Researchers clash over whether EU frontier parity really needs 7GW of compute
eliebakouch · x · 2026-09-15
Elié Bakouch challenges a European AI compute report: catching up with OpenAI/Anthropic's internal models shouldn't need 7GW — 1GW is roughly Colossus 2's scale, and 2GW of the report's proposed 14GW (1M B300s) would be more than enough. He adds that datacenter building and inference can be decoupled from training. Report author Anton Leicht responds that the figures target actual frontier parity, assuming US intervention cuts off hyperscaler compute rentals past some point. A follow-up notes the report cites 45GW.
More from Infra
- effect-harness on Cloudflare: run agent loops, tools and credentials on separate VMs — samgoodwin89 · 2026-10-31
- Google's Project Suncatcher boots TPU in orbit and beams back first AI response — sundarpichai · 2026-10-31
- Together AI launches Post-Training platform for agentic RL with tool-use environments — togethercompute · 2026-10-31
- General Compute raises $100M for inference capacity as CoreWeave stock hits 12-month low — YvesMulkers · 2026-10-31
- Analyst warns of unstable AI compute feedback loop as Kioxia's NAND price warning triggers flash crash — bookwormengr · 2026-10-31
- SOMA launches enterprise prompt-compression layer claiming 20%+ inference savings — markjeffrey · 2026-10-31