Hacker News Reader

More importantly, openly shared public-weight models limits the scale of the success that closed-weight private AI services can manage. Eventually it'll become cheaper and easier to pay some AI data centre to use their capacity plus an open-source harness rather than to pay Anthropic to pay an AI data centre etc. etc.

IOW China almost definitely sees their own open-weight models as a pressure release valve to ensure the US AI companies, and by extension the US economy, doesn't get too big, and to ensure that these companies don't become the dominant global force (the way that e.g. US cloud companies did originally).

100% this. I would assume if any government purchased an ai company and wasn't completely hands off (which from the sound of this post, it wouldn't be), I would assume that company would fail.

Could it be that the goal is not to be competitive, but to be responsible?

They're responsible for the Norwegian people's retirement savings, and burning $800 billion of it just to render an American AI company uncompetitive is not that.