Dario Amodei on AI bankruptcy risk

by Vojtech: Dario Amodei

The gap between Anthropic's $50 billion commitment and OpenAI's over $235 billion in commitments isn't just about ambition levels; it's a direct result of Dario Amodei refusing to match the compute spending levels of competitors. His reasoning comes down to a brutal math problem where insufficient revenue relative to compute purchases would lead to bankruptcy regardless of hedging strategies. He outlines a hypothetical scenario where revenue grows 10x annually, reaching $100 billion by end of 2026 and $1 trillion by end of 2027. Amodei calculates that purchasing a trillion dollars of compute starting at the end of 2027 would require five years of such revenue. Being off by even one year or having revenue of $800 billion instead of $1 trillion would make bankruptcy unavoidable when buying that amount of compute. "if my revenue is not $1 trillion, if it's even $800 billion, there's no hedge on earth that could stop me from going bankrupt if I buy that much compute." This explains why he refuses to purchase that level of compute based on that uncertainty despite internal doubt about whether 10x growth will continue. The author argues that missing a projected growth curve by a single year creates a risk of insolvency. It frames the situation as a quiet warning regarding the AI boom's financial fragility. The text suggests Anthropic is growing 10x a year, making $1 trillion revenue by late 2027 a logical endpoint if the trend holds, yet the cost of being wrong is too high.