Go further · Chapter 10 · 10b · A step deeper
Has Microsoft built a topological qubit?
The short answer (tap to open)
Not as far as independent scientists can confirm. Microsoft says its Majorana chips use topological states that resist errors. Critics, including a paper published in Nature in June 2026, argue the key test had flaws and the signals could have ordinary explanations. Microsoft stands by its results. Independent replication, or a working qubit with measured performance, would settle it.
What has happened, and when?
Last checked: 2 October 2026.
- 2001: Alexei Kitaev shows how Majorana zero modes at the ends of a special wire could protect a qubit.
- 2012: a team at Delft reports first signs consistent with Majoranas in a nanowire. Excitement grows; Microsoft invests heavily.
- 2018 to 2021: its authors retract a high-profile Nature paper reporting a key Majorana signature, in 2021, after other researchers find problems with its data.
- 2023: Microsoft publishes its "topological gap protocol", a test to decide whether a device is in the right state.
- February 2025: Microsoft announces Majorana 1 and a Nature paper on measuring the qubit's state. Nature's editors add a note that the paper doesn't show evidence for Majorana zero modes.
- 2025: physicists including Henry Legg (St Andrews) publish criticisms of the gap protocol.
- June 2026: Nature publishes Legg's critique, alleging analysis-code errors and selective presentation; Microsoft publishes a reply standing by its results, and announces Majorana 2 with new performance claims (not yet peer reviewed).
- Throughout: DARPA keeps Microsoft in the final stage of its benchmarking programme.
What's Microsoft's case?
Its devices pass the protocol it designed for the topological phase; its measurements show the behaviour a topological qubit should have; it has published in leading journals; and it says newer chips show much longer-lived qubits.
What's the critics' case?
The protocol can be passed by ordinary, non-topological effects; the analysis contained errors; data that didn't fit was left out; the field has seen earlier retractions; and no independent group has reproduced the key results.
What would settle it?
What would settle it?
Would convince sceptics
Wouldn't settle it
A company press release
The answers, in words
- A company press release: Wouldn't settle it. A claim isn't evidence until others can check it.
- A second, independent lab reproducing the signals: Would convince sceptics. Would help a lot.
- A two-qubit gate with measured accuracy: Would convince sceptics. Would help.
- "Braiding": moving Majoranas around each other and seeing the predicted effect: Would convince sceptics. Would be decisive.
- A bigger chip announcement: Wouldn't settle it. More of an unconfirmed thing is still unconfirmed.
Check
One quick one
Collect
Your aha card
A sound idea, a disputed device.
Independent replication is what science waits for.