PsiQuantum said on July 22, 2026, that it had signed a $125 million expanded agreement with the Defense Advanced Research Projects Agency (DARPA) under the agency’s Quantum Benchmarking Initiative (QBI). Under the agreement, DARPA will fund testing and evaluation of PsiQuantum’s utility-scale hardware designs and critical components, the company said. The agency defines utility-scale operation as computational value exceeding cost.
PsiQuantum said it is one of two companies in Stage C, QBI’s final stage. In March, DARPA said that two performers from a related program, Underexplored Systems for Utility-Scale Quantum Computing (US2QC), had advanced to Stage C.
Both Stage C companies, PsiQuantum and Microsoft, entered through US2QC without passing through QBI’s first two stages. DARPA has described Stage C as the phase in which its evaluators test companies’ hardware.
US2QC
DARPA selected Atom Computing, Microsoft, and PsiQuantum for US2QC in January 2023. The agency said it would use the program to determine whether an underexplored approach to quantum computing could reach utility-scale operation “much faster than conventional predictions.”
“The goal of US2QC is to reduce the danger of strategic surprise from underexplored quantum computing systems,” Joe Altepeter, the program manager, said at the time. DARPA said each company would present a design concept to a DARPA-led test and validation team in the first phase.
PsiQuantum said in January 2024 that DARPA had awarded it a contract for the second phase, focused on its design for a fault-tolerant prototype. A fault-tolerant computer corrects its own errors as it runs.
In February 2025, DARPA said it had selected Microsoft and PsiQuantum for the final phase of US2QC after more than 50 experts on its test and evaluation team examined their approaches. The agency said that phase “has the same technical goals as the final stage of QBI.”
QBI’s stages
DARPA announced QBI in July 2024 as an outgrowth of US2QC and a second program, Quantum Benchmarking. The agency later said it would use QBI to verify whether any approach can reach utility-scale operation by 2033.
In April 2025, DARPA said it had chosen nearly 20 companies for Stage A, a six-month sprint to characterize their concepts for a fault-tolerant quantum computer. Atom Computing, the third company selected for US2QC in 2023, was among them.
DARPA selected 11 companies, including Atom Computing, for Stage B in November 2025. The agency described Stage B as yearlong, with each company developing a research and development plan, identifying risks, and specifying the prototypes needed to reduce them.
In Stage C, DARPA’s independent verification and validation team will test the companies’ hardware, evaluating components, subsystems, and algorithms, the agency said in April 2025. In the same release, DARPA said companies that make it through Stages A and B will enter Stage C.
In March 2026, DARPA said it had evaluated approaches from 20 companies since QBI’s launch. It also opened a new Stage A solicitation for organizations that QBI had not yet funded. Micah Stoutimore, QBI’s managing director, said it “now seems likely that someone will build a utility-scale quantum computer by 2033.”
Published hardware results
PsiQuantum is developing photonic quantum computers that encode qubits in particles of light. In a Nature paper published in February 2025, PsiQuantum researchers reported benchmarks for chip modules, including 99.98% fidelity for state preparation and measurement, a measure of how accurately a photonic qubit is set up and read out. They reported 99.22% fidelity for two-qubit fusion, an entangling measurement on two photonic qubits, and 99.72% for a chip-to-chip interconnect, “conditional on photon detection and not accounting for loss.”
PsiQuantum said in July that DARPA’s earlier work with it included on-site reviews, real-time demonstrations, and hands-on component testing. The company also said it had announced a $31.8 million agreement in September 2025 to continue on-site testing and evaluation under Stage C.
Microsoft is developing topological qubits, a design meant to protect quantum information from local noise. In a Nature paper in February 2025, Microsoft researchers reported a measurement of parity, the property in which the company’s qubits would store information. Nature’s editors noted in the peer review file published with the paper that the results did not represent evidence of Majorana zero modes, the states the design requires.
In a June 2026 preprint, Microsoft researchers reported a parity switching time of about 20 seconds in a newer device. The switching time is the typical interval before the parity flips on its own. The researchers reported only Z measurements, one of the two measurement types needed to demonstrate a qubit, Science News wrote.
On June 24, Nature published a comment in which Henry Legg of the University of St Andrews disputed data behind the 2025 paper. The journal published a reply from Microsoft researchers alongside it. Chetan Nayak, who leads Microsoft’s quantum hardware work, told Scientific American that DARPA had “moved Microsoft into the final phase of its Quantum Benchmarking Initiative” after evaluating the company’s public and proprietary results.
Atom Computing builds machines from arrays of neutral atoms. In a June 2026 preprint, the company and collaborators reported running an error-correcting code called the toric code for up to 90 cycles, with lost atoms replaced during operation. In comparisons over up to eight rounds, they reported a lower absolute logical error rate for the larger of two code sizes.
Update, November 12, 2026: On October 7, DARPA said it had selected Atom Computing, Diraq, IBM, and IonQ for Stage C. The agency said the four join Microsoft and PsiQuantum, “which advanced to Stage C following participation in a pilot precursor” to QBI, and that it expects more companies to advance to the stage.
