Physicists are mostly unconvinced by Microsoft’s new topological quantum chip

ANAHEIM, CALIF. — At the world’s largest gathering of physicists, a talk about Microsoft’s claimed new type of quantum computing chip was perhaps the main attraction. 

Microsoft’s February announcement of a chip containing the first topological quantum bits, or qubits, has ignited heated blowback in the physics community. The discovery was announced by press release, without publicly shared data backing it up. A concurrent paper in Nature fell short of demonstrating a topological qubit. Microsoft researcher Chetan Nayak, a coauthor on that paper, promised to provide solid evidence during his March 18 talk at the American Physical Society’s Global Physics Summit.

Before the talk, the chair of the session made an announcement: Follow the code of conduct; treat others with respect. The room, jam-packed with hundreds of eager physicists filling the seats and standing along the walls, chuckled knowingly at the implication that decorum might be lost.

Topological quantum computing has had a dark shadow cast upon it by a series of retracted claims. Nevertheless, the concept holds great promise. The qubits that make up quantum computers are notoriously fragile and error-prone. Qubits that harness the concepts of topology, the mathematical discipline that describes structures with holes or loops, might improve on this. With topological quantum computing, “you can have very low error rates,” Nayak, of Microsoft’s Station Q in Santa Barbara, Calif., said during his talk.

Scientists were not wowed by the data he presented.

A key plot looked like random jitter, rather than an identifiable signal. Nayak claimed that an analysis of that apparent randomness revealed a pattern underlying the noise, suggesting a working qubit. That argument wasn’t enough to flip the harshest critics.

“The data was incredibly unconvincing. It is as if Microsoft Quantum was attempting a simultaneous Rorschach test on hundreds of people,” says physicist Henry Legg of the University of St. Andrews in Scotland, one of the fiercest critics of Microsoft’s work. 

Still, others were optimistic that, with additional effort, Microsoft could improve their device to produce a clearer signal. “I felt like it was maybe a bit premature to call it a qubit,” says physicist Kartiek Agarwal of Argonne National Laboratory in Lemont, Ill. But “there’s very many positive signs.”

The draw — and pushback — of topological qubits 

Quantum computers promise to unlock new types of calculations, but only if they can be made reliable. The idea of building a qubit that is intrinsically less error-prone has excited scientists. “It’s one of the more creative, more original approaches to quantum computing, and in this sense, I’ve really been rooting for it,” says physicist Ivar Martin of Argonne National Laboratory.

But the idea has struggled to get off the ground, trailing decades behind more conventional qubit technologies. 

Creating a topological qubit requires provoking electrons in a material to dance just-so. The electron collective behaves like a hypothetical, particle-ish thing: a quasiparticle known as a Majorana. But creating Majoranas, and proving they exist, has been extremely challenging. 

Microsoft has made impressive strides, Martin notes. But “as far as demonstrating things which people at this meeting would care about the most — really convincingly showing physics of Majoranas — it’s underwhelming to many.”

If it’s possible to be less-than-underwhelmed, that would describe Legg, who gave a talk the day before Nayak’s. He expressed doubts about the very foundation of Microsoft’s method in a room filled to bursting — albeit a significantly smaller room than Nayak’s headliner venue. 

In his talk, squeezed into the meeting’s schedule at the last minute, Legg listed a litany of criticisms. The critique centered on the method used to demonstrate that the device is topological in the first place — the “topological gap protocol,” laid out in a 2023 Microsoft paper in Physical Review B. That protocol was flawed, he argued in his talk and in a paper submitted March 11 to arXiv.org. For example, Legg argued, the protocol gives different results for the same data, depending on the range of the parameters included, such as the spread of magnetic field or voltage values. 

“Any company claiming to have a topological qubit in 2025 is essentially selling a fairytale, and I think it’s a dangerous fairytale,” Legg said. “It undermines the field of quantum computation and, in general, I think it undermines, actually, the public’s confidence in science.”

During a Q&A immediately after Legg’s talk, Microsoft researcher Roman Lutchyn rose with a forceful rebuttal: “A lot of statements here are just simply incorrect,” he said, ticking through several of Legg’s claims, which he also addressed in a LinkedIn post. “We stand behind the results in these papers.”

Disorderly conduct

At their most basic level, Microsoft’s devices consist of aluminum nanowires, just 60 nanometers wide, laid atop a semiconductor. When cooled, this aluminum becomes superconducting, allowing it to transmit electricity without resistance. This induces superconductivity in the semiconductor, creating ideal conditions for Majoranas. Once the device is tuned to particular values of magnetic field and voltage, Majoranas should theoretically appear at each end of the nanowires.

Disorder in these devices is a big problem for topological qubits. Surface roughness or material defects can result in spurious signals or ambiguous results. In recent years, Microsoft’s devices have improved enormously in that regard, says physicist Sankar Das Sarma of the University of Maryland in College Park. But, he says, “some more improvement is needed.… I think disorder still needs to go down by another factor of two.” 

When the aluminum threads are arranged in an H shape, they create a qubit with Majoranas at each of its four ends. To claim a working qubit, Microsoft needed to show that they could perform measurements on it. This involves probing quantum dots, hot dog–shaped nanoparticles laid out near the nanowires. Two types of measurements, known as X and Z, are necessary.

In the February Nature paper, Microsoft demonstrated a Z measurement, which involves probing the quantum dot associated with a single wire. Repeated Z measurements revealed the qubit switching between two possible states, the expected outcome for a topological qubit. These transitions purportedly indicated flips in parity, essentially reflecting whether there were an even or odd number of electrons within a wire. 

During Nayak’s talk, he unveiled their X measurement, which probes a quantum dot adjacent to two nanowires. The plot of these data looked random, lacking the same obvious flip-flopping between two values. 

The audience did not seem particularly impressed. During the Q&A, Cornell University physicist Eun-Ah Kim said, “I would have loved this to just come out screaming at me that there’s only two, but I don’t think that’s what I see.” 

Nayak said that a statistical analysis of the random-looking data revealed a hidden pattern. But, in an email, Kim questioned the validity of Nayak’s method for teasing out this pattern.

Even regarding the clearer Z measurement, scientists still don’t agree whether this flipping constitutes evidence for Majoranas. “I’m persuaded,” Das Sarma says, “but people of goodwill could disagree.”

During the talk, attendees raised smartphones high to snap photos of Nayak’s slides, which rocketed around the physics community. Just after the presentation, physicist Sergey Frolov of the University of Pittsburgh, who was not at the meeting, posted a detailed rebuttal on the social media platform BlueSky. 

“[T]he data shown are … just noise. They are simply disappointing,” wrote Frolov. This, he suggested, doesn’t bode well for the chip containing eight qubits that Microsoft announced in February: “That chip cannot possibly work, given what we saw today.”

Not all scientists are quite as critical as Legg and Frolov. Agarwal, for example, thinks Microsoft’s topological gap protocol, the foundation of their current work, is sound. But, he notes, the device Nayak presented is impractical, given that its values appear essentially random. “It certainly can’t be used as a qubit in its present state. That’s also clearly obvious,” Agarwal says.

Nayak is confident that his team will improve their devices further, until skeptics are convinced. Frolov, for one, is confident that more paper retractions are coming.