Ex-OpenAI Researcher: DeepSeek, Kimi, MiniMax, Ling All Trained on Nvidia Chips

FateOfMuffins · reddit · 2026-11-13

Former OpenAI researcher Will Depue claims that Chinese models including DeepSeek, Kimi, MiniMax, and Ling were all trained on Nvidia chips.

The claim counters the narrative that Chinese labs rely on domestic compute, but it is an unverified personal assertion with no supporting evidence.

Related event: OpenAI researcher claims forensics show four Chinese labs trained models on banned Blackwell chips(13 posts)→

Original post →

More from Infra

• Cerebras Q3: cloud revenue up 272%, 700MW data center capacity under contract — Sethwinterroth · 2026-11-13
• Critic faults chip export-control research: ByteDance lines up 36,000 B200s in Malaysia — nitarshan · 2026-11-13
• Tracked GPU acquisition likened to SR-71's Soviet titanium, sourced via hundreds of willing neoclouds — bookwormengr · 2026-11-13
• DynaDFlash: diffusion drafter boosts speculative decoding acceptance length by 5% — charles_irl · 2026-11-13
• Report: Nvidia planning frontier-class open models, sources claim — StewartalsopIII · 2026-11-13
• Celesto CI runs GitHub Actions on microVMs for ~90% less per minute — aniketmaurya · 2026-11-13

Infra channel →