For most of the last decade, quantum computing was measured the way early PCs were: by a single number on a slide. More qubits meant more progress, and every vendor had a roadmap promising a bigger one next year.
The number that actually matters
That framing has quietly collapsed. What separates the leading labs now is the ratio between physical and logical qubits — how much messy hardware it takes to synthesise one qubit stable enough to run an algorithm on.
The shift changes who looks competitive. A 1,000-qubit machine with a poor error floor can be beaten, on real workloads, by a smaller system with a well-characterised noise model and a decoder that runs fast enough to keep up with the hardware.
The decoder is now part of the computer. If it cannot keep pace with the qubits, the qubits are decoration.
None of this makes the field less exciting. It makes it legible. Error rates are measurable, comparable and hard to spin — which is exactly why the next two years of announcements will be more honest than the last five.
Reader comments 0 comments
Sign in to comment.
No comments yet. Be the first.