Quantum error correction
Experiments in controlling quantum errors are getting stronger, and how they are done matters as much as the qubit count. An IBM-linked study, revised in September, reports a verifiable fidelity bound after discarding runs with detected errors. That is useful progress, and continuously correcting errors while running an economical, general-purpose computation has yet to be shown.
Research snapshot · updated Nov 12, 2026
State of the art (2026)
Google's Willow chip showed in December 2024 that error correction works in practice. A distance-7 surface-code memory ran below the error threshold, and its logical qubit lasted longer than the best physical qubit on the chip. Real-time decoding was shown on a smaller, distance-5 memory. Quantinuum launched its Helios machine in November 2025. IBM's Starling plan targets 200 logical qubits running 100 million gates by 2029, using qLDPC codes to cut the number of physical qubits needed. Fault-tolerant computing at a commercially useful size has yet to be shown, and Microsoft's Majorana qubit claims remain disputed.
Logical qubit error rate replaces physical qubit count as headline metric
Google Willow Dec 2024 below-threshold demonstration set the framing. 2025-2026 vendor headlines (IBM Heron, Quantinuum Helios, PsiQuantum Omega) increasingly cite logical error rate (10^-3 to 10^-6) rather than physical qubit count. This is the right metric for usefulness: 200 logical qubits with 10^-9 error rate enable quantum chemistry simulations of small molecules; 10,000 physical qubits with 10^-3 error rate do not.
IBM Starling 2029 + PsiQuantum 2027-2028 + Quantinuum Helios are the three credible 100+ logical qubit paths
IBM Starling 2029 target: 200 logical qubits, 100M gates (publicly committed roadmap). PsiQuantum Brisbane 2027-2028 commercial fault-tolerant photonic device (with A$940M federal backstop). Quantinuum H2/Helios extension to ~100 logical qubits 2027 (trapped-ion scaling depends on multi-chamber architecture). Microsoft + AWS + Atom Computing + QuEra are credible but with less explicit 100+ logical milestone commitment.
Magic state factory throughput becomes the new bottleneck
Surface codes don't natively support T-gates → magic state distillation factories required, ~10:1 overhead per T-gate. Quantinuum + IBM + Google demonstrated factory preparation 2025-2026. Algorithmic improvements (game-changing biased-noise codes, transversal magic state injection) over the next 24 months determine whether useful applications need 10⁵ or 10⁶ logical qubits.
Explore the deeper analysis in CanaryIQ
Explore what is covered below. These previews show the structure; log in to read the full analysis.
Signal stack
Evidence stacked leading → lagging
9 signals
Signal stack
Evidence stacked leading → lagging
Technology-native KPIs
Metrics that predict trajectory, tracked over time
4 tracked
Technology-native KPIs
Metrics that predict trajectory, tracked over time
Landscape map
Who builds what, and who depends on whom
75 players · 6 layers
Landscape map
Who builds what, and who depends on whom
Catalyst calendar
Dated events that will move the position
8 ahead
Catalyst calendar
Dated events that will move the position
Technology roadmap
Milestones on the path to maturity
8 milestones
Technology roadmap
Milestones on the path to maturity
Watchlists
Companies, people and papers, each with a condition for removing it
20 · 2
Watchlists
Companies, people and papers, each with a condition for removing it
Decision frameworks
What our view means for your job
Locked
Decision frameworks
What our view means for your job
Thesis changelog
When our view changed, and why
9 updates
Thesis changelog
When our view changed, and why
Change our mind
5 disconfirming conditions
You've read the verdict. The file is much deeper.
The full signal stack, technology-native KPIs tracked over time, the landscape of who depends on whom, the dated catalyst calendar, decision frameworks for every desk, live watchlists and a record of every time our view on Quantum error correction has changed. All of it is in CanaryIQ.