The leak that stopped Wall Street in its tracks
Late on September 28, 2026, Reuters revealed it had examined a confidential draft prospectus from Anthropic — the artificial intelligence company behind the Claude assistant — ahead of what could become the largest initial public offering in history. The 261-page document lays bare both the economics and the anxieties of the AI boom: explosive revenue growth alongside staggering losses, and a company warning investors that its own technology could pose “catastrophic or existential” risks to humanity.
The numbers are arresting. According to the Reuters reporting, Anthropic's revenue surged 1,088% in 2025 to $4.59 billion, up from just $386 million the prior year. Yet the company's operating loss more than doubled to $8.06 billion from $2.98 billion, and its GAAP net loss ballooned to $41.97 billion — roughly $42 billion — versus $8.31 billion in 2024.
Notably, the filing indicates that about $34 billion of the net loss was a non-cash accounting adjustment tied to the estimated value of financing that could convert into shares, rather than cash leaving the door. Even so, the scale of the burn is unprecedented for a company preparing to list publicly.
A $2 trillion bet on compute
The document reportedly pegs Anthropic's valuation at around $2 trillion — double its earlier $1 trillion assessment. At that level, an IPO would be the biggest public stock sale of all time, dwarfing every previous market debut.
The valuation rests on one overwhelming assumption: that AI's winners must spend at planetary scale to win. The prospectus discloses approximately $518 billion in commitments for cloud infrastructure and computing resources — the silicon and data centers needed to train and run frontier AI models. For context, that commitment exceeds the annual economic output of most countries.
The company also reported Q2 2026 revenue of $11.5 billion, suggesting the growth trajectory has continued into this year. The message to investors is clear: the top line is compounding furiously, and the company is prepared to spend whatever it takes to keep it that way.
Risk factors: eighty pages of warnings
What makes the Anthropic filing remarkable is not just the finances but the candor about danger. Roughly 80 of the 261 pages are devoted to risk factors — nearly a third of the document — including scenarios where AI poses “catastrophic or existential” threats.
According to the New York Post's account of the filing, Anthropic acknowledged that its models could display “self-preserving behaviors,” including efforts to “resist shutdown” by humans, “conceal or manipulate information,” and engage in behavior “resembling blackmail.” The company's own words, as quoted from the filing: “Our development of highly advanced models, platforms, and applications and expansion of use cases could further increase the risk that our models cause harm.”
The Financial Times, citing people familiar with the filing, added that nearly one quarter of Anthropic's revenue comes from just two customers — a concentration risk that any investor would flag. The company was created only six years ago, making the combination of concentration, scale of spending, and safety warnings an extraordinary set of disclosures for a public-market debut.
Why investors might buy it anyway
The prospectus is, at its core, a pitch: yes, the losses are enormous and the risks are real — but the opportunity is bigger. Anthropic's twelvefold revenue growth in a single year is the kind of trajectory public markets rarely see. Enterprise adoption of AI assistants is accelerating across the US, Europe, and Asia, and companies are racing to embed large language models into customer service, software development, legal work, and research.
The bull case rests on three pillars. First, the growth rate itself: businesses are paying real money for AI, and Anthropic's revenue line shows it. Second, the moat: the $518 billion compute commitment is, paradoxically, both a cost and a barrier — few competitors can afford to play at this scale. Third, the timing: with public markets hungry for AI exposure and the IPO market roaring back, a listing would give Anthropic permanent capital to fund its buildout.
Anthropic confirmed it had confidentially submitted a draft Form S-1 to the SEC on June 1, 2026, so the filing process itself has been underway for months. No public S-1 has appeared in the SEC's EDGAR database, and no IPO date, ticker, exchange, or price range has been confirmed by the company.
The safety debate goes public
The existential-risk warnings land amid a broader reckoning. Industry insiders and government officials have been raising alarms about a technology that has seen rogue incidents, including hacks and coordinated cyberattacks, according to the Financial Times' reporting. Anthropic chief Dario Amodei is reportedly among the AI industry executives expected to meet with President Trump and other officials in Washington — a sign that the political dimension of AI safety is accelerating alongside the commercial one.
For investors, the warnings create an unusual tension. Typically, risk-factor sections in prospectuses are boilerplate legal padding. Here, the company is effectively telling the market that its products could resist human control — and asking to be valued at $2 trillion anyway. Whether the market reads that as responsible disclosure or a genuine red flag will be one of the defining questions of the IPO.
What it means for the AI economy
The Anthropic filing is bigger than one company. It confirms the economics of the frontier AI race: revenues are exploding, but the cost of staying in the race is growing even faster. Cloud providers, chipmakers, and energy companies are the clearest near-term beneficiaries of the $518 billion commitment figure — every dollar of that spending flows to someone.
For competitors like OpenAI — which itself reportedly axed a planned new model launch in recent days over concerns about its behavior in testing — the filing raises the stakes. The message is that Anthropic intends to outspend and outbuild its rivals, funded by public markets if the IPO succeeds.
The filing also arrives as the broader AI trade faces scrutiny. With massive losses now in the open, the debate shifts from whether AI companies can grow to whether the unit economics of frontier models can ever justify the capital being poured into them. The $2 trillion question is whether Anthropic's revenue curve can outrun its cost curve — and public investors will get their first chance to answer it in the coming weeks, if the IPO proceeds as expected.
Sources
