
The AI Race Nobody Can Afford to Win
Earlier this week Dario Amodei of Anthropic, published an essay, We Must Pace the Frontier, arguing that the development of frontier AI may need to slow down to give safety research, testing, and regulation more time to catch up.
His pitch isn’t “stop AI” but more “let’s catch our breath before the tech sprints into the unknown.” But there’s a twist. Amodei believes the U.S. can only tap the brakes if it stays ahead of China.
His concern is mainly military. He is afraid that China could use AI to transform warfare, particularly through autonomous systems and swarms of intelligent drones. To preserve the advantage, he argues that the United States should restrict China’s access to advanced AI chips (see Nvidia) and to semiconductor manufacturing equipment, while also targeting chip smuggling and access to computing power through data centres outside China. The UAE, for example, is currently building enormous AI data centres, backed by major global tech players. This, of course, raises the question of whether China could benefit from these data centres through cloud access, partnerships, or neutral hosting.
In any case, Amodei thinks that these measures could stretch US lead for 3–5 years enough time for the US to negotiate from a position of strength.
Amodei isn’t naïve, of course. He knows cooperation will be difficult, but he proposes a few areas where the two countries might find common ground:
- A mutual ban on using AI for bioweapons.
- Mandatory testing of frontier models for cyber, biological, and autonomy risks before release, and
- Rules limiting how AI can accelerate the creation of even more powerful AI. A kind of digital arms‑control treaty.
The problem is obvious. What happens if one country slows down while the other quietly keeps accelerating? AI training is not a missile silo. One cannot spot it.
And then there is another issue. The AI race will not necessarily remain a simple contest between Washington and Beijing. In today’s globalised world, tech innovation no longer follows a hierarchical path, meaning that middle powers can exploit their competitive advantages and innovate far beyond what their size might suggest. As countries like Turkiye, Israel and Ukraine have shown, new ideas and adaptation are not constrained by geography, wealth, or population size.
If that sounds like a geopolitical thriller, that is because it is. The question is: Can humanity slow down just enough to stay safe, without letting everyone else race ahead?
Beyond Amodei’s post, I took a lot from ideas in the book, Bytes and Bullets: Global Rivarlies, Private Tech and the Future of Modern Warfare by Steve Feldstein, which I highly recommend to anyone interested in geopolitics and the future that’s coming, whether we asked for it or not.
Thank you for reading. If you liked this piece or found it useful, share it, or consider subscribing. Or buy me a virtual coffee as caffeine keeps this whole enterprise running.
✍️ Written by
Newsletter
Stay updated with new articles.

No comments yet. Be the first to comment!