Originally published here.
It is utterly wild to me that (other than Mistral’s valiant effort) Europe as a whole is seemingly just giving up on the fight to build dominant players at the AI hardware and model layer. When over a year ago, @leopoldasch said he had situational awareness about the path towards AGI, I feel like I have increasing situational anxiety about what Europe’s lack of tech champions means for this continent. Don’t get me wrong: I fully understand how hard it is to compete on this front. But by now I am convinced that this is the most important technology shift in the history of humanity. We are only three years into GenAI and are already refusing to even enter the playing field because the chances of success are…low? That seems far more insane than not even trying.
Picture an imaginary competition with the morbid twist that anyone who fails to win a medal is dropped into a pool full of sharks. Whether you like it or not, you are forced to compete in three disciplines: Archery, ski jumping and fencing. The problem is: you have not done archery or fencing in years, and you have never ski jumped in your life. Panic sets in. The clock is ticking, the competition begins in four weeks. What do you do? Refusing to compete means you will certainly be fed to the sharks on game day. Whether you like it or not, the only option is to compete. So, you put on your adult pants and spend every remaining minute of every remaining day training as if your life depends on it, since, well, it actually does.
On game day, you take a deep breath, pick up the bow and arrow, and you shoot.
That is how Europe should approach AI, or rather: how we must approach AI. Indeed, the chances of success are slim and shrinking by the hour. But even so, if this were any other competition with stakes this high and odds this low, we would still try.
This makes me think the only logical explanation for our inaction is that we do not see losing the AI race as truly catastrophic - either because we doubt the race exists in the first place, or because we do not believe losing it would spell disaster for the continent. For if we did, we would never allow ourselves the option of sitting out and welcoming defeat.
The US approach to AI stands in stark contrast to this. Sriram Krishnan, Senior White House Policy Advisor on AI, recently summarised their geopolitical strategy for AI: complete and utter dominance.
“One of the measures we have been thinking about is in the future, let’s say the world inferences 10 quadrillion tokens a month. What share of those tokens are being influenced on American hardware, on American models? And how do we maximise that market share? That’s the mental model I have been playing with. How do we make sure this American stack is dominating that market of token inference?”
Meanwhile, Europe is fragmenting itself into ever more regional AI clusters, none of which have the talent density or scale to significantly move the needle for the continent. Imagine if the strongest talent behind notable European AI companies had been united in one company, rather than split up among many different ones? Instead of drowning in layers of regulation that may prove weak anyway, especially since the Trump administration has already signalled it will pressure the EU to lower barriers for US tech companies, Europe could channel that energy into building an attractive alternative: minimal regulation, significantly lower taxes, fast-tracked international work visas, and full talent mobility to draw in the best talent worldwide.
Make no mistake: these are our Hunger Games, and each of our numbers has just been drawn. No more talk of “we focus on the application layer”. Such feebleness is an insult to all the extraordinary European talent that has built the world we benefit from today. We owe it to ourselves to at least try, no matter how unlikely the chances of success.
