Peter Diamandis comments on reported plans for Anthropic’s IPO structure, focusing on governance and control in a company building frontier AI models. He says Anthropic already has an unusual control arrangement via a “long-term benefit trust”, contrasting that with reports the firm is considering super-voting shares that would preserve founder influence after listing. Diamandis places the debate in a historical context, recalling that founders once typically relinquished special voting rights at IPO, before dual-class structures became common with major Silicon Valley listings such as Google and Meta. He then argues that installing or expanding founder super-voting rights at this stage would be atypical, adding that he has not previously encountered it. The segment also includes speculative discussion about the rationale for concentrated control at an AI company: one line of argument presented is that founders may believe they are best placed to manage safety risks and prevent outsiders from redirecting the firm’s mission. Diamandis counters with a concern about the broader implications of concentrating decision-making power in a small group at a company pursuing highly capable AI systems, describing the prospect as “bizarre”. No documents are presented in the audio, and the claims about Anthropic’s IPO planning are framed as based on reporting rather than first-hand involvement.
Peter Diamandis argues reports that Anthropic may add founder super-voting shares at IPO are unusual. He says super-voting became common in Silicon Valley listings but claims he has not seen it installed retroactively at the point of going public.
Read this moment →