Anthropic is working both sides of the AI power dynamic. On one side, it’s building the most capable models. On the other, it’s lobbying to make sure few others can follow.
David Sacks argued on *All-In* that Anthropic’s push for a “permissioning regime” in Washington is a classic play for regulatory capture. The rules it advocates would require labs to get government approval before releasing new models or selling chips, a hurdle only well-funded incumbents could clear.
This lobbying push runs parallel to its technical sprint. The company confirmed the existence of Claude Mythos, a model it calls a “step change” in performance, after details leaked from an unsecured database. Mythos is built for heavy-duty coding, academic reasoning, and cybersecurity.
David Sacks, All-In with Chamath, Jason, Sacks & Friedberg:
- Anthropic is sort of the most AGI-pilled of all the frontier labs.
- They made this bet on coding as their way to get to recursive self-improvement.
The coding bet is its commercial engine. By dominating developer tools and APIs, Anthropic reportedly added $6 billion to its annual run rate in a single month. Its new “Computer Use” feature moves the model from a chat interface to an autonomous desktop agent, embedding it deeper into enterprise workflows.
While Anthropic guards Mythos for security testing, OpenAI is retreating from consumer experiments. It shelved an “adult mode” after safety failures, consolidating around enterprise and coding to compete directly. OpenAI’s revenue is mostly consumer subscriptions; Anthropic’s is nearly all enterprise API usage.
Chamath Palihapitiya, All-In with Chamath, Jason, Sacks & Friedberg:
- OpenAI is three-quarters consumer subscriptions and a quarter API.
- Anthropic is almost the exact opposite.
The race is escalating toward a liquidity event. Rumors point to an Anthropic IPO as early as October, with Sam Altman wanting OpenAI to go public first. This financial pressure prioritizes moat-building - through both regulation and technical superiority - over open exploration.
Anthropic’s strategy is clear: advance the frontier, then lobby to fence it off. The result is a market where technical prowess and political influence compound, leaving little room for new entrants.
