DARPA Reopened Its Quantum Entry Stage. The Stage A Ceiling Is $1 Million — and Two Deadlines Are in Play

by Margaret Ashworth · September 14, 2026

Senior Analyst, Frontier Tech Wire

The entry door into what DARPA itself calls the world's largest evaluation of quantum computing closes in just over two weeks — at least on one of the two dates the agency has published. Full proposals for the Stage A Quantum Benchmarking Initiative Topic — solicitation number DARPA-PA-26-02-02, shortened by the agency to Stage A QBIT — are due Sept. 30, 2026, according to the Defense Advanced Research Projects Agency's own announcement of the reopening. For small and mid-cap quantum companies, and for the private firms that compete with them for the same customers, the date is one of the few fixed points on the autumn calendar.

It is also one of the most consistently misread. The Quantum Benchmarking Initiative is routinely described in market commentary as a funding programme, and selection into it is routinely announced by participants as a form of government endorsement. DARPA's published description is narrower than that. The agency's programme page states that QBI "is designed to rigorously verify and validate whether any quantum computing approach can achieve utility-scale operation — meaning its computational value exceeds its cost — by 2033." The same page names Micah Stoutimore as the programme's manager.

DARPA announced the reopening in a news item dated March 10, 2026, headlined "Quantum Benchmarking Initiative expands quest to separate hype from reality." That item sets two dates in a single sentence: "Abstracts are due July 31, 2026, and full proposals for the Stage A QBIT are due Sept. 30, 2026." The July abstract date has passed. The September proposal date has not. Neither date is published in that item with a cut-off time or a time zone, and the Stage A QBIT frequently-asked-questions document instead refers applicants to "the date and time stated in the QBIT PA" — the programme announcement itself. Anyone filing should take the clock from that document, not from a news page.

The Quantum Insider, writing up the reopening on March 11, 2026, reports the same two dates — July 31 for abstracts, Sept. 30 for full proposals. We compared the two write-ups directly; on the deadlines they do not diverge.

In the same announcement, Stoutimore is quoted saying: "In fact, it now seems likely that someone will build a utility-scale quantum computer by 2033, but it remains unclear exactly which team or teams might get across that finish line." The emphasis on "someone" is DARPA's own. The conditioning matters and is easy to strip out in a headline: the sentence asserts that someone is likely to succeed, while explicitly declining to say who.

The dates sit inside a longer-running announcement, and they do not agree. DARPA's opportunity listing for the parent programme announcement, DARPA-PA-26-02, is posted with a date of Nov. 14, 2025 and carries a deadline of Dec. 30, 2026 — three months later than the Sept. 30 proposal date attached to the QBIT topic beneath it. The Stage A QBIT frequently-asked-questions document describes "a rolling admission process such that any abstracts submitted prior to the date and time stated in the QBIT PA will be considered," but that line names neither date and is about abstracts rather than full proposals. This publication did not find a DARPA statement reconciling the two dates in plain language, and we are not resolving them here. Prospective applicants should read the solicitation itself rather than rely on either figure alone.

The money attached to each stage comes from DARPA's own briefing decks, not from secondary reporting. The programme-manager briefing DARPA published for the QBI proposers day, dated Sept. 3, 2024 on its face, lays out three stages: "QBI Stage A: Plausibility," at "6-months" and "Up to $1M"; "QBI Stage B: R&D Plan," at "12-months" and "Up to $15M"; and "QBI Stage C: Validation and Co-Design," with "Duration tailored to performer" and "Up to $300M." A DARPA Microsystems Technology Office briefing dated July 25, 2025 repeats the same three ceilings and the same three durations. In both decks each dollar figure carries an asterisk directing readers to the QBI programme solicitation for details and conditions — the ceilings are ceilings, not award sizes.

The Stage A QBIT frequently-asked-questions document is consistent with that ladder. It states that "six months is the expected Stage A term" while allowing that milestones may be completed early, that "the total payments in Stage A do not exceed $1M," and that "multiple awards are anticipated." One caution on the paper trail: DARPA hosts this file among its March 2026 attachments, but the document carries an original posting date of March 24, 2025 on its face. We are describing it by its solicitation number rather than by a year.

The more consequential line for anyone modelling what a QBI selection is worth appears in DARPA's published QBI questions-and-answers document — the file DARPA hosts among its April 2025 attachments, stamped as originally posted Sept. 6, 2024. Asked at Question 72 whether DARPA would fund a particular research or risk-mitigation effort, the agency answered: "QBI is not significantly funding research, it focuses verification and validation of ongoing research and development efforts. Please see Section II.D for Stage B and Stage C funding limits and constraints." On DARPA's own account, then, this is a programme that pays to check work, not a programme that pays to do it.

A widely repeated claim about QBI cost-sharing deserves a careful reading, because of where in that document it actually sits. The one-third and one-half fractions people cite appear in Question 75 — that is, inside the text of a question a prospective proposer submitted to DARPA, not inside anything DARPA wrote. The submitted question reads: "We anticipate getting funding from other sources; would accepting government grants over private funding impact our ability to participate vis a vis the requirement that DARPA not fund more than ⅓ in Stage B or ½ in Stage C of our effort?" That is the proposer's own characterisation of the solicitation. DARPA's answer, quoted here in full, does not restate either fraction: "DARPA anticipates all performers will raise funds on their own. The source of your other funds should not affect your participation, apart from following relevant laws and regulations." To be unambiguous: we have found no DARPA statement of a one-third or one-half cost-share rule, only a proposer asserting one inside a question. Readers who need the binding text should go to Section II.D of the solicitation.

What is not in doubt is the direction of DARPA's answer: performers are expected to raise their own money. That reframes the headline ceilings. Up to $15 million over twelve months is not a runway for a company burning tens of millions a quarter; it is a contribution toward an evaluation exercise the company is largely financing itself.

Xanadu Quantum Technologies, which trades on Nasdaq and the TSX under XNDU, offers the cleanest published illustration. In results issued Aug. 5, 2026, the company reported second-quarter 2026 revenue of $1.5 million — down from $2.8 million in the first quarter of 2026 and up from $1.1 million a year earlier, with the release attributing that year-on-year increase primarily to DARPA Stage B revenue. Note the distinction: DARPA money is credited with the increase, not with the whole of the line. Against that, Xanadu reported a net loss of $42.1 million for the quarter and cash and cash equivalents of $312.8 million at June 30, 2026, in U.S. dollars under U.S. GAAP. The DARPA line is real revenue. It is also under four per cent of the quarter's net loss.

The attrition between stages is visible in DARPA's own pages. The agency's April 3, 2025 announcement said "nearly 20" companies had been chosen for the initial stage and named seventeen: Alice & Bob, Atlantic Quantum, Atom Computing, Diraq, Google Quantum AI, Hewlett Packard Enterprise, IBM, IonQ, Nord Quantique, Oxford Ionics, Photonic Inc., Quantinuum, Quantum Motion, QuEra Computing, Rigetti Computing, Silicon Quantum Computing Pty. Ltd. and Xanadu, with at least one further selection still in negotiation at publication. DARPA's Stage B selection page, dated Nov. 6, 2025, says the agency "selected 11 companies to enter the second stage (Stage B)" and names them: Atom Computing, Diraq, IBM, IonQ, Nord Quantique, Photonic Inc., Quantinuum, Quantum Motion, QuEra Computing, Silicon Quantum Computing Pty. Ltd. and Xanadu. Several names on the first list, including Rigetti Computing, do not appear on the second. DARPA does not publish reasons for individual outcomes, and we are not attributing any.

Rigetti's own disclosures show why a $1 million or $15 million line item can be a small part of the picture even for a company of that size. In results dated Aug. 6, 2026, Rigetti Computing (Nasdaq: RGTI) reported second-quarter 2026 revenue of $5.1 million and a GAAP net loss of $52.6 million for the three months ended June 30, 2026, alongside cash, cash equivalents and available-for-sale investments of $541.3 million at that date. The GAAP figure is not all operating: the release's reconciliation shows a $29.6 million change in the fair value of derivative warrant liabilities inside it, and a non-GAAP net loss of $16.0 million once that and other adjustments are stripped out. The same release notes that in May 2026 Rigetti signed a letter of intent with the U.S. Department of Commerce for an award of up to $100 million over three years, under the CHIPS Research and Development Office broad agency announcement pursuant to the CHIPS Act, with the department taking an equity stake corresponding to the total funding amount. That is several times the QBI Stage B ceiling, from a different agency, under a different statute, and on very different terms.

At the other end of the range, IonQ (NYSE: IONQ) reported second-quarter 2026 revenue of $80.1 million in an Aug. 5, 2026 release, described as 287 per cent year-on-year growth, with cash, cash equivalents and investments of $3.0 billion at June 30, 2026. The same release states, on a GAAP basis: "Net loss was ($1,867.7) million and GAAP EPS was ($5.08) for the second quarter." We re-opened IonQ's release to confirm the figure and its units, and it is stated as written — millions, not thousands, for the quarter. The release does not break out what drives a loss of that size; it separately reports an adjusted EBITDA loss of $120.3 million for the quarter, a non-GAAP measure on the company's own definition. We are not reconstructing the gap between the two. For a balance sheet of $3.0 billion, a Stage B award of up to $15 million does not move the cash position; its value to a company like this is as a credential rather than as funding.

Eligibility for the reopened stage is broader than some readers assume. The Stage A QBIT FAQ states that performers from prior QBI or US2QC efforts "are not precluded from submitting new proposals under this PA if they believe their approach offers new or distinct contributions," and adds that those who "did not move beyond Stage A should contact DARPA via QBI@darpa.mil prior to submission." On foreign participation, it states that "Non-U.S. organizations and/or individuals may participate to the extent that such participants comply with any necessary nondisclosure agreements, security regulations, export control laws, and other governing statutes applicable under the circumstances" — a materially different posture from grant programmes restricted to domestic entities.

None of this is a view on any security, and nothing here is a forecast. Most of the companies inside QBI are pre-profit, several are private, and the publicly traded ones carry the ordinary risks of speculative technology issuers: large operating losses, dilution, dependence on capital markets that can close, and a commercial timeline DARPA itself frames around 2033. Advancing a stage is not a contract to build anything; DARPA's programme page describes Stage C as the point at which performers "Work with the Government to Verify and Validate that their utility-scale quantum computer concept can be constructed as designed and operated as intended." The immediate, checkable events are the Sept. 30, 2026 proposal date on the QBIT topic, whatever DARPA subsequently announces about Stage A QBIT selections, and the unreconciled Dec. 30, 2026 date on the parent announcement.