Founder control, Trump risk, giant losses: Anthropic’s IPO filing

Anthropic, an artificial intelligence company, has released its IPO filing, revealing staggering financial figures and unique governance structures. The company reported a net loss of $81 billion for the first half of the year, while also having over $400 billion in future obligations related to chips and data center leases, many of which are yet to be constructed.

The IPO filing highlights the company's deep ties with other players in the AI ecosystem, with Amazon and Google as early investors. However, Anthropic's governance structure is quite distinct, with its seven co-founders holding control through a single golden share of supervoting stock. A separate council of "financially disinterested trustees with expertise in AI safety" will select a majority of the board of directors.

This setup stands out among tech peers, as it gives founders a significant amount of control compared to other companies.

Anthropic faces considerable risks, including a key-man risk posed by US President Donald Trump. The filing references a February spat between Trump and CEO Dario Amodei over the use of Anthropic's models in war, which could lead to government restrictions on the company's models. Additionally, nearly half of Anthropic's revenue comes from cloud service providers, making them vulnerable to changes in contracts and policies.

The filing also touches on the company's stance on AI safety, warning of potential existential risks posed by sufficiently advanced AI systems. However, China is only briefly mentioned, despite being a major competitor in the AI field. The introduction letter from Anthropic's leadership was found to be human-written, suggesting an authentic voice behind the company's mission.