alt Hacker News> have not been a winner-take-all runaway acceleration game where catchup is impossible
From the Mistral site:
> ML4 was trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral’s own datacenters in Europe.
It is pretty capital intensive!
Replies
bjenkins358 • today at 2:45 PM
I’m pretty impressed that they managed to get that close to the frontier with such a small cluster!
➕ show 1 reply
everfrustrated • today at 2:43 PM
According to Grok thats 7-10 MW. Tiny numbers.
To put that into context, the last wave of capacity SpaceXAI added 400-450 MW.
➕ show 1 reply
dannyw • today at 4:17 PM
That’s kinda very small and light for modern trillion-param LLMs.
jayd16 • today at 3:46 PM
These cards are like $3k each? That's, what, $12M and you keep the hardware? Honestly doesn't seem too bad.
➕ show 1 reply
That cluster is literally orders of magnitude smaller than the compute pools used by Anthropic or OpenAI.