United States – October 24, 2026 -- DARPA has advanced 11 quantum computing companies to Stage B of its Quantum Benchmarking Initiative (QBI), giving each team one year to show that a useful, fault-tolerant quantum computer can be built by 2033.
Launched in 2024, the QBI is designed to test whether any quantum approach can reach utility-scale performance on an economically sensible timeline. It is not a winner-takes-all contest. Several teams can succeed in parallel, or none at all.
Stage B follows a six-month feasibility screen of nearly 20 companies
Stage A selected nearly 20 companies to propose fault-tolerant architectures that could realistically be built within a decade and deliver net value. Over six months, DARPA worked with each team to test the feasibility of its concept. The 11 Stage B companies now come with funding and tighter scrutiny. The source does not disclose the funding amounts.
Trapped-ion and superconducting platforms: IonQ, Quantinuum, IBM and Nord Quantique advance
IonQ advanced on proven real-world performance and a credible scaling roadmap for its trapped-ion qubits. Quantinuum, which builds modular trapped-ion systems on the QCCD architecture, advanced for a mature roadmap and a de-risked path to fault tolerance.
On the superconducting side, IBM advanced with a validated fault-tolerant architecture and a detailed scaling plan built on modular, chip-linking design. Nord Quantique advanced for a hardware-efficient approach that uses bosonic error-correcting codes in superconducting cavities.
Photonic and silicon spin teams make up five of the 11 selections
Xanadu advanced with a fault-tolerant photonic roadmap that has defined execution milestones. Photonic Inc. advanced for a distributed architecture that links silicon spin qubits through optical entanglement.
Three silicon-spin developers also advanced:
- Diraq, for what DARPA judged the most scalable and cost-effective silicon-based design.
- Quantum Motion, for a plausible route to a fully CMOS-compatible quantum chip with cryo-CMOS control electronics.
- Silicon Quantum Computing, for atom-precision fabrication and high algorithmic fidelity in its donor qubits.
Neutral-atom developers Atom Computing and QuEra join the cohort
Atom Computing advanced for high-coherence arrays of nuclear spin qubits in optical lattices and a credible scaling strategy. QuEra Computing advanced for an energy-efficient platform based on optical tweezers and Rydberg interactions, backed by a strong experimental foundation.
The spread across trapped ions, superconducting circuits, photonics, silicon spin qubits and neutral atoms shows that no single approach has won out yet.
Independent verification will test each blueprint before Stage C hardware builds
To pass Stage B, teams must deliver a detailed, credible blueprint for a 2033-era machine. They must also show, through intermediate prototypes or simulations, that no critical obstacles remain unaddressed. DARPA's support includes independent verification and validation by expert teams, who will check whether performance targets and risk mitigations are realistic.
Teams that meet the Stage B objectives will be invited into Stage C. There they must build and test components and subsystems to show that the envisioned machine can be constructed and will perform as expected.