Microsoft says it will have a useful quantum computer in three years - the Majorana 2 claim explained

[BigDog]

Microsoft has dramatically accelerated its quantum computing roadmap after unveiling Majorana 2 at Build 2026. The new chip achieves mean qubit lifetimes of 20 seconds compared to milliseconds on the original Majorana 1 - a roughly 1,000x improvement. Microsoft now targets a scalable, commercially useful quantum computer by 2029, cutting its previous 2033 target in half. The key materials change was replacing aluminium with lead as the superconductor, a switch that Microsoft's own Discovery agentic AI platform helped engineer. Chetan Nayak, who leads Microsoft's quantum effort, said: 'We will have a quantum machine in 2029 that can solve commercially viable, reasonable problems.'

Physicists remain divided. The topological qubit approach is still controversial and Microsoft has not yet published enough public data to fully satisfy independent scrutiny of the Majorana results.

Microsoft Says It Will Have A Useful Quantum Computer In Three Yearswww.forbes.com

[Amy96]

1,000x improvement in qubit lifetime is an extraordinary number. Milliseconds to 20 seconds is not incremental progress, that is a qualitative change in what becomes possible. If independently verified this is significant

[Luca76]

The scepticism from physicists is legitimate and should not be brushed aside. Microsoft's topological qubit claims have had credibility problems going back years and claiming a 1,000-fold improvement deserves rigorous independent review before the roadmap acceleration is taken at face value

[Tracey]

Replacing aluminium with lead is a materials science insight that took years to make work according to their own account. The fact that agentic AI accelerated the materials discovery is itself an interesting data point about what Discovery is actually capable of

[Northernah]

2029 lines Microsoft up exactly with IBM which committed $10 billion last month to quantum machines on the same timeline. Two major players targeting the same year with very different architectures is going to make for an interesting four years

[StevenArroyo]

A four-qubit array is still a prototype. The gap between a prototype that demonstrates a materials breakthrough and a commercially useful machine with enough logical qubits to solve real problems is enormous and Microsoft has consistently underestimated it in previous roadmaps

[TeaSpiller]

If the 2029 goal holds, the encryption implications become a near-term concern rather than a theoretical future one. Harvest now decrypt later attacks on encrypted communications become more strategically viable the closer a real fault-tolerant machine gets