Vol. III·Government Tracker·DARPA QBI

Stage B advancers · Stage C decisions Q4 2026

Boulder · Colorado
10 min read

DARPA Quantum Benchmarking Initiative

The only program in which a sovereign technical buyer with no equity exposure spends a year auditing each performer's R&D plan against a single, unambiguous test.

Can this approach deliver an industrially useful quantum computer by 2033?

— the single question DARPA's Quantum Benchmarking Initiative is built around

On November 6, 2025, DARPA announced that 11 of 18 Stage A performers advanced to Stage B — a 12-month engineering audit with funding up to $15M per team. In parallel, PsiQuantum and Microsoft continue in the related US2QC final phase.

Stage B selection has become an unofficial Series-D-equivalent due-diligence stamp: a single credible signal that DARPA's technical team judged a company's roadmap plausible. The list of who is not on this roster — Google, Rigetti, HPE were Stage A participants and did not advance — is just as load-bearing as the list of who is. Stage C advancement decisions, expected Q4 2026, are the single most important catalyst for the sector this year.

The cohort · 11 of 18 advanced

Stage B performers

IBM

Superconducting (modular)

High

BenchmarkModularity at scale + realism of the 2029 fault-tolerance roadmap

RecentGambetta framed Stage B advancement as "firm validation"; most detailed public roadmap in the industry

ConcernsLeast dependent on DARPA capital — Stage B is reputational, not financial. 2029 FTQC date leaves thin margin for modular interconnect losses.

Quantinuum

Trapped Ion (QCCD)

High

BenchmarkValidate Lumos utility-scale design + Apollo 2029 FTQC milestone

RecentHelios deployed Nov 2025 with record physical and logical fidelities; high-Tc materials simulations

ConcernsQCCD ion-shuttling speed is a known scaling bottleneck; Apollo 2029 aggressive given current zone counts

QuEra Computing

Neutral Atom (Rydberg)

High

BenchmarkMulti-thousand-atom continuous operation + algorithmic fault tolerance

RecentAlgorithmic FT breakthrough; below-threshold 4-round circuits (2.14× below threshold) with Harvard

ConcernsAtom loading rates and reconfiguration speed at multi-thousand scale; commercial revenue still nascent vs IonQ/Quantinuum

FundingUp to $15M

Atom Computing

Neutral Atom (Yb)

Above-average

BenchmarkScaling from ~1,000 phys / ~50 logical to >10,000 phys / >100 logical

RecentJan 2025 on-premise Microsoft system delivered 24 logical qubits; NVIDIA NVQLink integration

ConcernsReliance on Microsoft's logical-qubit stack creates correlated exposure to Majorana strategy; mid-circuit measurement / atom-loss replenishment unproven at scale

IonQ

Trapped Ion (chip-based + photonic)

Moderate

BenchmarkPath to 2M physical / 80K logical qubits by 2030

Recent99.99% 2Q fidelity (world record); AQ 64 milestone on Tempo three months early; $130M FY25 revenue

ConcernsThe 2M-qubit / 2030 number is the most aggressive in the cohort. Narrative-to-engineering gap is the widest.

Xanadu

Photonic (GKP)

Moderate

BenchmarkGKP fidelity, loss budget, Aurora-to-FTQC scaling plan

RecentAurora end-to-end (12Q across 35 chips, 13km fiber); 60% optical-loss reduction in 2025; HyperLight TFLN partnership

ConcernsLoss-per-component is the dominant blocker. 2029 FTQC commitment is aggressive even for photonics.

FundingUp to $15M

Photonic Inc.

Silicon T-centre spin-photon

Moderate

BenchmarkRemote entanglement fidelity between T-centre modules for distributed surface codes

RecentThree-qubit register on silicon photonic chip (Nature Nano Jan 2026); $200M+ round at $2-2.7B valuation

ConcernsT-centre emission yields and remote-entanglement rates are the dominant scaling questions

Diraq

Silicon CMOS spin

Plausible

BenchmarkFoundry-fabricated devices clearing >99% 2Q fidelity consistently across many devices

RecentCryogenic electronics integrated with qubits (Jun 2025); foundry-fab spin qubits >99% fidelity (Sep 2025 Nature)

ConcernsLongest-duration roadmap in the cohort — qubit count today is tens, proof point is industrial repeatability

Quantum Motion

Silicon MOS spin

Plausible

BenchmarkTile repeatability and yield on foundry process

RecentSep 2025 delivered industry-first full-stack silicon-CMOS quantum computer to UK NQCC; $160M Series C closing May 2026

ConcernsSame long-cycle risk as Diraq; current qubit count is small

Silicon Quantum Computing

Precision atom donor

Lower-middle

BenchmarkDemonstrating precision-placement scaling without degrading per-qubit fidelity

RecentDec 2025 study showed silicon-based processor scales without fidelity loss; record-setting processor announced early 2026

ConcernsMost boutique manufacturing — STM atom placement not yet a foundry process; near-term qubit counts very small

Nord Quantique

Multi-mode bosonic

Binary

BenchmarkBosonic redundancy genuinely substituting for surface-code overhead at scale

RecentContinued bosonic error-correction results through 2025; expanded Sherbrooke facility

ConcernsThe 1:1 ratio claim is the most aggressive theoretical claim in the cohort. IV&V will scrutinize hidden overhead.

Funding$5M initial + up to $15M total

The final phase · Stage C-equivalent

US2QC track

PsiQuantum

Silicon Photonics (Omega chipset)

High

BenchmarkBuild a real million-qubit-scale fault-tolerant utility computer

Recent$1B Series E (Sep 2025); Omega manufacturable chipset; broke ground at IQMP

ConcernsCapital intensity vs timeline; component-loss budgets at million-qubit scale remain the central physics question

Microsoft

Topological (Majorana 1)

Highly uncertain

BenchmarkDemonstrating actual topological qubit operation at system scale

RecentFeb 2025 Majorana 1 announcement (8 topological qubits) — peer reviewers and Nature editorial team publicly disputed claim

Concerns2018 underlying Nature paper retracted. Microsoft's parallel Atom Computing partnership functions as effective hedge. IV&V is the most rigorous test the topological claim will face.

The differential signal

Stage A only

The most decision-relevant section of the tracker. Six Stage A names did not appear on the November 2025 Stage B roster. Two were absorbed into Stage B advancers in the months around the announcement — the cleanest M&A signal QBI has produced. DARPA has stated the program is not a winnowing competition; companies may be added in future rounds, but the Stage B cut still functions as the first real differential signal.

Eliminated · stand-alone

Rigetti Computing

RGTI

Superconducting

Public pure-play, Stage A only. Cleanest negative differential signal in the program. No 8-K tying stock-price guidance to QBI non-advancement has been identified.

Alice & Bob

Cat qubit (bias-noise)

Stage A only. Did not advance despite Nord Quantique (also bosonic) making the Stage B cut.

Hewlett Packard Enterprise

HPE

Superconducting / hybrid HPC

Stage A only. Immaterial to HPE's enterprise mix.

Acquired into a Stage B participant

Atlantic Quantum

Fluxonium superconducting

Acquired by Google Quantum AI (October 2025). Team folded into Google's QBI participation.

Oxford Ionics

Trapped ion

Acquired by IonQ (~September 2025, ~$1.075B). Team folded into IonQ's QBI line.

Joined Stage A late · Stage B status unclear

Google Quantum AI

GOOGL

Superconducting

Added to Stage A in September 2025 — too late for the November Stage B cut. Acquired Atlantic Quantum a month later. May be considered in a future Stage B round.

What changed this year

2026 program updates

Leadership transition

Micah Stoutimore replaced founding program manager Joe Altepeter in early 2026, driven by DARPA's mandatory PM tenure limits. No change in stated technical criteria.

New Stage A solicitation

DARPA-PA-26-02-02, opened March 9, 2026. Open only to organizations that have not previously received QBI funding. Abstract deadline July 31, 2026. Signals widening aperture rather than narrowing.

Directional shift

"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."

— DARPA, March 2026

The rail · past · next · 2033

Stage B → Stage C → 2033

Feb 2023
DARPA announces US2QC collaborations with Atom Computing, Microsoft, PsiQuantum
Jul 2024
QBI program launched as successor / expansion to original Quantum Benchmarking program
Feb 2025
Microsoft + PsiQuantum advance to US2QC Validation & Co-Design (Stage C equivalent)
Apr 2025
Stage A announced (~18 performers, up to ~$1M each)
Sep 2025
Google Quantum AI added to Stage A; IonQ acquires Oxford Ionics (~$1.075B)
Oct 2025
Google acquires Atlantic Quantum team
Nov 6 2025
Stage B — 11 advance, up to $15M / team
Early 2026
Stoutimore replaces Altepeter as QBI program manager
Mar 2026
New QBIT Stage A solicitation (DARPA-PA-26-02-02) opens for organizations not previously funded
Q3 2026
Stage B mid-term reviews
Q4 2026
Stage C advancement decisions
2027–2029
Stage C — Build, Verify, Validate (independent IV&V)
2033
Utility-scale verification deadline

Primary sources

For full sector context, see /companies, /qnt-ipo-watch, and /roadmaps.