Microsoft’s quantum computing claims face new scrutiny from Nature critique

Microsoft has spent years betting that a particular flavor of quantum computing, one built on exotic particles called Majorana zero modes, will be the path that actually works at scale. A new critique published in Nature suggests the company’s evidence for that bet might be less solid than advertised.

Henry Legg, a researcher at the University of St. Andrews, published the critique on June 24, raising pointed questions about a Microsoft paper that appeared in the same journal back in February 2025. His core argument: the software Microsoft used to detect minute energy gaps in hybrid semiconductor-superconductor nanowires produced inconsistent results, and a fuller dataset released later looked more like random noise than proof of the energy gap Microsoft claimed to have found.

The ‘Jesus in toast’ problem

Legg’s analysis suggests the software essentially found patterns that weren’t really there. He likened the misleading outcomes to “finding an image of Jesus in toast,” a comparison that depends entirely on how you set your analysis parameters. Adjust the knobs one way and you see a breakthrough. Adjust them another way and you see noise.

This isn’t the first time Microsoft’s quantum research has faced this kind of challenge. The company has previously dealt with two retractions from Nature and alerts about papers published in both Nature and Science.

The March 2025 paper was characterized not as a claim of discovering Majorana particles outright but rather as a specific methodological innovation.

Microsoft pushes forward anyway

Microsoft isn’t backing down. The company has defended the disputed research, describing the software in question as an effective practical tuning tool that’s already been implemented in operational quantum chips.

On June 2, Microsoft announced the Majorana 2 chip. The company has also set a target of developing a useful quantum system by 2029.

DARPA’s Quantum Benchmarking Initiative is among the bodies evaluating quantum computing claims from various companies, adding another layer of accountability to a field where hype has historically outpaced results.

The competitive landscape is shifting

A paper published in June 2026 by Microsoft and Quantinuum reported an 11x to 800x reduction in logical error rates using error-correction codes.

The history of two prior Nature retractions is particularly relevant here. One retraction can be an honest mistake. Two start to look like a systemic issue with how results are being validated internally before publication. A third round of serious questions, even without a formal retraction, creates a credibility gap that engineering announcements alone can’t close.