alt Hacker NewsAssuming you are right, the "build-up" of more physical datacenters isn't really the difficult part, training frontier models is. The open-weight Chinese models are several months behind frontier models at this stage and as long as you're not interested in burning cash to stay in the lead you can just wait it out. Many signs (upcoming ridiculous IPO figures and benchmarks) point to the current LLM paradigm reaching a plateau.
I think that currently the EU is focusing on strengthening itself from a military perspective due to US policy shifts, so the EU just doesn't have the funds to divert to AI without impacting day to day life of people, thus prioritizing safety and perhaps energy production, as electricity is considerably more expensive here than in the US or China. If you ask the people who vote, they'll put safety, grocery and gas prices way above EU's LLM capabilities.
> so the EU just doesn't have the funds to divert to AI without impacting day to day life of people
The US Government is not a major funder of frontier labs (who are almost exclusively VC and private money funded)...
> as electricity is considerably more expensive here than in the US or China
Yes, there are government policy reasons behind why this is the case. Some would argue that these policy decisions are a net negative for the EU.
> If you ask the people who vote, they'll put safety, grocery and gas prices way above EU's LLM capabilities.
Fair point, these are good things to focus on. However, EU citizens are going to love it when there are no frontier-tech opportunities left in their countries, with everything in this area having to be imported from the US and China (note that just because Chinese firms make open weight models available now, this could easily change very quickly in the future for a variety of reasons).