alt Hacker NewsIsn't building a 90% frontier model relatively cheap for EU?
I feel like EU could start a company, start from available open weight models, feed 2bln a year into it (1% of the EU budget) and make a compelling almost SOTA model for the EU market. This company could partner with datacenter providers and sell it hosted in the EU or somewhere else with EU protection terms. The budget for this company would easily double with the added revenues and you are creating an ecosystem of providers that can compete with US big-techs and have a 500 million people market that can't wait to ditch US companies for them, given the current mood.
The model can be open weight and it's an easy way to compound the efforts we are seeing in China without even having to talk to each other. Maybe there is a way to make it work not open weights but I am not sure how would that work.
These are those kind of decisions that seem such no brainers to me, which probably means I am completely out of touch with reality.
Replies
Europeans are risk averse and don't have access to that many deep pocketed risk-taking VCs. On top of that the US is poaching all the talent since most EU firms won't or can't match the salaries offered in the US to AI researchers.
This may change in the future as AI gets more commoditized and the current US admin keeps shooting itself in the foot but they are still very far ahead right now
> I feel like EU could start a company,
That's not how market-based economies work...
> feed 2bln a year into it and make a compelling almost SOTA model
...and the reason is, if you give a bunch of people €2b a year and tell them "go try and make something", they'll make a ton of paperwork covering their asses and very little actual output.
This is irrespective if those people are European ("european google killer"), American ("cost plus" old US aerospace companies) or Chinese (which is why they do it a little different).
If there are no incentives to really try really hard, they won't do it.
In many high-tech cases in Europe, the formula for "let's subsidise the hell out of research and hope a commercially-viable business comes out" has a really poor track record.
Your second option - and possibly the best bet - is to find an existing company that already showed they're capable, and shower them with money, which is what French are doing with Mistral.
Good idea in theory, but in reality 90% of this 2bn budget would just be swallowed up by the bureaucracy that would surround this.
1% of the budget is a really big number. The EU has a lot than a hundred responsibilities that demand money.