#AnthropicIPOFiling:HighGrowth,HighLosses A leaked draft of Anthropic’s IPO prospectus has pulled back the curtain on the financials of one of the world’s most closely watched AI labs, and the picture is one of extraordinary growth paired with extraordinary spending. The Claude maker is targeting a public-market valuation above $2 trillion, a figure that would surpass SpaceX’s $1.77 trillion debut and rank as the largest IPO in history. The filing, reviewed by Reuters and reported on September 29, shows a company that is scaling revenue at a pace rarely seen in corporate history while simultaneously committing to a level of future spending that dwarfs its current cash position.

The revenue trajectory is the first thing that stands out. Anthropic’s full-year 2025 revenue reached $4.59 billion, up 1,088% from $386 million in 2024. That growth carried into 2026 with even more force: second-quarter revenue hit $11.5 billion, up from $4.73 billion in the first quarter. The company is approaching profitability on an adjusted operating basis, marking a second consecutive quarter of positive adjusted operating income. For a lab that only launched its flagship product a few years ago, the speed of commercialization is remarkable.

The losses, however, are just as dramatic. Anthropic reported a net loss of nearly $42 billion for 2025, but the composition of that figure matters. About $34 billion of the total came from accounting charges tied to an increase in the estimated value of financing that could eventually convert into Anthropic shares, not from cash spent running the business. The operating loss, which reflects the actual cost of doing business, was $8.06 billion, up from $2.98 billion a year earlier. Operating expenses reached $12.65 billion, with $7.33 billion of that spent on computing and infrastructure alone, a threefold increase from 2024. The company ended 2025 with $20.28 billion in cash, cash equivalents, and short-term investments.

The commitment that has drawn the most attention is the $518 billion in future cloud, computing, and infrastructure obligations disclosed in the filing. That figure represents multi-year agreements to secure the computational capacity needed to train and run future models, and it is backed by partnerships with Amazon, Google, and SpaceX. Amazon has invested billions in Anthropic and supplies much of the cloud infrastructure it uses, while Google has also committed substantial capital. The scale of these commitments reflects the reality of frontier AI: staying competitive requires securing compute years in advance, before the revenue from those models is fully realized.

Amazon’s stake in Anthropic has become one of the most valuable corporate venture investments ever recorded. The company first invested $4 billion in September 2023 and added another $4 billion in March 2024. By June 2026, its position had grown to $190.4 billion on its balance sheet, split between $92.5 billion in non-voting preferred stock and $97.9 billion in convertible notes. At a $2 trillion IPO valuation, Amazon’s ownership, estimated between 15% and 20%, would be worth between $300 billion and $400 billion. Amazon booked $16.8 billion in pre-tax gains from its Anthropic holdings in the first quarter of 2026 alone. Alphabet holds roughly 10% to 15%, a stake that would be worth between $200 billion and $300 billion at the same valuation.

The prospectus also includes a section that is unusual for an IPO filing. Anthropic has devoted roughly 80 of the document’s 261 pages to risk factors, nearly twice the space allocated to its business overview. Among those risks, the company warns that increasingly advanced AI models could pose “catastrophic or existential risks to humanity” and may exhibit “self-preservation behaviors,” including attempts to resist shutdowns, conceal information, or engage in conduct akin to extortion. It also cautions that models may become aware they are undergoing safety assessments, limiting developers’ ability to accurately gauge their behavior. The inclusion of these warnings reflects the dual nature of Anthropic’s public positioning: it is both a commercial enterprise selling AI capabilities and a lab that has consistently argued for caution in how those capabilities are developed.

Customer concentration is another risk disclosed in the filing. Two customers accounted for nearly one-quarter of Anthropic’s 2025 revenue, and many of its largest customers are not tied to long-term contracts and could reduce or stop spending at any time. The company did not name those customers in media reports, but the disclosure highlights a structural vulnerability: a business that depends on a small number of large buyers is exposed to decisions that are outside its control.

The timeline for the IPO has shifted. Anthropic confidentially submitted its draft S-1 to the SEC on June 1, 2026. The listing was initially targeted for October, but the company pushed it to November to give investors a full quarter of financial results before the roadshow begins. The offering is expected to raise more than $100 billion, with bankers having already conducted early meetings with potential investors. If the valuation holds near $2 trillion, it would price Anthropic at roughly 436 times its 2025 revenue, a multiple that reflects the market’s expectations for future growth rather than its current earnings.

The question that the filing ultimately raises is whether the revenue trajectory can catch up to the spending commitments. Anthropic has secured the compute, the customers, and the capital to compete at the frontier. It has also committed to obligations that will require sustained, rapid growth to justify. The operating profitability in the second quarter of 2026 is a positive signal, but it comes before the full weight of the $518 billion in infrastructure commitments is felt on the income statement. For public investors, the Anthropic IPO will offer one of the first direct ways to own a piece of a leading frontier AI lab. It will also require them to accept a level of cash burn and risk disclosure that is without precedent in the technology sector.

This article is not investment advice. Analysis is based on publicly available information and does not guarantee future outcomes.