Anthropic

Anthropic enters the ToTheMoon conversation later than the other labs, but by 2026 it is one of the central characters: a company betting on expensive professional work and enterprise customers, publishing research on how models behave, and the first to find its model restricted as a strategic asset. The hosts ask whether that position pays off and who ends up setting the rules: the company, the state or the paying user.

In late 2025 the discussion was about money. Episode 088 set OpenAI's and Anthropic's revenue plans side by side and recalled that Anthropic's last recorded revenue was a billion. Episode 096 called the rivalry with OpenAI a war of business models, advertising versus expensive professional work, and episode 126 noted that roughly 80% of AI investment is concentrated in two companies because the rest of the ecosystem cannot exist without base intelligence. Episode 124 added the DXC partnership and a move into banks and legacy enterprise code.

In the summer of 2026 the channel covered Anthropic weekly: episode 120 discussed the admission that Claude helps build its own systems, episode 122 the shutdown of Fable 5 at the demand of the US government — a right it had itself requested. Toward autumn the hosts turned to research: episode 153 covered experiments where individually safe agents form an unsafe system, and episode 156 the Guideline assessment, where its monitoring earned a three. Episode 166 added two signals from inside: Jacob Coxon's departure with a warning about self-improving superintelligence, and the report of five suspicious biological requests to Claude, after which restrictions were tightened. Episode 168 added three storylines: Amodei's call to admit outside checkers, a court ruling that the ban on Anthropic in the US defence complex is unlawful, and the report on cheaper cyberattacks with Claude.