A commercially viable quantum computer demonstrates fault-tolerant computational advantage on a practical task before 2030.
While noisy quantum systems exist, users query whether practical quantum computers are truly functional yet. Resolution hinges on achieving fault tolerance on real-world industrial tasks.
Across 6 frontier models answering independently, the panel has all but ruled it out: the median is 11%.
They broadly agree, 23 points between GPT at 29% and Claude at 6%.
Nothing is settled until 16 Nov 2031, when the outcome is checked against sources and every number above gets a permanent score.
What each one said
-
GPTOpenAI
It turns on whether logical-qubit error correction scales fast enough to beat specialized classical hardware on a useful workload.
- IBM, Google, Quantinuum and PsiQuantum must deliver many high-fidelity logical qubits and a reproducible application result by 2030.
- No platform has yet shown fault-tolerant advantage on an industrial task; prior advantage claims were narrow sampling benchmarks.
- A credible 2027–28 demonstration of sustained error suppression with dozens of logical qubits would materially raise the odds.
29% -
GeminiGoogle
Scaling to thousands of logical qubits required for practical advantage faces brutal hardware and error-correction bottlenecks.
- Leading roadmaps from IBM and Google target 100 to 1,000 logical qubits by 2029-2030, but real industrial utility demands far more.
- Decoherence, control electronics at scale, and high physical-to-logical overhead make the pre-2030 timeline exceptionally tight.
- A breakthrough in neutral-atom or topological architectures achieving 1,000+ error-corrected qubits by 2028 would sharply raise odds.
19% -
DeepSeekDeepSeek
Fault-tolerant logical qubits at scale are the bottleneck; nobody has shown even one below-threshold logical qubit at useful scale.
- Google's 2024 Willow result showed error decreasing with more physical qubits, but only at ~100-qubit scale, far from the ~10^4-10^6 logical qubits needed for practical…
- IBM's roadmap targets 200 logical qubits by 2029 and 2000 by 2033; even optimistic vendors don't claim fault-tolerant advantage on industrial problems before 2030.
- The gap between a single below-threshold logical qubit and a commercially useful fault-tolerant machine is roughly 3-5 orders of magnitude in physical qubit count and…
12% -
QwenAlibaba
Requires fault-tolerant logical qubits at scale beating classical on a real industrial problem—no one is close to that threshold yet.
- IBM and Google roadmaps target ~1,000 logical qubits by ~2029-2030, but demonstrating advantage on a practical task requires running a full algorithm end-to-end, not…
- The gap between Quantinuum's 2024 logical-qubit error-rate demos and a commercially viable fault-tolerant system solving, say, a chemistry or optimization problem is at…
- A surprise would be a breakthrough in error-correction overhead (e.g., LDPC codes cutting qubit requirements 10x) combined with a narrow practical problem needing fewer…
9% -
GrokxAI
FTQC advantage on a useful industrial task needs ~10^3–10^4 logical qubits that no 2026–29 hardware path yet supplies.
- IBM, Google, Quantinuum and PsiQuantum must ship and run a logical-qubit algorithm that beats classical solvers on chemistry/optimization by 31 Dec 2029.
- Expert surveys and published roadmaps still place first useful FTQC in the mid-2030s; 2019–25 ‘advantage’ claims were NISQ and non-practical.
- A sudden 10× fidelity or qubit-count leap published in 2027 would be the only development that could move this above ~20 %.
8% -
ClaudeAnthropic
Hinges on achieving true fault tolerance with practical (not academic) task advantage within ~4 years, a very high bar.
- Current logical qubit demos (Google, IBM, Quantinuum, IonQ) remain small-scale error-correction proofs, not industrial-task advantage.
- Roadmaps from major players target thousands of logical qubits by early-to-mid 2030s, not 2029-2030, for truly practical fault-tolerant workloads.
- A breakthrough like a surprise error-correction scaling leap or a company claiming verified commercial fault-tolerant advantage would change this, but no credible signal…
6%
6 of 6 models answered · 23 points between the highest and lowest. None was shown the market price.
Question sourced from a news sweep on 17 Nov 2026. Forecast by google/gemini-3.8-flash, anthropic/claude-sonnet-5, openai/gpt-5.6-terra, x-ai/grok-4.6, deepseek/deepseek-v4.1-flash, qwen/qwen3.8-max-0902 via OpenRouter.