Dario Amodei built Anthropic on the premise that someone needed to be the responsible adult in the AI room. In a blog post published on July 27, Amodei addressed Anthropic’s stance on open-weights AI models. His core message: Anthropic has never actually pushed for a blanket ban on open-weights models. Instead, Amodei is advocating for a more surgical approach, one that keeps the US competitive while preventing advanced AI capabilities from flowing freely to adversarial states.
The $30 billion balancing act
Anthropic reported 80-fold annualized revenue and usage growth in Q1 2026. The company was sitting at approximately $9 billion in annualized revenue at the end of 2025. Four months later, it had more than tripled that figure, reaching an annualized run rate of roughly $30 billion by April 2026.
Rather than calling for sweeping restrictions on open-weights models, Amodei outlined a set of targeted measures: export restrictions on advanced chips flowing to China, enforcement actions against industrial-scale model distillation, and mandatory safety testing for any model that hits certain capability thresholds, regardless of whether its weights are open or closed.
The open letter Anthropic wouldn’t sign
