Anthropic’s business plan: Build God or die trying

The Verge
Anthropic's S-1 reveals massive compute commitments that require explosive growth, possibly through recursive self-improvement, to avoid bankruptcy.

Summary

Anthropic's S-1 filing shows the company has committed to approximately $850 billion to $1 trillion in compute and infrastructure over the next 20 years. To meet these obligations, the company must sustain its current rapid revenue growth, which grew fourteen times to $16.3 billion between the first half of 2025 and the first half of 2026. While Anthropic has been more disciplined than competitors like OpenAI, focusing on enterprise clients and achieving a 900 percent net revenue retention rate, it faces intense competition from other frontier labs and open-source models. The company has no alternative revenue paths like Google's, and its model releases drive revenue spurts, forcing continuous training. Anthropic's risk factors acknowledge that using AI to develop more AI may advance capabilities faster than safety measures, and the business plan appears to require dramatic recursive self-improvement. If such self-improvement is not achievable, meeting compute commitments and turning a profit becomes significantly harder.

(Source:The Verge)