Can Open Models Keep Up With the AI Frontier? | Nathan Lambert

Compute access, not talent or distillation, is the main structural edge keeping OpenAI and Anthropic ahead of fast-following open-model labs.

• Compute is the real moat: Nathan Lambert says Chinese labs have talent "comparable" to U.S. labs but lack compute — a top Chinese pretraining researcher reacted with disbelief on hearing OpenAI researchers can access ~5,000 H100-equivalents each.
• Distillation is overstated: Lambert argues the public literature shows distillation only buys a few months' catch-up, since open post-training recipes still lean on an "outdated Qwen" teacher model with unclear gains from stealing frontier reasoning traces.

Themes, sentiment and reasons are AI-written paraphrases, checked so they never quote an episode. They may contain errors. Not investment advice. Episodes remain the property of their creators.