Nature published a comment on June 24, 2026, in which Henry Legg, a physicist at the University of St Andrews and the University of Basel, disputed the data behind a 2025 Microsoft paper on the devices meant to host the company’s topological qubits. Such qubits are designed to protect quantum information from noise. The journal published a reply from Microsoft Quantum researchers, with Chetan Nayak among the authors, alongside Legg’s comment.
“We stand by our results and our roadmap,” Nayak, Microsoft’s technical fellow and corporate vice president for quantum hardware, said in statements to Scientific American and The Register.
In the 2025 paper, Microsoft researchers reported a measurement of parity, whether the number of electrons in a nanowire is even or odd. When Nature published the paper, the journal’s editors noted that the results did not represent evidence of Majorana zero modes, the states Microsoft’s qubit design requires. Fortune and Physics World reported in February 2025 that the editors’ note was in the peer review file published with the paper.
The paper under dispute
Microsoft published the paper, “Interferometric single-shot parity measurement in InAs–Al hybrid devices,” on February 19, 2025. The same day, the company announced Majorana 1 and called it “the world’s first Quantum Processing Unit (QPU) powered by a Topological Core.”
Microsoft’s topological qubits would store information in Majorana zero modes, states predicted to form at the ends of semiconductor nanowires coated with a superconductor. Information shared across a pair of such states is predicted to resist local noise.
The qubits would encode that information in the nanowire’s parity. Microsoft measures parity through changes in the capacitance of a quantum dot coupled to the nanowire. A quantum dot is a tiny pocket that traps electrons.
Legg’s argument
Legg’s comment appeared as a Matters Arising, Nature’s format for peer-reviewed critiques of papers it has published. St Andrews said in a release that Legg found flaws in Microsoft’s topological gap protocol, an automated test the company uses to decide which regions of a device are in the topological phase.
Depending on where the measurement windows are set, the protocol can class the same region as gapped or as gapless, the university said. A gapped region has a clean superconducting gap, an energy range with no available electron states. Topological protection requires such a gap.
Legg also examined raw conductance data that Microsoft had not shown in the paper. According to the university, he concluded from those data that Microsoft’s devices were disordered, with possible quantum dots that could account for the measurements without topological states.
“There were two pretty basic Python programming errors that hid these alternative regions,” Legg said in comments published by The Register. He said plotting code had been set to highlight only the largest region the protocol identified as topological, hiding other regions that passed it. The second error, he said, was in code that reversed an array of bias-voltage data by index position instead of using the physical voltages.
Microsoft’s reply
In the reply, according to The Register, the Microsoft authors called one of the coding errors a minor off-by-one-pixel bug in their processing for the protocol. They described Legg’s critique as a selective examination of transport tune-up procedures. They also wrote that Legg had offered no alternative physical model that reproduces their capacitance signal.
Nayak told Scientific American that Microsoft had taken part in dialogue. He said Nature had accepted and published the company’s rebuttal. He also said DARPA, the Defense Advanced Research Projects Agency, had “moved Microsoft into the final phase of its Quantum Benchmarking Initiative” after evaluating the company’s public and proprietary results.
DARPA said in February 2025 that it had selected Microsoft and PsiQuantum for the final phase of a related program, Underexplored Systems for Utility-Scale Quantum Computing. The agency said that phase has the same technical goals as the final stage of its Quantum Benchmarking Initiative.
Responses
“They attempt to dismiss these issues as minor bugs, and retrospectively adjust their evidence hierarchy,” Legg said of the reply, in the comments published by The Register. Sergey Frolov, a physicist at the University of Pittsburgh, told Scientific American that Legg’s comment “makes it painfully apparent that the paper in Nature has no scientific value.”
Majorana 2 and the 2029 target
On June 2, Microsoft researchers reported a characteristic parity switching time of about 20 seconds in a device built with lead in place of aluminum, in a preprint that has not been peer reviewed. The switching time is the typical interval before the parity flips on its own.
In a blog post the same day, Nayak wrote that Majorana 2 “contains qubits that are 1,000x more reliable” than those in Microsoft’s previous processor. He wrote that Microsoft had cut its timeline in half and now aims to deliver a scalable quantum computer by 2029.
Demonstrating a qubit requires two types of measurement, called X and Z, Science News wrote. According to the magazine, Microsoft’s researchers reported only Z measurements in the June preprint. “For Majorana 2, one has to ask why they do not report the X-measurement,” Legg said in the comments published by The Register.
In the Majorana 2 work, Microsoft researchers used a new method to probe nonlocal properties of the states, meaning features shared across a pair of them, Science News reported. Kartiek Agarwal, a physicist at Argonne National Laboratory, told the magazine on June 2 that he regards the method as support for the case that the states are Majoranas. “This is fantastic progress,” Agarwal said.
Update, November 12, 2026: On September 22, Microsoft said it would provide Majorana 2 qubits for independent testing and that DARPA “will have full access.” On October 7, DARPA named Microsoft among the companies already in Stage C, the final stage of its Quantum Benchmarking Initiative.
