Europe will not lose the AI race for lack of engineers or models. It will lose it because a better product is not allowed to show up uninvited.
Everyone counts the models. Fine, but a model that's ahead this quarter is on sale to the whole world the next, so it settles nothing. What settles it is who a buyer picks, and a buyer can only pick what has reached them.
Picture a three-person team with a tool that does a job better than what some company pays for today. Nobody there has heard of them. The only way in is to write to a stranger. Most messages like that are a nuisance and some are spam, and once in a while one is the better product. The team has no friends in that building yet, so the message is often all they've got.
Now make that message risky or illegal. The careful companies stop sending it, and business goes back to introductions, to people who already know each other, to the same conference every year. If you're inside the circle that's lovely. If you aren't, being better doesn't help, because nobody is comparing.
What bothers me is that you can't see it happen. There's no rejection. The product that never got a meeting just fades out, the incumbent keeps the account because it was already there, and the numbers say the market chose. I suspect a good chunk of "Europe doesn't produce scale-ups" is this, filed under funding.
I haven't measured any of it. It's how selling works as I understand it, and the number I'd need is how many first meetings start with a stranger here, against somewhere with looser rules. I don't want no rules at all either, spam is real. And I sell software, so of course I want the door open. If I were the incumbent with the warm relationships I'd probably call the same rule protecting customers from noise.
— Simon