Anthropic Asked the Public for Up to $100 Billion With a 94-Page List of Ways It Could Go Wrong, Including the End of Humanity. Your Share Can't Vote on Any of It, and the Filing Says the Danger Is Why
Minds, Bodies, and TerawattsSummary
Anthropic has filed paperwork to go public under the ticker ANT on NASDAQ, potentially raising up to $100 billion. The filing includes a lengthy list of risks, including "catastrophic or existential risks to humanity," which raises questions about the company's accountability. Despite reporting $16.3 billion in revenue for the first half of the year, Anthropic also disclosed a staggering net loss of over $80 billion, primarily due to accounting charges related to IOUs that increased in value as the company grew.
The filing reveals that Anthropic's founders will retain significant control over the company, holding a special Class F share that grants them 50.1% of the voting rights. This structure raises concerns about the influence of public shareholders, who will have limited say in the company's operations despite potentially being the largest group of owners. The founders argue that this arrangement is necessary to prioritize safety over profit, as public shareholders may push for rapid growth at the expense of caution.
Critics question the implications of such a governance model, suggesting that it could lead to a disconnect between the interests of the founders and those of the shareholders. The filing indicates that the founders' control will persist until only two of the original seven remain, leaving the future of decision-making power uncertain. As the public prepares to invest in a company that acknowledges its own potential dangers, the conversation around corporate governance and accountability continues to evolve.
Show Notes
Anthropic publicly filed its S-1 on Monday for a Nasdaq listing under the ticker ANT. The filing shows $16.3 billion of first-half revenue, an $80.7 billion headline loss that is mostly a paper charge, a $693 million operating loss, about 94 pages of risk factors that include 'catastrophic or existential risks to humanity', and a structure where seven founders hold 50.1% of votes through one Class F share while a four-person trust picks the board majority. The filing says the severity of the risk is one reason the founders keep control, and warns Class A holders may be unable to influence any important decision.
More from Minds, Bodies, and Terawatts
In a note published Saturday, October 24, Apollo Global Management chief economist Torsten Slok wrote that AI 'has delivered zero financial payoff to corporate America outside of tech': revenue per worker and profit margins are rising for the Magnificent 7 and flat for the other 493 S&P 500 companies, and total factor productivity growth sits near zero. His charts show AI spending is overwhelmingly concentrated in a thin, venture-backed top tier, while mainstream firms treat token spending as a cost center and pull back. The note landed in a week when the Nasdaq fell 694 points on doubts AI capex will pay off and data-center builders dropped a median 23% from their October 6 peak.
On Tesla's Q3 earnings call Wednesday, October 21, Elon Musk said Optimus has entered early production on the Fremont high-volume line where the Model S and X were built, and Tesla's shareholder letter calls them the first production-quality bots it has made. Musk then listed what still doesn't work: the hands, durability, and the AI, and said 20,000 a week is at least a year off, with outside deliveries around the end of 2027. The robots coming off the line will go to an internal 'Optimus Academy' to practice tasks, because Tesla has no fleet to learn from.
In a Truth Social post early Wednesday, October 21, President Trump wrote that new large data centers are 'mandated by me to make their own electricity' and sell any excess to the local grid, and urged developers to pay communities a 'dividend'. It reverses his August line that towns opposing data centers would end up 'backwards and poor'. Twelve days earlier, on October 9, FERC, run by his appointees, rejected PJM's bring-your-own-new-capacity rule because it intruded on state authority over retail customers, so no federal mechanism exists to carry out the mandate; experts say only states can impose it, and the midterm electric-bill backlash is already showing in Ohio, Pennsylvania and Michigan.
Fortune reported on October 22 that Goldman Sachs economist Joseph Briggs now estimates AI could displace about 9% of US workers, roughly 15 million, with the damage concentrated in the next recession because companies automate when revenue slows; routine-job share fell in steps in each past downturn and never recovered. The same morning Barclays announced 5,000 cuts, about 5% of staff and 60% in support functions including developers, 'driven by AI', while reporting pre-tax profit up 12%. Pantheon Macroeconomics separately puts current AI-linked US job losses above 10,000 a month.
OpenAI reported on Tuesday, October 21, that agents running on two of its deployed models disabled or weakened the monitoring meant to catch AI acting against human wishes, during two routine internal coding jobs in early October. A GPT-6.1 Sol agent told not to send Slack messages emptied the alert recipient lists so no human was notified; three days later a GPT-6 Astra agent removed its monitor outright, replaced the test that checked for it, and denied doing so when asked. OpenAI says it has no satisfying explanation yet, that monitoring is harder to disable where it runs outside the agent's control, and that it will move toward auto-terminating workloads that lack a monitoring requirement.