Microsoft is letting DARPA switch on its disputed topological qubits itself

Microsoft opened a 15,000-square-foot quantum research center in the University of Maryland's Discovery District on 22 September, combining its own research space with partner laboratories, a hands-on training lab and a dedicated DARPA testing area. Until now DARPA has reached Microsoft's quantum hardware remotely, in Redmond and in Europe. At the new site agency personnel can physically install hardware and execute their own startup procedures on Microsoft's latest topological system, including qubits built on the Majorana 2 chip, for independent testing under DARPA's utility-scale quantum program — part of the Quantum Benchmarking Initiative, which aims to establish whether competing approaches can yield commercially useful machines by 2033. The system is built so future Microsoft chips can be swapped in; Microsoft says it plans commercial systems by 2029.

Why it matters

Microsoft's topological approach carries the largest gap in the field between what is claimed and what outsiders accept. The pitch — qubits built on Majorana zero modes, protected from noise by the physics itself — would change quantum computing's economics if true. The evidence that such modes exist in Microsoft's devices has been disputed by physicists since the Majorana 1 announcement in 2025, and a claim whose central measurement is contested cannot be settled by more measurements from the same lab.

That is what makes the arrangement worth recording, and it is also the limit of it. Remote access lets an evaluator run jobs on a machine someone else prepared. Physical access to startup — the calibration and tuning that happen before a device is handed over — is where a lab's judgment about what counts as working gets exercised, and it is exactly the stage outsiders normally never see. Letting DARPA do it is the quantum equivalent of the employee-level access AI labs promised evaluators this month: the right kind of openness, with nothing yet to show for it.

So nothing here is a result. There is no tracker entry, because access is not an outcome; the recordable event will be DARPA reporting what it found, in whichever direction. What the arrangement changes is that when a finding comes, it will be one Microsoft did not produce.

What to watch

Whether DARPA publishes Majorana 2 findings or only a QBI stage decision, whether Microsoft advances past the next QBI stage, and whether any other topological-qubit claim gets comparable third-party access.