Anthropic Prepares to File Its IPO Prospectus as Soon as Friday
Anthropic plans to make its IPO prospectus public as early as Friday, according to people familiar with the matter, a step that would put the financial internals of the Claude developer in front of any investor with a browser. Reports of the timing circulated Thursday, citing three people briefed on the plans.
A public filing is a formality with consequences. Once the S-1 lands in the Securities and Exchange Commission’s database, the company’s revenue, losses, customer concentration and risk factors stop being a matter of speculation among venture investors and become a document that carries legal weight. A listing could follow in November, depending on market conditions, though the date remains tentative.
The numbers in the draft are the ones drawing attention. Anthropic generated $16.3 billion in revenue in the first half of 2026, according to a draft prospectus reviewed by The Wall Street Journal, up from $1.1 billion in the same period a year earlier. Its net loss over those six months reached $80.7 billion.
That loss figure needs reading before it needs reacting. Much of it reflects changes in the valuation of convertible notes rather than cash leaving the business, according to accounts of the filing, a common feature of pre-IPO accounting at companies whose private price rose sharply within the reporting period. Analysts said the operating picture is better captured by the cash spent on computing, which came to nearly $9 billion in the first half, more than the $7.3 billion Anthropic spent on infrastructure in all of 2025.
The commitments are larger than the spending. The company has locked in roughly $450 billion of cloud computing agreements over the next decade, including $111 billion with Google Cloud, $110 billion with Amazon Web Services and $31 billion with Microsoft, according to the Journal. It told prospective investors it could spend as much as $750 billion in coming years on additional capacity, including deals still being negotiated.
Those obligations are noncancelable in the main, which converts a growth story into a fixed-cost business with a long tail. For every dollar of revenue the company books, a substantial share is already pledged to the three cloud providers that also happen to be its largest strategic partners and, in two cases, its investors. That arrangement concentrates both supply and governance in a small circle.
Expectations around the offering have been expansive. Wall Street estimates have put the company near a $2 trillion valuation with a raise of as much as $100 billion, figures that would place it among the largest listings in history. Reports this week said some investors are marking their expectations down toward $1.5 trillion as they work through the risks disclosed in the draft.
Those risks are the usual ones for a company at this stage, plus a few specific to frontier AI. Customer concentration tops the list: a business with $16 billion of half-year revenue is still dependent on a limited set of large enterprise accounts and API customers. Capital intensity runs a close second. Third is exposure to liability tied to how the models are used, an area where the legal perimeter is being drawn case by case and no reserve is easy to size.
The revenue trajectory itself invites comparison. Growing from $1.1 billion to $16.3 billion in twelve months is the steepest climb any enterprise software company has recorded at this scale, and it happened while the company was building data centers and hiring researchers at the same time. Reports on the draft said third-quarter revenue continued to rise and that the company reached adjusted profitability in the period, a measure that excludes some of the costs public investors will care about most.
For the AI industry, a public Anthropic sets a mark. Competitors raising private capital will be priced against a disclosed set of financials rather than a narrative, and the cloud providers that host these workloads will face questions about how much of their backlog depends on customers with this cost structure. Employees holding equity will get a market price for the first time, and early investors will get liquidity they have been waiting years for.
There is also a strategic dimension to filing now. A public company must disclose quarterly, forecast publicly and answer to shareholders who may not share a safety-first governance philosophy. The structure Anthropic has built around its board and its long-term benefit trust will be examined line by line by lawyers looking for conflicts, and the answers will shape how other labs structure their own conversions from private to public.
Chief executives who take a company public on this scale usually describe the moment as a beginning. The document says something narrower: that the cost of staying at the frontier is now a matter of public record, and that the money to pay for it has to come from somewhere other than the venture funds that built the business.
Whether public investors will underwrite $750 billion of future computing commitments on the strength of a single year of revenue growth, or demand that the pace of spending slow to something a quarterly earnings call can defend, is the question the roadshow will answer.
