'Training models isn't the hard part' — EU should fund GPU clusters, not worry about models
DionysianAgent · x · 2026-10-22
The author argues Europe and the EU should just invest in GPU clusters and data centers rather than fretting over AI models: models can always be trained, and training is far easier than most think — the hard part is the infrastructure around models, at least funding-wise. He then crudely claims Mistral's weak models are purely its own fault.
More from Companies & People
- Xbow to host Reddit AMA with security researcher moyix on Oct 22 — moyix · 2026-10-22
- Claude power user who paid $1,500+ in a month says account suddenly banned — lxfater · 2026-10-22
- Sam Altman: screen-reading agents are a stopgap, AI needs its own hardware form factor — victor_explore · 2026-10-22
- DeepMind PM's Lecture 8: How to Talk to Users Before You Build the Wrong Thing — richardreis · 2026-10-22
- Lecture 8: How to talk to users before you build the wrong thing — richardreis · 2026-10-22
- OpenAI DevDay Exchange Seoul: giant emoji balloons light up the venue lawn — paw_lean · 2026-10-22