Microsoft puts a Majorana 2 machine where DARPA can design its own tests

Maryland's $52.5 million from the FY2026 budget is the first money in, and it is more than half of the ceiling ^[11][12]. If DARPA and the state together reach the $100 million upper bound over four years, up to $47.5 million is still to come ^[18], which averages about $25 million a year across both funders ^[19]. The report does not say what Microsoft is spending on the center.

Test design moves with the hardware. DARPA's evaluations of Microsoft's quantum systems had been remote, run from Redmond or from facilities in Europe ^[4], and according to Crypto Briefing the hands-on machine lets DARPA researchers stress-test it under conditions they design instead of conditions Microsoft sets up for a demo ^[7].

Microsoft's bet is topological: qubits built on Majorana particles, theoretically more error-resistant than the superconducting qubits IBM and Google use ^[5]. Majorana 2 is the company's latest attempt to show that the theory survives contact with hardware ^[10]. Being one of two finalists in the US2QC program keeps the defense money attached to that approach ^[8]. The program targets practical, deployable quantum computers by 2033 ^[15], and Microsoft's own public commitment is commercial systems by 2029 ^[6], four years earlier ^[20].

The term worth a second look is the upgrade path. The system in Maryland is built so future Microsoft chips can be swapped in without rebuilding the facility ^[14]. A weak Majorana 2 number therefore does not end the tenancy, because the next chip goes into the same room.

The Quantum Benchmarking Initiative is framed as hardware-agnostic standard-setting ^[9]. If its numbers stay inside the government, the outside check sharpens DARPA's procurement and never reaches a commercial buyer. If the standard travels, it lands on the neighbours too, because IQM and Quantum Motion have set up in the same Discovery District, inside commuting distance of the Pentagon and Capitol Hill ^[17]. And the $52.5 million is state economic development money under a "Capital of Quantum" banner ^[13], which pays for talent and tenancy whether or not topological qubits work.

On the evidence, the substantive change here is narrow and real: DARPA writes the tests, Microsoft sells the hardware. A test DARPA controls is worth more than a demo and less than a result, since no benchmark has been published. The optimistic case falls apart if a good Majorana 2 number is followed by a missed 2029 date, because that would mean measurement was never the binding constraint. The facility's stated remit runs to cryptography, logistics optimization, sensor networks, and materials modeling for defense systems ^[16].