TL;DR
For decades the honest answer about quantum computers was "someday," because the machines made too many mistakes to finish a long calculation. That is starting to change. Two companies using two different kinds of hardware now report that their error correction works the way theory says it must: the bigger the protective code, the fewer the errors. One says its protected qubits now fail about five times in a million operations and targets a fully fault-tolerant machine by 2029. At the same time, AI models have begun discovering quantum gates and redesigning quantum circuits to need far fewer operations. None of this has been independently checked, and no quantum computer yet does useful work better than a supercomputer. But the gating problem is moving, and if it keeps moving, the first real payoff, simulating molecules for drugs, batteries and catalysts, could arrive in the early 2030s.
Below the line, for paid subscribers: The full story: where it stands, how we got here and what's dated next; our Odds board on 5 milestones, starting with "Quantinuum's 99.9995% error-correction result passes peer review by 2027"; our labeled speculation on what it adds up to; and every source.