Chips for Humanity launches nonprofit compute cluster for AI safety researchers with ~2 GPUs each vs ~200 at Anthropic
Miles_Brundage · x · 2026-11-12
AI safety researcher heyanuja, with Nicky and Alexa, launched Chips for Humanity, a nonprofit that procures compute for safety researchers outside frontier labs, with an open application form and a LessWrong post.
Key arguments from the site:
- Stark gap: outside safety researchers average 2 GPUs per person (up to 8 at the best-funded teams), versus roughly 200 per safety researcher at Anthropic based on public figures; frontier labs put only 2% of research compute into safety.
- Hostile compute market: Geodesic Research spent six months securing one cluster — contracts locked at 1–3 years, prices rose 10% a month from May to August, offers vanished in 1–2 days.
- Consequences: slower feedback loops, no compute-heavy methods, and false negatives — problems that only emerge in larger models go undetected.
- Proposed fix: funders should establish or back an entity that owns compute contracts and allocates GPU time across grantees.
Former OpenAI safety researcher Miles Brundage amplified the initiative.
More from Infra
- Six techniques that decide LLM serving performance, explained — techNmak · 2026-11-12
- MiniMax H3 Needs 50 GB — Here's How It Runs on a 12 GB GPU — fa6637 · 2026-11-12
- PISA: Prompt-Adaptive Cost-Aware LLM Routing Tops Four Routing Benchmarks — Saul Urso · 2026-11-12
- Join ordering, Kaplan-Meier decode estimates and trie attention yield 10x analytics inference speedup — charles_irl · 2026-11-12
- Bernstein's Rasgon: Nvidia's growth comes down to Jensen Huang being a visionary — ceciletamura · 2026-11-12
- Lossless MXFP4 scale compression frees 50.2 GiB across 8 GPUs running Kimi-K3.1-Pro — AlpinDale · 2026-11-12