How the US-China AI Race Will Play Out – Leopold Aschenbrenner

Leopold Aschenbrenner argues a US lead of even a few years over China in AI could be militarily decisive — and a neck-and-neck race risks catastrophic instability.

• A 3-month US-China AI lead is dangerous; a 2-year lead could be as decisive as the Gulf War’s 100:1 kill ratio.
• Superintelligence plus robotics could compress a century of technological progress into under a decade.
• AI could enable finding and destroying stealthy nuclear submarines, breaking current nuclear deterrence entirely.
• CCP will spend billions and deploy thousands of operatives to infiltrate US AI labs once stakes become clear.
• 100GW data centers are acutely vulnerable to cruise missile strikes — a strong incentive for preemptive attack.
• Aschenbrenner’s preferred endgame: Democratic Coalition establishes clear lead, then offers China a compute-ratio treaty and galaxy-slice deal.
• Historical norm is intense great-power competition; post-WWII peace is the anomaly, not the baseline.

2024-06-06 · Watch on YouTube