Anthropic's IPO Just Revealed Something TERRIFYING?

Summary

The presenter reviews Anthropic’s purported S-1 filing ahead of a possible IPO and questions whether the company can justify a $2 trillion valuation. He describes Anthropic’s revenue growth as striking but argues that infrastructure costs, financing arrangements, long-term commitments, customer concentration, and competition make its prospects uncertain.

Financial figures and obligations cited

• Revenue: Anthropic reportedly generated $16.3 billion in the first half of 2026, more than 3.5 times its revenue for all of 2025, according to the video.
• Operating loss: $693 million in the first half of 2026.
• Infrastructure and financing costs: The presenter says Anthropic obtains computing capacity from cloud providers including Google and Amazon, partly through convertible notes that may result in equity issuance rather than immediate cash payment.
• Reported accounting impact: He cites more than $72 billion in value associated with these notes and another $8 billion in other expenses, which he adds to estimate an $80 billion net loss for the first half of 2026. This is the presenter’s interpretation of the figures discussed in the video.
• Long-term commitments: More than $500 billion in AI computing obligations over the next 10 years. The presenter says most are non-cancellable, potentially leaving Anthropic liable even if it does not use the capacity.
• Model costs: A state-of-the-art model training run may cost around $100 million, and the presenter says ongoing inference costs are substantial. He also questions the economics of the $200-per-month Claude Max plan, which he describes as subsidized.

Business and market risks

• Counterparty concentration and conflicts: Google and Amazon are described as suppliers, investors, customers, and competitors at the same time. The presenter says these intertwined relationships could threaten access to computing capacity or create other conflicts.
• Customer concentration: The top five customers reportedly account for more than one-quarter of revenue, while the top 1% of business customers account for almost half of business spending.
• Dependence on continued growth: The presenter argues that a $2 trillion valuation would require revenue to keep expanding rapidly. He says the filing’s risk disclosures do not give him confidence that triple-digit growth can continue.
• Other risks: The presenter cites public opposition to data-center construction, energy constraints, export controls and other government restrictions, and competitors developing cheaper or more capable models.
• Competition from open-weight models: He says Moonshot’s Kimi K3.1 performs as well as or better than Claude Opus 3.5 on many tasks at roughly one-third the price. He argues that cheaper models users can download and run themselves could put pressure on Claude’s market share.
• Specialized data: The presenter says expanding into more white-collar professions may require proprietary training data. He argues that companies with valuable data may keep it for their own AI products rather than provide it to Anthropic.
• AI safety disclosures: He says nearly a quarter of the filing concerns risk factors, including AI concealing information, blackmail, creating security vulnerabilities, resisting shutdown, or causing human extinction. He also says Anthropic paused training on a next-generation model for more than a month after internal security alarms were triggered.

Presenter’s view and investment stance

The presenter says he doubts Anthropic can sustain even double-digit revenue growth over the coming years, much less triple-digit growth. He personally would not buy Anthropic shares in its IPO. This is his stated opinion, not a formal valuation analysis or recommendation to viewers.

Companies, products, and instruments mentioned

• Prospective IPO: Anthropic; no ticker is provided.
• Companies: Google, Amazon, Uber, and Airbnb. Uber and Airbnb are mentioned as examples of companies that had operating losses before going public.
• AI products and models: Claude, Claude Max, Claude Opus 3.5, and Moonshot Kimi K3.1.
• Financial instruments: Convertible notes or promissory notes tied to computing capacity and potential future equity issuance.

Disclosure

The presenter says this is not financial or investment advice and that he is not qualified to provide it.

Presenter and source

• Presenter: Asian Dad Energy channel host, who describes himself as a former software engineer at a large technology company with 25 years of industry experience.
• Source discussed: Anthropic’s S-1 filing, as characterized by the presenter.