Nathan Lambert: distillation buys a few months, Chinese labs lack compute not talent
natolambert · x · 2026-10-15
Nathan Lambert, former post-training co-lead at the Allen Institute for AI and now launching nonprofit open lab Trillium, joined a16z's Lisha Li to discuss whether open models can keep up with the frontier.
- Telling Chinese researchers an average OpenAI researcher reportedly has 5,000 H100s, "their jaw just falls to the floor" — Chinese labs have talent but are short on compute
- He thinks the distillation story is overblown: research puts its edge at a few months, and "there's a bit of cope" in believing only American labs can build good models
- Echoes Jensen Huang's take: distillation isn't robbery, it's competition
A deep discussion on open vs frontier, the compute gap, and what distillation is actually worth.
Related event: Ex-AI2 Post-Training Lead Nathan Lambert Starts Nonprofit Open Lab Trillium(4 posts)→
More from AGI Musings
- AI cost falling 54x faster than electricity ever did, five years in — LinusEkenstam · 2026-10-15
- Herbert Simon's 40-year-old advice on computers nails how to use AI today — msg · 2026-10-15
- AI researcher Thomas Dietterich debunks multi-agent analogy: 1000 copies of me can't build a toaster — tdietterich · 2026-10-15
- Scholar's long post on Anthropic's societal-challenges push: earnest aims, ecosystem worries — sethlazar · 2026-10-15
- AI Auditing Pioneer Deb Raji Joins Stanford HAI to Tackle AI Harm Accountability — StanfordHAI · 2026-10-15
- Will the frontier-lab and closed-vs-open capability gap widen in six months? — panickssery · 2026-10-15