G.L.A. INVE$T RE$EARCH
The Infinite Innovation Newsletter
with Coach Kyle
October 8, 2026
DARPA JUST PICKED ITS QUANTUM FINALISTS. WHO MADE THE CUT, AND WHO DIDN’T.
Watch the breakdown first: https://www.youtube.com/watch?v=sg5BQ2Mw6dw(The Quantum Digest)
Disclaimer: This newsletter is not financial or investment advice. It is for educational and entertainment purposes. G.L.A. INVE$T RESEARCH CONSULTING is not a licensed advisor or broker.
Wall Street spent the last two years treating quantum like a slogan. DARPA just treated it like a contract.
On October 7, 2026, the Defense Advanced Research Projects Agency advanced four more companies into Stage C of the Quantum Benchmarking Initiative. That is the final stage. The test is simple and brutal: can any of these machines reach utility scale by 2033, meaning the computation is worth more than it costs.
Stage C is where plans stop and proof starts.
THE CUT
These four just moved up, each on a different qubit bet:
Atom Computing, Boulder, Colorado. Neutral atoms. Ceiling reported up to $300 million. Microsoft is supplying algorithmic support and error-correction codes.
Diraq, Sydney, Australia. Silicon CMOS spin qubits, built on the same manufacturing stack as ordinary chips. Initial funding tranche reported at $51 million.
IBM, Yorktown Heights, New York. Modular superconducting processors. The only mega-cap already inside the final verification room on this round.
IonQ, College Park, Maryland. Trapped ions. Company says the Stage C agreement is worth up to $300 million through 2029, with generations of its Superion systems put against the roadmap. Funding past the first obligation depends on appropriations.
They join the two that were already in Stage C from the earlier US2QC pilot:
Microsoft. Compact superconducting topological qubit architecture.
PsiQuantum. Silicon photonics.
Six names. Six different physics bets. That is the board.
WHO DID NOT MAKE THIS ROUND
DARPA’s own performer list, current as of October 8, still has these companies in earlier stages. Not eliminated. Not promoted.
Still in Stage B: Nord Quantique, Photonic Inc., Quantinuum, Quantum Motion, QuEra Computing, Silicon Quantum Computing, and Xanadu.
Still in Stage A: Google Quantum AI, on superconducting transmon qubits, and Quandela, on hybrid spin-optical.
Read that again. Google is still in Stage A. Quantinuum, the Honeywell-backed trapped-ion name a lot of institutions treat as the quality leader, is still in Stage B. Xanadu and QuEra are still in Stage B.
DARPA was explicit. QBI is not a winner-take-all contest. Managing director Micah Stoutimore said Stage B puts the plan under a microscope, and Stage C is where the government checks whether the hardware does what the roadmap claims. He also said more promotions into Stage C are expected, and that the field is moving fast enough that someone building a utility-scale machine by 2033 is now the base case. Which team, and which physics, is the open question.
A miss on this list is not a death sentence. It is a timing signal. In this market, timing is the trade.
WHY THIS MATTERS
Utility scale is the line between a lab demo and an industrial tool. DARPA’s definition is the one that counts for defense, intelligence, and eventually the public market: value exceeds cost, on a path that can be built and operated as designed, by 2033.
Stage A is the concept. Up to about $1 million. Stage B is the R&D plan, the risks, and the prototypes. Up to about $15 million. Stage C is independent verification and validation. Up to $300 million.
Since mid-2024, DARPA has looked at more than 20 companies. This is the first time the public can see a real short list with government money attached to proof, not slides.
THE INVESTOR BOARD
Public names you can actually buy:
IonQ (IONQ). Pure-play trapped ion. Now sitting in the same verification room as IBM and Microsoft, with a reported agreement ceiling up to $300 million through 2029. That is a credibility event. It is also a concentration event. One contract does not make a utility-scale computer.
IBM (IBM). The incumbent. Modular superconducting. Already sells quantum access. Stage C is validation that the roadmap survived a year under a microscope, not a reason to re-rate the whole company overnight.
Microsoft (MSFT). Already in from the pilot, topological qubits, and now also the error-correction partner on Atom Computing’s neutral-atom path. Two seats. One balance sheet.
Private names that belong on the pre-IPO radar, the same way we track Anduril, PsiQuantum-class infrastructure, and the rest of the defense-tech stack:
PsiQuantum. Silicon photonics. Already Stage C. The photonics path is the one that looks most like a foundry story if it works.
Atom Computing. Neutral atoms, Boulder. Microsoft on the software side. Reported ceiling up to $300 million.
Diraq. Silicon spin qubits on existing CMOS. If that manufacturing claim holds, it is the cheapest path to scale on the board. Australian, not American. Worth knowing.
The names still outside the room are not scrap. Quantinuum, QuEra, Xanadu, and Google Quantum AI are the “not yet” pile. In a program DARPA says will keep promoting teams, “not yet” is a position, not a verdict.
COACH KYLE’S READ
I have sat through every quantum cycle since the first time a sales desk tried to bolt the word onto a semiconductor pitch. Most of those cycles were marketing. This one is different because the customer is the Pentagon’s research arm, the test is utility versus cost, and the deadline is 2033.
Three things I want the network to understand.
Modality is the moat argument, not the logo. Neutral atoms, trapped ions, superconducting, silicon spin, silicon photonics, topological. DARPA kept all of them alive on purpose. Anyone telling you the architecture is settled is selling a position.
The public pure-play is IonQ. The public platforms are IBM and Microsoft. The asymmetric upside, if you can get it, is still private: PsiQuantum, Atom, Diraq. That is the same split we use on the rest of the innovation book. Our approach is to favor the listed operators and track the privates before the bankers do.
Do not confuse a Stage C badge with a product. Stoutimore’s line is the whole note: plans to proof. Proof is what Stage C is for. Until a machine’s value clears its cost, this is a research and national-security program, not a cash-flow story.
Quantum is not a APEX 10 or SEXY 6 replacement. Tesla, Palantir, Nvidia, Spacex ,Strategy, Robinhood, and Bitcoin are still the compounding core. Quantum is the next layer of the compute stack, the same way we treated AI infrastructure before the multiple showed up in the prints. The investors who win this one will be early on the architecture and late on the hype.
WATCH IT
The Quantum Digest walks the cut, the misses, and why the obvious name is not the obvious winner: https://www.youtube.com/watch?v=sg5BQ2Mw6dw
If you want the private-market follow-up on PsiQuantum, Atom Computing, and Diraq let Me know!
Stay in the know!!
Coach Kyle
Founder, G.L.A. INVE$T RESEARCH CONSULTING
The Infinite Innovation Newsletter
DISCLAIMER: THIS NEWSLETTER IS NOT INVESTING OR FINANCIAL ADVICE. ALL INFORMATION IS FOR EDUCATIONAL AND ENTERTAINMENT PURPOSES. G.L.A. INVE$T IS NOT A FINANCIAL ADVISOR OR LICENSED BROKER.
Original edition: Kyle’s Beehiiv newsletter. Author opinions, scenarios and dated estimates are retained; limited editorial clarifications do not verify every source claim.