Anthropic filed an amended S-1 on Monday, and there’s a lot in it.
The money
- Q3 revenue was about $17 billion, up from $11.5 billion the quarter before. Revenue for the first nine months was $33.2 billion, up 1,206% year over year.
- Anthropic warns that growth like that won’t continue and that it can’t promise revenue will keep growing at all.
- Compute ate 58% of revenue in Q3, up from 48%. The operating loss grew too.
- Compute and infrastructure commitments are close to $900 billion and could top $1 trillion. About $447.5 billion of that is expected by the end of 2030.
The warning
TechCrunch spotted new risk-factor language about recursive self-improvement (RSI), the point where AI meaningfully speeds up building better AI. Anthropic says it uses AI in its own R&D and may have built models in the past whose jumps in capability it didn’t recognize at the time. It also says capabilities could start advancing faster than its safety work can keep up.
My take
Claude is one of the tools I lean on, so I want them healthy. But a $1 trillion compute bill on top of “we might not notice what our own models can do” is a lot to put in one filing. I’d love to see the safety side spelled out as clearly as the money.
God Bless, and May the Source Be with You!
