Will any China-domestic AI chip reach ≥80% of NVIDIA H100 perf on a public benchmark before 2026-12-31?
RégulationIAWeeklyChineTech2mois
Manifold MarketsSans KYCDonnées de résolution vérifiéesAssez bien calibréAlerte qualité des donnéesDonnées source dégradées
Données au 25 oct. 2026, 20:50 UTC · politique pm-quality-4
Prévision communautaire actuelleNon 81.9%
Prévisionnistes
6
Type de question
binary
Méthodologie
Play-money forecasting platform
Type de source
Prévision
Données du marché
Mis à jour il y a 16 heures
Obsolète4 mai 26, 20:22 UTC31 déc. 26, 23:59 UTC
Tendances
Marchés Associés
Questions liées pour le contexte, et non une comparaison de contrats équivalents.
[Metaculus] Will a Chinese firm make a large order of domestic AI chips before 2027?
Manifold Markets3. China’s AI chip sector make significant strides, planting seeds for eventual decline on Nvidia [See full title!]
Manifold MarketsWill an "AI chip export licencing" bill like the AI OVERWATCH act become law in 2026?
Manifold MarketsWill Apple release an AI chip for iPhones in 2026?
Manifold MarketsWill the US government force another frontier AI model shutdown for allied users by mid-2027?
Manifold MarketsWhat will happen in the world throughout the 2020s
Manifold MarketsRègles
Resolves YES if, on or before 2026-12-31, any China-domestic AI accelerator (Huawei Ascend, Cambricon, Biren, Moore Threads, Iluvatar, etc.) is reported to achieve ≥80% of NVIDIA H100 effective performance on a publicly disclosed benchmark.
- Acceptable benchmarks for resolution:
- MLPerf (mlperf.org) — official submitted results
- LLM training/inference benchmarks published by Epoch AI, CSET, or peer-reviewed academic papers (arXiv, conference proceedings)
- Vendor-published technical specifications + neutral third-party verification (e.g., research lab reproducing the result)
- "Effective performance" means application-level throughput on standardized AI workloads (e.g., GPT-class training tokens/sec, Llama-class inference tokens/sec, ResNet-class images/sec), not just raw FLOP/s peak.
