Anthropic’s $2 trillion IPO question: How much growth can investors buy?
The AI company is on course for roughly $60 billion in annual revenue, but its latest financial disclosures also reveal steep computing costs, billions in underlying losses and hundreds of billions in infrastructure commitments.
Another day, another leak from Anthropic’s anticipated IPO prospectus. Investors are waiting for the official filing, which is expected to shed light on the finances of the AI giant, much of whose business has remained hidden from public view as a private company. If Anthropic intends to go public in 2026, ahead of rival OpenAI, it faces a narrowing window. Thanksgiving is approaching, followed soon afterward by the Christmas season. There has been talk within Anthropic’s orbit of a roadshow, a series of formal meetings with prospective investors, beginning on November 9. That timetable would require the company to publish its prospectus imminently, giving investors time to examine the figures before the offering.
Some observers suspect that the financial details are being released in a carefully timed drip-feed, gradually preparing investors for the contents of what could be one of the most consequential financial documents of the year. The theory is that the disclosures will allow the market to absorb the surprises in advance, reducing the risk of a sharp reaction when the full picture emerges. If so, the strategy is likely to be especially useful for the less flattering numbers.
The Wall Street Journal’s report this week offered the most comprehensive picture yet, moving beyond individual metrics to provide a broader view of Anthropic’s financial position. The revenue figures were particularly striking. Third-quarter revenue reached $16.8 billion, up 45% from the second quarter and exceeding the $16.3 billion reported for the first half of 2026. For comparison, Anthropic generated $1.1 billion in revenue during the first half of 2025.
The latest figures help address concerns raised by an earlier leak suggesting that growth had begun to slow. Even if the monthly growth rate has moderated, Anthropic’s expansion remains extraordinary by almost any conventional measure.
There are, however, vulnerabilities beneath the headline numbers. Anthropic remains heavily dependent on a relatively small group of customers. Just two accounted for 20% of its revenue in the first half of the year. Such concentration creates risk if a major customer reduces its spending or changes suppliers. The counterargument is that, as more businesses integrate AI into their everyday operations, Anthropic could build a broader and more diversified customer base.
The expense figures are harder to dismiss. They are alarming even against a backdrop of investor expectations that frontier AI models will require enormous amounts of capital. That is particularly notable given recent suggestions from Anthropic representatives that the company had achieved profitability in the previous quarter.
In the first half of 2026, Anthropic reported a loss of $80.7 billion. Some $72 billion of that amount was an accounting loss associated with debt expected to convert into equity at a later stage. Reuters has reported an accounting loss of $42 billion for the previous year, with the increase reflecting the rise in Anthropic’s valuation following a subsequent private funding round.
Stripping out that accounting effect still leaves a loss of $8.7 billion. The bulk of the underlying expense burden comes from computing. Anthropic spent $9 billion on cloud services, chips and other computing costs in the first half of 2026, up sharply from $2.3 billion in the same period a year earlier.
Even those figures may not capture the full scale of the financial commitments facing the company. Anthropic has signed contracts for AI infrastructure totaling $600 billion over the coming decade and is negotiating additional agreements worth another $150 billion. Google and Amazon are each reportedly tied to contracts valued at $110 billion, while Microsoft has a smaller agreement worth $31 billion.
For Anthropic, these commitments represent a formidable obligation to keep spending, even if growth slows or the economics of AI change. For the broader industry, however, they offer a powerful source of reassurance. Chipmakers, cloud providers and data center operators can point to contracts that underpin at least part of the investment pouring into AI infrastructure.
This is also where the industry’s circular financing comes into focus. Investors backing infrastructure providers have a financial interest in ensuring that major AI customers such as Anthropic remain sufficiently funded to honor their commitments. Capital flows into AI companies, which spend heavily on computing infrastructure, generating revenue for other companies that may themselves be backed by the same pool of investors. The arrangement helps finance rapid expansion, but it also creates dependencies that could become problematic if the flow of capital slows.
Anthropic’s revenue growth and mounting losses reveal a phenomenon rarely seen at this scale. A company founded only six years ago is on course to generate approximately $60 billion in revenue this year. That is a scale reached by established companies such as Unilever, pharmaceutical giant Eli Lilly, Cisco and Netflix only after years of building their businesses.
Yet the same figures that make Anthropic an extraordinary growth story also raise two central questions ahead of a potential IPO. Can the company sustain its expansion long enough to reach profitability? And can it justify a valuation of $2 trillion while continuing to incur such substantial losses?
The answers will depend partly on the broader market environment. Bond yields, Federal Reserve interest-rate decisions and the possibility of further escalation in the Middle East could all affect investor appetite for a high-priced technology offering, regardless of Anthropic’s own performance.
Assuming no major external shock, one of the most important reference points is the SpaceX offering led by Elon Musk last summer, which valued the company at $1.77 trillion. On the numbers alone, Anthropic has a strong argument for comparison. Despite its losses, it has already surpassed SpaceX in revenue and growth rate, even though Anthropic is only six years old.
SpaceX, for its part, was not profitable in the quarter preceding its offering. It returned to profitability in its most recent quarter, according to results released yesterday. That history offers a reminder that investors may be willing to accept substantial losses when they believe a company’s growth prospects justify an ambitious valuation.
There are also reasons to be cautious about drawing too direct a comparison. SpaceX’s valuation initially proved difficult for the market to absorb, with its shares trading below the offering price during the first few months. Following the latest results and discussions about a long-anticipated integration with Tesla, its valuation has since climbed above $2 trillion.
The doubts surrounding SpaceX’s offering were not unlike those that accompanied Google’s IPO two decades ago, when a valuation exceeding $20 billion was considered exceptionally ambitious. In both cases, the question was whether a company’s potential justified a price that seemed excessive by conventional standards.
Anthropic also has to contend with a rival benchmark. OpenAI, which has postponed its IPO for now, is working on a private funding round at a reported valuation of $1.4 trillion. That figure could shape how investors assess Anthropic, although a private financing valuation does not necessarily translate directly into the price public-market investors will be willing to pay.
Ultimately, Anthropic’s decision may come down to a familiar dilemma for companies preparing to go public. Investment bankers often advise companies to “leave some money on the table,” allowing new shareholders to benefit from an initial rise in the stock rather than trying to extract every last dollar at the offering.
If the year-end rally continues, an offering at $1.5 trillion that leaves room for the stock to rise could prove more attractive than insisting on $2 trillion and risking a flat or disappointing debut. For a company whose growth has already exceeded almost every conventional benchmark, the challenge may be less about proving its importance than convincing investors that its future profits will eventually justify the extraordinary cost of getting there.
