The Defense Advanced Research Projects Agency (DARPA) has selected four quantum computing companies for the final stage of its Quantum Benchmarking Initiative (QBI), an independent evaluation of whether their technologies can deliver commercially useful quantum computing.

The four companies, IBM, IonQ, Atom Computing and Diraq, will undergo extensive testing by DARPA researchers, who will evaluate their hardware, system designs and ability to build fault-tolerant quantum computers.

The announcement marks a major milestone for QBI, which was launched in July 2024 to determine whether a commercially viable quantum computer can be built by 2033.

DARPA defines a commercially useful quantum computer as one that produces more computational value than the total cost of building and operating it. Quantum systems must demonstrate not only advanced computing capabilities but also a practical business case.

The four companies join Microsoft and PsiQuantum, which previously qualified for the equivalent final evaluation through DARPA’s Underexplored Systems for Utility-Scale Quantum Computing program.

Four Competing Quantum Technologies

The four newly selected companies use distinctly different approaches to developing quantum processors, a sign of the continuing debate over which architecture offers the best path to commercially successful quantum computing.

IBM is developing superconducting quantum processors, with plans to launch its Quantum Starling system in 2029. The company is focusing on error correction, modular processor design and cryogenic infrastructure to support a fault-tolerant system.

IonQ uses trapped-ion technology, which manipulates individual charged atoms to perform quantum calculations. Its development roadmap includes the Superion 256 architecture, and the company has reported achieving 99.99% accuracy in two-qubit gate operations.

Diraq is pursuing silicon-based spin qubits, using manufacturing techniques adapted from the semiconductor industry. Its roadmap targets 150,000 physical qubits by 2029 and more than two million qubits on a single chip by 2031.

Atom Computing uses neutral atoms controlled by optical systems. Its technology includes arrays containing more than 1,200 neutral-atom qubits, with Microsoft supporting the development of error correction technology.

DARPA Shifts to Hardware Testing

The companies have already completed two evaluation stages. Stage A examined their technical concepts, while Stage B assessed their development plans, engineering challenges and strategies for addressing potential failures.

Now, stage C focuses on performance of actual hardware and engineering systems. DARPA’s independent verification team will examine whether the companies’ systems deliver the results promised in their technical roadmaps. As QBI Managing Director Micah Stoutimore put it, “Stage C is where we begin moving from plans to proof.”

This includes evaluating quantum processors, control systems and error correction technology. Error correction is particularly important because quantum computers are highly sensitive to disturbances that can produce inaccurate calculations.

DARPA’s evaluation is not intended to crown a single industry winner. Rather, the agency wants to identify every credible technology capable of meeting its 2033 target.

The latest announcement does not eliminate other participants. Seven companies from the previous evaluation group, including Quantinuum, QuEra Computing and Xanadu, have not yet advanced. DARPA expects additional companies to qualify for Stage C.

For the quantum computing industry, the shift to independent hardware testing is likely to significantly impact the sector’s future. Companies have promoted ambitious development roadmaps, and some quantum vendors are generating revenue from hardware sales, cloud-based access and research contracts. In this still emerging tech, the DARPA test could also help enterprises determine when quantum computing might become a practical alternative for specialized workloads.