates what the theorem establishes from what it has already been asked to carry, names the width and duration under which the cost dominates, and states what it changes for federal migration schedules, which is nothing.

A $290 million announcement, read against its own awards

On 25 August 2026 the US National Science Foundation announced more than $290 million across eight Quantum Leap Challenge Institutes — $28 million to $37.5 million each over five years, three of them new and five renewed, spanning 36 higher-education institutions in 19 states with the Department of Energy national laboratories, NIST and the Department of War as federal collaborators and more than 30 companies alongside. The number travelled as a marker in the quantum computing race. The award titles read differently.

What the money is actually aimed at

Princeton's new institute, MARQUIS, receives $27.9 million over five years to reinvent one component: the Josephson junction, two superconducting electrodes separated by an oxidized insulating barrier only a few atoms thick, whose dominant construction — aluminum with an aluminum oxide barrier — Princeton describes as essentially unchanged for more than 25 years. Yale's PRACTIQAL takes practical error correction. Chicago's QuBBE takes quantum sensing for biophysics. Colorado's Q-SEnSE takes $37.5 million for atomic clocks, the most stable lasers in the world, and sensors that read disease in a patient's breath. Not one of the eight is denominated in qubit count.

Two states, one week, opposite ends of the curve

In the same week India's C-DOT unveiled fourteen indigenous quantum-secure products, which its chief executive characterized as production-grade with revenue already booked. One state announced a product line; the other funded the layer underneath it. Both answers are defensible and they are answers to different questions — and only one of them has products to point at this year.

A refrigerator result, precisely dated

On 19 August 2026 IBM said it had connected and cooled its first two modular cryogenic systems into a single environment. Assembled, the two box-shaped modules stand more than eight feet tall and eight feet wide. They reached 4 kelvin in under five days and then went below 15 millikelvin, and each module’s vacuum enclosure offers up to twelve times more wiring space than IBM’s most widely used quantum systems — room sized for the chip-to-chip connections its L-coupler interconnect is meant to carry.

What was not announced

IBM reports no installed processor and no qubit result for this test. The release carries no qubit counts, no fidelities and no error-correction figures for the coupled pair, and IBM Quantum Nighthawk processors are due to go into the cells later in 2026 for operational testing. Filed accurately, this is a cryogenics and packaging result under a fault-tolerance roadmap, and it sits at the layer where long roadmaps either hold their dates or quietly slip.

The calendar it runs beside

IBM’s stated targets are L-coupler-linked processors totaling at least 1,000 programmable qubits by 2027 and Starling, billed as the first fault-tolerant quantum computer, in 2029. Beside those company dates runs a standards calendar that moves with neither: NIST’s draft transition report IR 8547 deprecates classical public-key algorithms at 112-bit security strength after 2030 and disallows them after 2035. Reading the two together is where quantum readiness stops being a posture and becomes a schedule with dates on it.

Two governments, one control point

On 31 July 2026 China’s State Council published the Exit and Entry Administration Provisions, nineteen articles that take effect on 15 September. One of them lets competent State Council departments bar a citizen from leaving the country where that person violates export control or technology import/export rules in a way that “may endanger national industrial or technological security.” Seventeen days later, on 17 August, the Office of the Under Secretary of War for Research and Engineering notified thirty American universities to audit their institutional ties to foreign entities listed under Section 1286 of the fiscal 2019 defense authorization, with mitigation — up to terminating the partnership — reported by 31 August or future federal research funding is at risk. The same day, the White House published a national security science and technology strategy whose protected-technology list names quantum information technologies.

Why the wording matters

American law has treated a person as a channel for export since the deemed-export rule, so the direction is not new; a fundamental-research carve-out has kept university science largely outside it. What is changing sits at the edges of that carve-out. One instrument conditions a university’s research funding on the foreign affiliations in its own files; the other attaches an exit consequence to an individual on a forward-looking standard, with no fixed time limit where other categories of Chinese exit ban run six months to three years.

An unusually legible test case

Neither measure singles out quantum: the strategy lists it as one of fourteen protected areas, and the Chinese trigger is written around export control generally. Quantum research is simply where the change is easiest to observe, because the field is small, concentrated in a countable number of laboratories, and heavily co-authored across borders. A collaboration agreement signed in 2024 now carries exposure on both ends, and an affiliation cannot be recalled the way a shipment can.

A program that publishes its verification loop

On 19 August 2026 Korean industry press reported that Baobab AiBIO had been named lead research institution of a Ministry of Science and ICT / National Research Foundation program applying quantum computing to targeted protein degradation, with LigaChem Biosciences, Yonsei University and the Institute of Molecular Design as co-institutions. The part worth reading twice is the workflow the program describes: design, then synthesis and evaluation, then Cryo-EM structural work, then redesign. Quantum-assisted molecular design is being pointed at a wet lab instead of at a benchmark table.

What that does and does not settle

Taking predictions into synthesis is not the same as demonstrating quantum advantage. That claim would additionally need a declared classical comparator, a stated endpoint and a resource comparison, and none of those is public yet. Nor is wet-lab work unprecedented here: a 2025 Nature Biotechnology paper reported fifteen molecules designed with a quantum-classical generative model, synthesized and assayed, two of them promising against KRAS. The Korean program adds public money on the testing half of the loop and a named synthesis partner.

The infrastructure running alongside

IBM joined and cooled its first two modular cryogenic units below 15 millikelvin the same week, on a roadmap to at least a thousand programmable qubits in 2027. That is parallel context and not a dependency: the Korean work names Yonsei's existing IBM QPU environment. For any reader judging a quantum-medicine claim, the useful test is whether a check has been scheduled.

A four-year objection, answered outdoors

On 6 August 2026 Optica published a result whose lead author, Cheng Li, says has “met with repeated doubt and pushback” since the project began: polarization-entangled photon pairs generated with natural sunlight as the pump, measured outdoors, at a Bell-state fidelity of 0.939 and a CHSH value of 2.5408 against the classical bound of 2. The apparatus is a Fresnel collector about the size of a household window, a custom all-glass concentrator cone, a fiber about as wide as a human hair, and a ppKTP crystal at the end of it. The same lead author had shown in 2022 that an incoherent LED could pump weak entanglement; a Xiamen group showed in 2025 that sunlight could drive down-conversion at all. This joins those two records.

A substitution, not yet a subtraction

The energy overhead of the pump laser is the motivation the authors name, and running a source on light that is already falling on the instrument is attractive for exactly that reason. What the experiment swaps in, though, is a collector, a tracking mount, spectral filtering, a custom concentrator and crystal temperature control, and no published comparison yet shows that trade winning on total mass, power, thermal load or reliability.

What is not yet on the record

The preprint reports roughly 1,600 pairs per second per milliwatt, comparable to laser-pumped setups once normalized for effective phase-matching bandwidth. What is missing is mission-level throughput, a secure key rate, a sustained duty cycle across sky conditions, and an independent replication. Those are the numbers to watch before anyone writes a passive entanglement source into a space-segment work package.

Nine non-binding letters, not nine closed deals

On 21 May 2026 the US Department of Commerce announced nine letters of intent worth $2.013 billion in CHIPS incentives across the American quantum computing industry — $1 billion to IBM for a 300-millimetre quantum wafer foundry in Albany, $375 million to GlobalFoundries, and $100 million each to Atom Computing, D-Wave, Infleqtion, PsiQuantum, Quantinuum and Rigetti, with up to $38 million to Diraq. Attached to each is a condition that changes the instrument: in exchange for the award, each recipient would issue equity securities to the Department. The letters are non-binding, and the definitive documents are still unsigned.

What one filing answers

Commerce described every stake as minority and non-controlling, and said little more. Several recipients have since added something of their own — D-Wave that it would issue $100 million in common stock, GlobalFoundries that the Department's position is about one per cent. Quantinuum's prospectus goes furthest by a distance, describing securities that would be non-voting so far as the law permits and freely transferable, carrying anti-dilution, registration, redemption, exchange, conversion, participation, tag-along and information rights, with funding released against named milestones across a five-year performance period. For that one company the control question is largely designed out. The custody question is not, and no recipient has answered it.

Own, screen, or buy

Three governments reached for three different instruments this season. Washington moved toward ownership. Brussels chose control over ownership: Regulation (EU) 2026/1386 puts quantum technologies into the mandatory screening scope of all 27 Member States for the first time. Israel chose procurement, tendering a sovereign quantum computer in August. Ownership is the instrument whose internal governance is least visible, and it is the one the United States picked in a field whose cost assumptions moved again in July.