Anthropic Needs Twice Nvidia’s Profits for a $2 Trillion Tag

Before any long-term investor opens a prospectus, they ask a simpler question: is this a business, or is it a project wearing a business’s clothes? The Anthropic IPO forces that question with unusual clarity.

Research firm New Constructs wrote in a note this week that Anthropic’s IPO would be the “most ridiculous IPO of 2026,” posing a historic risk to financial markets. The firm said Anthropic would need profits equal to twice Nvidia’s trailing four-quarter total to support a $2 trillion valuation. Nvidia’s net income over the most recent four quarters is roughly $193 billion. Anthropic has not earned $193 billion in revenue across its entire existence, let alone generated that in profit.

Reuters, citing a leaked prospectus, reported 2025 revenue of $4.6 billion and a net loss of $42 billion. But Reuters also reported that a large share of that net loss reflected non-cash accounting charges tied to financing rather than the pure cost of running the business. The revenue number has moved fast since then. Reuters reported that Anthropic’s annualized revenue run rate topped $65 billion at the end of July. Reports citing investor materials and the New York Times have said the company could reach a revenue pace above $100 billion in annualized terms by the end of 2026 and that it recently posted an operating profit that excluded stock-based compensation. The bear case points out that operating profit calculated before stock-based compensation is a different animal than the kind of earnings a discounted cash flow model requires.

New Constructs noted that Anthropic still has not made its prospectus public, so the firm has not seen the actual filing. That means every number in circulation, good and bad, carries an asterisk. The Claude maker confidentially submitted a draft S-1 registration statement to the SEC on June 1, 2026.

Timing compounds the complexity. The Federal Trade Commission has opened a broad probe into the safety of AI systems, including Anthropic and OpenAI, looking into allegations of unfair or deceptive acts and potential harms to consumers. Regulatory overhangs rarely kill an offering, but they do sharpen the questions investors should be asking about long-term operating risk.

New Constructs also argued that the offering would mainly create liquidity for Anthropic’s Wall Street backers rather than generate wealth for public-market investors. That charge is leveled at almost every richly priced IPO and is often wrong. But the arithmetic here is harder to dismiss. Anthropic has been targeting an October 2026 listing with a valuation potentially as high as $2 trillion, against a Series H round in late May 2026 that valued the company at $965 billion. A doubling of valuation in five months, with no public financials and an open regulatory file, is not obviously a price that rewards patience.

The genuine bull argument is not that the $2 trillion figure is fair today. It is that Anthropic’s revenue trajectory is unlike almost anything in software history and that the enterprise API model generates stickier economics than the consumer chatbot businesses it is routinely compared to. That is a real competitive characteristic.

The question great investors ask is not whether Anthropic is growing. It clearly is. The question is whether the $2 trillion price already accounts for a decade of flawless execution and leaves nothing on the table for the buyer. New Constructs thinks the answer is yes, by roughly a factor of fourteen.