Four CHIPS Quantum Awards Closed. Their Public Scopes Converge on the Manufacturing Layer.

The U.S. Department of Commerce announced on September 8 that it had closed CHIPS and Science Act research awards with four quantum computing companies: D-Wave Quantum, Rigetti Computing, Quantinuum and PsiQuantum. Each carries a ceiling of up to $100 million, or up to $400 million across the four. Three agreements list an award date of September 4. Commerce's release on PsiQuantum gives no award date.

A ceiling is not a payment. The money moves in tranches released against milestones, and in each of the three disclosed arrangements Commerce takes stock in the recipient as a condition of the award. What Commerce executed are the definitive documents behind the letters of intent it announced in May, and no new appropriation was involved.

How many CHIPS quantum awards did Commerce close?

Quartz counted three companies and $300 million, framing the announcement around D-Wave, Rigetti and Quantinuum. The PsiQuantum finalization posted to the same NIST news series on the same day, at the same ceiling, to another quantum computing company; SCN counts four, and up to $400 million. GlobalFoundries closed a separate award of up to $375 million that day for a domestic quantum foundry. That one sits outside the total here because a fab is not a quantum computing company, though it returns below.

When SCN covered the letters of intent in May (nine of them, $2.013 billion, federal equity in every recipient), the piece closed with an open question about what "still requires final deal completion" would mean. Part of the answer is now on file. Three of the four recipients are SEC registrants and have filed current reports; D-Wave has also filed its award agreement as an exhibit, with the statement of work withheld. For PsiQuantum, the only public document is Commerce's summary.

What the public descriptions emphasize

Less of this paperwork is readable than the page count suggests. D-Wave's Other Transaction Agreement is filed as an exhibit to its Form 8-K, all 51 pages of it, and Attachment 1, the Statement of Work, appears in that exhibit as a single redaction mark. Quantinuum's 8-K lists three research activities and says the project consists of work related to those items "among other things," with the award agreement itself due as an exhibit to the company's third-quarter 10-Q. Rigetti's 8-K summarizes three project tasks, and it omits portions of the underlying agreement. PsiQuantum's award terms are not publicly disclosed in the materials reviewed here; the company is privately held, and the only filing of record SCN located at the SEC is a Form D from May 2025.

Together, these documents establish what the awards are described as covering. They do not establish what the awards leave out, and no reading of them supports a claim about absent work. What they do show is four companies on different hardware roadmaps whose federally funded work is described in nearly the same vocabulary.

Rigetti names three tasks: "(i) compressing readout electronics into an integrated, miniaturized package, (ii) expanding cryogenic capacity by orders of magnitude using a new cryostat architecture, and (iii) developing the fabrication capabilities for high-connectivity chip architectures." The middle task hits a limit that chip design cannot overcome. A superconducting processor sits at a few millikelvin inside a dilution refrigerator, and its control and readout signals arrive on wiring run down from room temperature through the cold stages. Each line carries heat with it. In a 2019 analysis of cryogenic setups for 100-qubit-scale superconducting systems, Krinner and colleagues at ETH Zurich treat that wiring heat load, set against the refrigerator's cooling power, as a governing constraint on how large a system one cryostat can host. Of the four recipients, Rigetti is the only one whose public description proposes rebuilding the cryostat rather than working inside one.

Quantinuum's award, per its Form 8-K, covers "(i) the development of low-loss integrated photonics at 422 nanometers, (ii) the prototyping of control application-specific integrated circuits (ASICs) on a high-voltage process for cryogenic operation and (iii) the development and packaging of reliable optical components at trapped-ion critical wavelengths." Trapped-ion machines move ions around a chip with electrode voltages and address them with laser light. Two scaling challenges cataloged in the standard review of the field by Bruzewicz and colleagues in Applied Physics Reviews are integrating the classical control electronics that generate those voltages and delivering laser light to many trap zones without an ever-larger optical table. Both appear in Quantinuum's list. Monarch Quantum's chief executive, Timothy Day, put the second one in his own words in Quantinuum's release: the road to large-scale trapped-ion machines "relies on moving away from complex, sprawling optical setups to scalable, reliable integrated photonics engines."

That release names the partners. GlobalFoundries will be "one of several" foundries fabricating Quantinuum's next-generation ion traps and other electronics on 300mm wafer technology; Monarch is to develop and manufacture lasers and optical components. Neither relationship is new. Day described the award as expanding an existing partnership, and SCN reported in August on a separate Quantinuum manufacturing agreement whose production terms were never disclosed.

Commerce's summary of the PsiQuantum award reads as the photonic version of the same list, addressing "mature and high-performance electro-optic materials, high-temperature single-photon detectors, and ultra-low-loss photonic packaging." Electro-optic materials change their optical properties under an applied voltage, which is how a photonic machine switches light on the timescales it needs. Photonic architectures are often described as escaping cryogenics, though their single-photon detectors generally are not, so raising detector operating temperature eases the refrigeration burden. Past that summary, the terms of PsiQuantum's award are not publicly disclosed in the materials reviewed here. The $100 million ceiling is the only number available.

The D-Wave documents most clearly show that the component description and the systems description refer to one project. The NIST release describes an award funding "the critical advancement of semiconductor process technologies for annealing and gate-model superconducting quantum computing systems, including qubit counts, error rates, and coherence through advanced dielectric material optimization, interface control, and high-density advanced packaging." Qubit counts and error rates are named there directly. So is the proposed route to them: the description pairs coherence with dielectric material optimization and interface control, which, on a plain reading, is the material and the figure of merit it is meant to move, written into one sentence. D-Wave's own press release states the targets at the top of that stack: a 100,000-qubit annealing system and a 10,000-qubit gate-model system "designed to enable 100 logical qubits capable of successfully performing more than one million operations." Its 8-K states the milestones at the bottom, tying tranche releases to tool installation, prototype fabrication, process integration, and the calibration and benchmarking of scaled processors. Logical-qubit counts are the units this field uses to state ambition, including in the DOE's 2028 fault-tolerance goal. A schedule of tool installations is what the first stretch of pursuing one looks like once it has to be audited.

The statutory plumbing points the same way. D-Wave's agreement cites 15 U.S.C. § 4656(d), under which the Secretary of Commerce "may make financial assistance awards, including construction awards, to strengthen semiconductor advanced test, assembly, and packaging capability in the domestic ecosystem," alongside the section establishing the national semiconductor technology center and the other-transaction authority at § 4659(a)(1). The packaging emphasis in these descriptions is written into the authority under which the award was made.

Sites, ownership and where things are made

The D-Wave agreement is the only one of the four that names its project sites, and the list separates three things that get run together in coverage of this program. The defined Project Sites are D-Wave's existing facilities in Palo Alto, New Haven, and Burnaby, British Columbia; the company's facility under construction in Boca Raton, Florida; SkyWater Technology's fab in Bloomington, Minnesota; SEMICAT's facility in Fremont, California; and NASA's Jet Propulsion Laboratory in La Cañada Flintridge, together with any other location the Department agrees to in writing.

One of those sites is in Canada, inside a federal award whose Attachment 5 is titled "Domestic Control and Production." D-Wave's most recent 10-Q lists four leased operating facilities in North America. Two are in British Columbia, at Burnaby and Richmond; the others are in Palo Alto and New Haven. The same filing says the company plans to move its corporate headquarters from Palo Alto to Boca Raton before the end of 2026 and open a US R&D facility there.

Attachment 5 attaches to intellectual property and to manufacture, not to where a research task is performed. Federally funded IP must stay owned by a domestic entity for the period of performance and at least ten years after it. Over the same window, it cannot be sold, transferred, licensed, or assigned to a foreign adversary. And subject inventions must be predominantly produced or licensed for production in the United States, waivable case by case on a showing that domestic manufacture is not commercially feasible. Where the research happens, who owns the IP, and whether a domestic supply of a given component exists are three separate questions.

One of those sites changed hands this summer. IonQ completed its acquisition of SkyWater Technology, whose Bloomington fab the agreement names as a Project Site, on July 31, five weeks before the September 4 award date. SCN has not sought comment from D-Wave, IonQ or Commerce, and draws no inference from the sequence.

The instrument, and how the equity works

All four awards rely on other-transaction authority, an instrument decades old at DARPA and increasingly common at Commerce, rather than standard grant authority. The three disclosed agreements run from the September 4 award date to the earlier of milestone completion or the fifth anniversary, and each allows for extension. D-Wave's period of performance is "subject to extension pursuant to the terms of the OTA." Rigetti may request an amendment extending it, with sixty days' notice. Quantinuum may request the same, and Commerce agreed not to withhold consent unreasonably. PsiQuantum's terms are not public, and none of these schedule provisions can be assumed to apply.

The filings add the disbursement schedule. D-Wave has $53,552,620 available shortly after the award date, then $9,075,000, $16,695,000, $20,390,000, and $287,380 as milestones are cleared. Rigetti has $43.9 million available on or as soon as practicable after the award date, then $29.9 million and $26.2 million, released only if Commerce "determines to its satisfaction in its sole discretion" that the milestones were met. Quantinuum's runs $56.0 million, then $32.0 million, then $12.0 million. PsiQuantum's is not public. Across the three that disclosed, $153.5 million is scheduled to be available near the award date, and the balance is contingent.

The difference between a ceiling and a payment is not academic for the recipients. All three public companies are selling systems and services and losing money doing it. For the six months ended June 30, 2026, D-Wave reported revenue of $5.9 million and a net loss of $66.4 million. Rigetti reported revenue of $9.5 million and a net loss of $19.5 million. Quantinuum reported revenue of $13.2 million and a consolidated net loss of $733.1 million, a figure dominated by $447.5 million of stock compensation expense recognized around its June listing; the portion of that loss attributable to Quantinuum Inc., as distinct from the non-controlling interest and the pre-listing predecessor, was $65.4 million. All three are compared here on consolidated net loss, the line each reports at the foot of its statement of operations.

The three securities agreements are close variants of one another, and they differ enough that summarizing them as a single arrangement would misstate at least one. In each case Commerce's ability to transfer shares is capped at the proportion of award funds drawn and not returned. In each case, the company may repurchase the shares corresponding to the undisbursed remainder for one dollar in aggregate if Commerce terminates: for convenience, in D-Wave's and Quantinuum's terms; on termination by the Department, in Rigetti's summary. Rigetti's agreement lets Commerce demand recovery of payments made as "a debt payable to the Department" if the company fails to complete a required project activity. Quantinuum's clawback covers research security, domestic control of intellectual property, domestic production, and failure to complete required project activities on time. On voting, D-Wave's and Rigetti's terms both carve out amendments adverse to the common stock and any merger or similar business combination. Quantinuum's does not: its exception covers changes to the authorized number of Class A shares, the par value, or the powers, preferences, or special rights of the Class A stock, and stops there.

The corporate structures differ too. Rigetti's awardee is Rigetti & Co, LLC, a subsidiary, while the shares come from Rigetti Computing, Inc. Quantinuum's awardee is Quantinuum LLC, and the shares are Class A stock of Quantinuum Inc. D-Wave contracted directly. Reading these awards by instrument rather than qubit technology is the same approach SCN took for Europe's quantum calls, which sort hardware by funding instrument as much as by qubit type.

The stakes, computed

No recipient disclosed a percentage. The figures below are SCN calculations, not disclosed figures. Each divides the shares going to Commerce by the sum of the company's most recent reported share count and those same shares, which gives the stake after assumed issuance. The reference counts are the cover-page figures from each company's second-quarter Form 10-Q, on the dates shown.

Two limits apply to those numbers. All three are common-share calculations, not comprehensive measures of economic ownership. And D-Wave's leaves out a second security: the company separately reports 3,176,096 exchangeable shares outstanding as of August 5, convertible into common stock one for one, at any time, for no consideration. Folding those into the denominator gives 7,095,721 ÷ 379,535,654, or 1.87 percent. That still rounds to 1.9.

D-Wave's status is the easiest to misread. Its 8-K says the company "will enter into a Securities Issuance Agreement" and that it will file an amendment "reporting the issuance of the Shares" once that agreement is executed. As of the filing, this is an agreed future issuance at a set price and share count, not stock transferred or a stake held.

Quantinuum needs its structure explained rather than a percentage quoted. Quantinuum Inc. listed on Nasdaq on June 5 at $60.00 a share and holds its business through Quantinuum Holdings, LLC. Class A carries one vote and economic rights. Class B carries one vote and, in the 10-Q's words, "is cancellable upon the redemption or exchange of one Common Unit" and "has no economic rights." There were 226,414,285 Class B shares outstanding on August 13 against 36,893,857 Class A.

Commerce's economic interest can be computed as of a single date. The certificate of incorporation and the Holdings LLC agreement require the company to maintain, at all times, a one-to-one ratio between the common units Quantinuum Inc. owns and its Class A shares outstanding, and a one-to-one ratio between Class B shares and the common units held by the continuing common unitholders. On the August 13 counts, those ratios imply 263,308,142 common units in Quantinuum Holdings. Add Commerce's 2,369,528 Class A shares and the units the ratio requires against them, and Commerce's look-through interest in the operating business works out to roughly 0.9 percent, against 6.0 percent of the listed class.

That figure rests on two assumptions, both of them SCN's. The first is that the one-to-one ratio held on issuance. The second is that nothing else moved between the August 13 count and the September 8 issuance. The 10-Q discloses one movement already, in July: 357,592 Class A shares issued to Holdings members redeeming an equal number of common units, with the matching Class B shares canceled. That leaves the unit total unchanged, and the August 13 counts already reflect it. A precise number would need a direct check with the company.

Pricing was disclosed unevenly. D-Wave's shares are priced at $14.093, struck off the lowest closing price across three reference dates and discounted 15 percent. Rigetti's filing states an implied issuance price of $12.92. Quantinuum's 8-K states no issuance price at all; $100.0 million across 2,369,528 shares works out to roughly $42.20, which is SCN's arithmetic and not a disclosed figure. The 8-K describes the letter of intent's formula as the lower of the IPO price less 20 percent and the award-date closing price less 15 percent, and does not say which applies.

Component supply

Up to $400 million across four companies is small relative to what building out domestic cryogenics or laser capacity costs, and research awards are not procurement. The named domestic suppliers are not new entrants handed a critical path from nowhere either. Monarch Quantum, founded in 2025 by Timothy Day and launched publicly in January 2026, already supplies Quantinuum. In a March 31 release, the company said a $55 million growth round, set against more than $60 million in customer contracts from Quantinuum, Infleqtion and NASA, had taken it past $100 million in combined capital and contracts within six months of founding. The award adds to an arrangement already in place.

The incumbent picture has moved as well. Bluefors is Finnish, and its Helsinki site is both headquarters and main production base, but in September 2024 the company opened expanded facilities in Syracuse, New York, and said the expansion would let it build dilution refrigerators in the United States for the first time. Oxford Instruments' NanoScience division, the other long-standing cryogenics name, was sold to Quantum Design of San Diego in a deal that completed on January 2, 2026. The business now trades as Quantum Design Oxford, from the same Tubney Woods site in Abingdon, Oxfordshire, selling the same Proteox dilution refrigerators; the old NanoScience web address redirects to the new one. Ownership moved; the factory did not. On the evidence available, SCN sees nothing to indicate that a lab's practical dependence on those instruments moved either. The domestic control provisions in these agreements are written against ownership and production, not against a supplier's passport.

What becomes checkable from here is the schedule. Each tranche is tied to a milestone determination: a dated event with a pass-or-fail outcome, which a roadmap slide is not. Whether those determinations reach the public is a separate matter. The reviewed reporting provisions require submissions to Commerce in the form of financial reports, annual technical milestone status reports, patent reports, and research security certifications; they do not require public announcements of individual milestone determinations. What would ordinarily surface for the three SEC registrants is the money, in periodic financial statements. Commerce has posted finalization releases for D-Wave, Rigetti, Quantinuum, PsiQuantum and GlobalFoundries, whose award of up to $375 million supports "a secure, domestic quantum foundry that will serve multiple quantum architectures and modalities," along with cryogenic CMOS process design kits, advanced packaging and heterogeneous integration. Commerce announced nine quantum letters of intent in May. It has posted five as finalized, and SCN draws no conclusion about the rest.