Are Useful and Error-Free Quantum Computers Really Only Two Years Away?

[Phoenix56]

New Scientist is running a piece asking whether fault-tolerant, useful quantum computers could arrive within two years, a significantly more optimistic timeline than most expert consensus from even 12 months ago. The basis for the question is real progress on multiple fronts. Microsoft's Majorana 2 announcement claimed qubit lifetimes of up to 20 seconds versus microseconds for current systems. A paper published in April 2026 by researchers working with neutral atom architectures found that fault-tolerant quantum computing might be achievable with as few as 10,000 to 20,000 physical qubits, drastically lower than previous estimates of millions. Microsoft and Atom Computing have committed to delivering error-corrected systems to research partners in 2026. QuEra has delivered an error-correction-ready machine to Japan's National Institute of Advanced Industrial Science and Technology.

The counter-argument is that the gap between error-corrected and genuinely useful at scale remains enormous. Yuval Boger from QuEra has said publicly that anyone claiming quantum computers are commercially useful today is being misleading. The machines arriving in 2026 are what the Microsoft framework calls Level 2, small error-corrected systems that can demonstrate scientific advantage on specific problems, not the Level 3 large-scale machines needed for most commercially transformative applications. Two years to useful is plausible for narrow scientific problems. Two years to broadly commercially useful is not.

[Dark Hawk]

The distinction between scientifically useful and commercially useful is the one that resolves most of the apparent disagreement between optimists and pessimists in this space. Both can be right simultaneously about different things