I knew Europe was behind. There was Europe 2031. Then there was A Transformative AI Strategy for Europe. Then I saw some of my most brilliant friends leaving Zürich for the US upon graduating.

This was much like watching the speedometres of three cars – US, China and Europe – and noting that Europe was going much slower. It’s different going for a ride, sticking your head out of the window and feeling the my-eyes-are-burning kind of speed.

Over the last few months, I’ve felt some of that burn. First, I visited San Francisco. After that, I spent some time trying to understand attitudes towards AI in China, speaking with Chinese PhD students in ML in Sweden and binge-listening to various podcast episodes about AI in China. You feel the speed in in-person conversations, twitter memes, signs in public spaces, feature articles. Here’s some of that speed. Fasten your seatbelts.

The death of the middle class #

Speed means worrying about a permanent underclass. If AI replaces all intellectual jobs, then wealth will be concentrated to whoever has AI company shares. If no one needs your talent and hard work, then what leverage do you have? Social mobility is only possible in a meritocratic society.

The idea of a permanent underclass isn’t really a concern in China. Chinese people are already used to cutthroat competition – not with AI, but with their peers. The government has also introduced policies to deal with AI labour displacement; for example, it is illegal to fire workers to replace them with AI.

Sharing is caring #

Speed means blindly trusting tech companies. In reality, you can already automate most digital work – you just have to let the AIs loose on your computer. Install whatever plugins you can find – at a bare minimum, get Google Drive, Gmail, Slack and GitHub – and grant the AI full read and write permissions.

Privacy concerns are so 2012 – the big tech companies already know you. If you’re really worried, just get a Mac Mini, so you can tinker with the latest stuff.

Sharing personal data isn’t just a matter of contextmaxxing – it’s also a way to support your country in the AI arms race. By sharing your data, you’ll help train the next generation of models.

Less leisure than ever #

Speed means working nonstop. Cognitive fatigue used to be a reliable stop signal; these days, you can vibe-work for 12h straight without feeling drained. Oh, and before going to sleep, be sure to kick off another batch of experiments.

In fact, depending on your business, you could also make agents do automatable work around the clock. There’s nothing stopping you and your fleet of agents from producing more… stuff.

Europe #

You don’t feel the speed in Europe. People just aren’t as excited to play around with the latest models.

Perhaps this is more worrying than Europe having no frontier AI, as it leads to bad collective epistemics. The best way to understand the transformative effects of AI is arguably by experimenting with Claude Fable 5.1 or GPT-6 Astra. I still remember my first experience with Claude Code – it was shocking¹. It helped me internalise how fast things move.

The new world order is the fast world order. If you think AI development is fast, remember that the labs would like to move twice as fast.