Skip to content

Quantum Observer

DARPA Advances Atom Computing, Diraq, IBM, and IonQ to the Final Stage of Its Quantum Benchmarking Initiative

The four join Microsoft and PsiQuantum in Stage C, where government evaluators test whether a proposed utility-scale machine can be built as designed. Of the four, only Diraq disclosed its initial funding, $51 million for the next year.

Illustration: Two machine assemblies sit inside a testing fixture, with four more queued beside empty bays.
Illustration: Quantum Observer (AI-generated)

The Defense Advanced Research Projects Agency (DARPA) said on October 7, 2026, that Atom Computing, Diraq, IBM, and IonQ had advanced to Stage C, the final phase of its Quantum Benchmarking Initiative (QBI). In Stage C, a government evaluation team tests whether a company’s proposed utility-scale quantum computer can be built as designed and operated as intended.

Microsoft and PsiQuantum entered Stage C earlier through a predecessor program, so six companies are now in the final phase. DARPA designed QBI to verify whether any approach to quantum computing can reach utility-scale operation by 2033. By utility-scale operation, DARPA means that a machine’s computational value exceeds its cost.

DARPA lists Stage C awards of up to $300 million per company. Of the four new entrants, only Diraq disclosed its initial funding, a $51 million tranche for the next year of work.

In Stage B, a yearlong review, DARPA examined each company’s research and development plan, including its technical risks and the prototypes needed to show progress.

The four new entrants

Atom Computing, of Boulder, Colorado, builds quantum computers from arrays of neutral atoms. The company said it could receive up to $300 million in Stage C but did not disclose an initial amount. Atom also said Microsoft had agreed to provide algorithmic support and error-correction codes.

In a preprint posted in June, Atom Computing and collaborators reported running an error-correcting code known as the toric code for up to 90 rounds of error checks, with lost atoms replaced from a reservoir during the runs. They compared two code sizes over up to eight rounds and reported that the larger code had a lower absolute logical error rate, the overall rate of errors in the encoded information.

Diraq, of Sydney, builds qubits from the spins of single electrons in silicon. The company disclosed an initial $51 million tranche within potential funding of up to $300 million.

In a paper published in Nature Communications in July, Diraq researchers and collaborators reported coherent control of an array of eight physical qubits made with a 300-millimeter foundry process compatible with CMOS, the standard technology for making computer chips. They also demonstrated a two-qubit gate between adjacent qubits. Diraq said it aims to reach utility scale by 2031 with more than 2 million physical qubits on a single chip.

IBM, a maker of modular superconducting processors, did not disclose an award amount or the length of its Stage C work. In its announcement, IBM said it expects to deliver the first fault-tolerant system in 2029. IBM has said that this machine, Starling, is designed to run 100 million quantum operations on 200 logical qubits, the error-protected units that a fault-tolerant computer builds from many physical qubits.

Trapped-ion company IonQ said it signed a Stage C agreement with a potential value of up to $300 million and that Stage C runs through 2029. IonQ did not disclose its initial amount. The company said “funding beyond the initial obligation remains subject to future appropriations and DARPA funding actions.”

DARPA’s evaluators will test several generations of IonQ’s Superion systems, the company said. When IonQ launched Superion in September, it said it had fabricated its first fully integrated processors with 256 physical qubits at its subsidiary SkyWater and had trapped the first ions in prototype systems. IonQ said it plans to deliver Superion systems to customers in 2027.

Microsoft and PsiQuantum

Microsoft and PsiQuantum reached the final phase through the Underexplored Systems for Utility-Scale Quantum Computing program (US2QC), a QBI predecessor. When DARPA announced their selection in February 2025, it said the final phase of US2QC had the same technical goals as QBI’s final stage.

Microsoft said on September 22 that it was providing topological qubits built on its Majorana 2 chip for independent testing at a new research center in Maryland. Topological qubits are a design intended to protect quantum information from local noise. Microsoft said DARPA will have full access to the system on site.

PsiQuantum, a developer of photonic quantum computers, signed a $125 million expanded agreement with DARPA in July, after a $31.8 million agreement in September 2025.

Next steps

“The quantum computing field is moving quickly, and we increasingly expect that someone will build a utility-scale quantum computer by 2033,” said Micah Stoutimore, QBI’s managing director, in DARPA’s announcement. “What we don’t yet know is which team or teams, or which technical approach or approaches, will get there.”

DARPA said it expects more companies to advance to Stage C as participants in its earlier stages progress. By November 2025, the agency had selected 11 companies for Stage B, including Quantinuum, QuEra, and Xanadu. DARPA said companies move through the stages on staggered timelines.

In a 2024 briefing for prospective participants, DARPA assumed for cost estimates that Stage C would last about 36 months. It wrote that companies completing Stage C would be those it deems ready to begin the final design and construction of a utility-scale quantum computer.