Anthropic is preparing to sell shares to the public in mid-November at a valuation that investors now discuss at around $1.5 trillion, down from the $2 trillion the company and its bankers had floated, while OpenAI races to close a roughly $30 billion private round at more than $1.4 trillion before Anthropic's numbers become public. This is the first time public markets will be asked to put a daily price on a frontier AI lab, and the answer will set the cost of capital for the entire industry. The two filings and the roadshow math, not the model benchmarks, are now the most important documents in AI.
What is actually happening with the Anthropic IPO
Anthropic confidentially submitted a draft S-1 to the SEC on June 1, 2026, after hiring Wilson Sonsini and talking to banks in late 2025. A draft prospectus was then seen by Reuters on September 28 and reported on September 29, revealing revenue, losses and a $518 billion infrastructure commitment. On October 14 the company held an investor day in San Francisco, the first time potential public investors saw the numbers directly. Trading could begin as early as next month, and bankers have said the company could aim to raise more than $100 billion, according to The New York Times.
At $2 trillion, that would be the largest IPO ever, passing SpaceX, which went public in June at a $1.77 trillion valuation. At $1.5 trillion it would still be the largest technology listing in history. The deal's lead banks are reported to be Morgan Stanley, Goldman Sachs and JPMorgan Chase, per GraniteShares research.
The timing matters because Anthropic's public filing has a hard rule attached. Under the JOBS Act, the company must publish an amended S-1 at least 15 days before it markets the offering to institutions. That is the moment the full financials become visible to everyone, including OpenAI's investors.
The revenue numbers behind the $2 trillion ask
The growth is genuinely extraordinary. Anthropic told investors at its October 14 meeting that third-quarter revenue rose 45 percent from the prior quarter to $16.8 billion, and that it was profitable on an adjusted basis, Reuters reported. Second-quarter revenue was $11.6 billion, up from $4.73 billion in the first quarter.
The annualized picture is what bankers are selling. Revenue run rate went from about $9 billion at the end of 2025 to $47 billion in mid-May and $65 billion by the end of July, according to GraniteShares. On the third-quarter number, that is roughly $67.2 billion annualized.
The historical income statement is less flattering. In fiscal 2025, Anthropic booked $4.6 billion in revenue against $13 billion in operating expenses, an operating loss above $8 billion and a net loss of $42 billion, largely from writedowns tied to earlier fundraising liabilities, per Fortune's read of the leaked S-1. It ended the year with $20.28 billion in cash. The company has said it expects a second straight quarter of adjusted operating profit, which is the number its bankers will lean on.
The problem is the second derivative. Revenue more than doubled from the first quarter to the second, then grew 45 percent in the third. For a company being priced on future growth, deceleration is the single most important input, and it is the reason the valuation conversation moved.
Why investors moved the price to $1.5 trillion
The arithmetic is simple enough to do on a napkin. At $2 trillion on $67.2 billion of annualized third-quarter revenue, Anthropic would trade at about 29.8 times sales. At $1.5 trillion, the multiple falls to roughly 22.3 times, Reuters calculated.
That is not obviously expensive by 2026 AI standards. LSEG data cited by Reuters puts Palantir at about 53 times 2026 estimated revenue, and SpaceX and Cloudflare near 41.6 times. The catch is what has to be true for the higher number. A $2 trillion valuation implies roughly 10 times the $190 billion to $200 billion of 2028 revenue Anthropic has projected internally, which requires the company to roughly quadruple again from its July run rate in about 18 months. Investors who attended the investor day told Reuters they were disappointed by the slowdown and argued for $1.5 trillion instead.
There is also a concentration problem buried in the filing. A quarter of Anthropic's revenue comes from just two clients, Fortune reported. That is normal for an enterprise-first AI business selling through cloud partners, but it makes the revenue base less diversified than the headline growth suggests.
The $518 billion compute bill
The most consequential number in the prospectus is not revenue. It is the commitment to spend at least $518 billion over a decade on cloud services, computing and infrastructure.
The breakdown, as reported by The Next Web, is $111.1 billion with Google, $110 billion with Amazon and $31.4 billion with Microsoft, spread over seven to ten years. Anthropic also carries about $161.2 billion in equipment leases tied to Broadcom, which has sought $60 billion in debt to fund chips. A deal with xAI could bring up to $84.5 billion of Nvidia-based capacity through 2029, but most of that is cancellable with 90 days' notice. AMD has committed to buy up to $5 billion of Anthropic stock and to supply capacity expected to exceed $20 billion.
Some of these contracts are take-or-pay. "If our actual spend falls short, we must pay Google the difference," Anthropic said in the filing, according to Reuters. That sentence converts a variable cost into a fixed obligation, and it is the kind of disclosure that public investors punish when growth slows. The company also warned that demand for AI will be "limited principally by the availability of compute," which frames the spend as a defensive necessity rather than an option.
Read together, the revenue and commitment numbers describe a business that must grow into an enormous fixed cost base. At $16.8 billion a quarter, the company is annualizing about $67 billion against a $518 billion decade-long commitment. The math works only if enterprise adoption keeps compounding for years, which is exactly the bet public shareholders will be asked to underwrite.
OpenAI's pre-emptive $30 billion raise
OpenAI is not waiting for that verdict. On October 19, Reuters reported that OpenAI is racing to secure investor commitments before Anthropic's mid-November debut, in an unusually fast round of about $30 billion that could value the ChatGPT maker at more than $1.4 trillion. SoftBank, Abu Dhabi's MGX and other existing backers are in talks to join, according to Reuters.
The motivation is partly defensive. Both labs are tapping the same pool of long-only funds and sovereign wealth funds, and OpenAI wants capital locked up before Anthropic's disclosures invite direct comparison. "OpenAI could want to set its own valuation and not have to worry about a comp," Edward Best, co-chair of capital markets at Willkie Farr & Gallagher, told Reuters. "It's really a two-horse race in terms of being public."
OpenAI is pitching acceleration after the launch of GPT-6 Astra. It told investors it had reached $50 billion in annualized revenue and expects $70 billion by year end, and that price cuts on cheaper models have boosted usage. It was last valued at $852 billion in March after a record $122 billion round that included a commitment of up to $50 billion from Amazon. The company also needs the money to fund infrastructure commitments and to shore up its balance sheet in case a weaker IPO market delays its own listing, which is planned for 2027.
One accounting fight will follow both companies into public markets. OpenAI records revenue net of partner shares, while Anthropic counts the gross value of contracts sold through cloud providers. OpenAI has accused Anthropic of inflating revenue; Anthropic says it follows established accounting practices. Public filings will settle that argument with audited numbers, and the answer will change how the two compare.
The market backdrop is not cooperating
None of this is happening in a calm market. The Philadelphia Semiconductor Index has shed $1.6 trillion in less than two weeks. CoreWeave's shares and debt tumbled as investors scrutinized nearly $50 billion in borrowings used to fund its expansion. AirTrunk REIT, Nvidia-backed Firmus and SoftBank-backed SB Energy have pulled or paused listings, and nuclear services company Holtec withdrew its US filing, citing a "perfect storm" around data-center demand, Reuters reported.
The macro backdrop is equally difficult: volatile oil prices, rising bond yields and growing concern about AI safety are all weighing on technology stocks. Bankers say a number of companies preparing to go public have decided to wait until after Anthropic's IPO rather than compete for demand. That is a striking reversal from the start of the year, when the AI IPO queue looked like a certainty.
Anthropic is therefore listing into a market that is skeptical of exactly the trade it represents: high growth, heavy capital intensity and a valuation that depends on forecasts two years out. If the deal prices well, it validates the entire neocloud and AI infrastructure complex. If it prices at $1.5 trillion or below, or breaks issue, the read-through hits every data-center landlord, chip supplier and private AI company at once.
What changes when the S-1 goes public
Anthropic devoted roughly 80 of 261 pages to risk factors, and more than a third of the document to warnings about the business, including the possibility that advanced AI poses "catastrophic or existential risks to humanity" and could "manipulate, blackmail and exhibit other unpredictable behaviours," according to the Financial Times as cited by Fortune. That language is unusual in an IPO document, and it signals that the company's own counsel sees model behavior as a material business risk, not a philosophical aside.
Public ownership changes the operating rhythm. Anthropic will report quarterly, guide on revenue and capex, and face questions about customer concentration, contract liabilities and the gap between adjusted and GAAP profit. Employees and early investors will eventually get lockup expirations, which will test the float. Index inclusion will bring passive demand, but only after the stock has traded for a period.
The near-term calendar is narrow. The public S-1 must appear at least 15 days before the roadshow, so the full numbers are likely to land in late October or early November. Pricing is expected around mid-November. OpenAI's round is expected to close faster than is typical for its size, which means both capital raises could be settled within weeks of each other.
The deeper question is what public markets will decide about AI economics. Private investors have been willing to fund losses on the promise of scale. Public investors will ask for a path to durable margins, and they will mark the stock every day based on whether the next quarter's growth supports the multiple. Anthropic's answer, and OpenAI's response to it, will determine whether the AI buildout is financed by equity markets at these valuations or forced to reprice.
The first real data point arrives in November. Until then, the only honest summary is this: the models are no longer the story. The contracts, the losses and the price of capital are.
Sources: Reuters, October 19, 2026; Reuters, October 15, 2026; The New York Times, October 14, 2026; Fortune, September 29, 2026; The Next Web; GraniteShares.