Anthropic is on track to exceed $100 billion in annualized revenue before the end of 2026, and it has quietly moved its initial public offering from October to November so that prospective investors can see a full third quarter of that growth before they are asked to price what could be the largest listing in history. Reports from The New York Times, The Wall Street Journal and Reuters on Friday and Saturday sketched a company whose revenue run rate has jumped from roughly $9 billion at the end of 2025 to more than $65 billion by the end of July, and whose bankers are now discussing a valuation of about $2 trillion and a raise of as much as $100 billion.
Anthropic closed its Series H in May at a $965 billion valuation, when its annualized revenue was about $47 billion. Two months later, that run rate had climbed past $65 billion, driven overwhelmingly by business customers adopting Claude for coding and other workplace tasks. Second-quarter revenue came in above $11.5 billion, roughly 2.5 times the first quarter and more than 14 times the $787 million booked a year earlier. The number of enterprises spending at least $100,000 a year on Anthropic software grew from about 1,500 at the end of 2025 to roughly 6,000 by the end of June, and more than 100 of them spent over $10 million each.
According to the Journal, advisers backed the later November window specifically so the company could put third-quarter results in front of investors. People familiar with the preparations said the timing was set before the recent public debate over whether AI is moving too fast, and that it could still shift. Reuters separately reported that the listing could slip past the November 3 midterm elections, though two people familiar with the matter said the election itself is not expected to have a major impact on the offering. Anthropic confidentially filed for its IPO on June 1 and, per the Times, could make its offering documents public within weeks.
The extra month is not only about showing a bigger revenue number. It is also about answering a competitive question that did not exist when Anthropic raised money in May. OpenAI released GPT-6 Astra on September 3, and the model has landed hard with enterprise buyers. Data from expense platform Ramp shows Astra accounting for about 13% of tracked enterprise AI spending, compared with about 8% for Anthropic's Claude Fable. On OpenRouter, developers spent more on OpenAI models than on Anthropic models last week, the first time that has happened in more than two and a half years. Reuters reported that three sources say Anthropic is now weighing whether to ship a new model of its own before the roadshow, with a safety evaluation of that model forming part of the deliberations.
That deliberation is awkward because of what Anthropic's chief executive said just days earlier. In a 3,800-word essay published September 12, Dario Amodei called on the entire industry to ease off the accelerator. “We must slow the pace at which we improve the capabilities of AI models,” he wrote, in a piece that drew public backing from Sam Altman and Elon Musk. Investors who will be pitched in the coming days are expected to ask directly how a self-imposed slowdown in product launches would affect a forecast that already calls for $190 billion to $200 billion in revenue by 2028.
Mike Volpi, a former partner at Andreessen Horowitz, said Anthropic could still emerge as a central player if the business models solidify, but was blunt about the shape of the market. “The AI market is not a cozy monopoly,” he said. Ramp co-founder Eric Glyman described how his company now rotates between models depending on the task, cutting its AI bill by 40%. “You don't need to rent a Ferrari to go grocery shopping,” Glyman said.
Anthropic declined to comment on the reports. Its cost base gives investors plenty to scrutinize alongside the revenue line: computing infrastructure spending rose 65% quarter over quarter to $5.6 billion, and a single agreement with SpaceX commits Anthropic to about $1.25 billion per month through May 2029 for capacity that includes more than 220,000 Nvidia GPUs.
Why It Matters
A $100 billion run rate, if it materializes, would put a five-year-old company on par with the largest enterprise software vendors in the world and make Anthropic the first pure-play frontier lab to test whether public markets will underwrite AI's capital intensity at scale. With OpenAI having ruled out a 2026 listing, Anthropic's November debut becomes the single reference price for the entire sector, the number every private AI valuation will be marked against.
The story also crystallizes the central tension of the moment. Anthropic built its brand on safety-first restraint and its CEO has now asked the industry to slow down, yet its revenue curve, its 2028 forecast and its ability to answer Astra all depend on continuing to ship. Open-weight models from China and from Meta are compressing prices at the same time that Meta, one of Anthropic's largest customers, is reportedly looking to reduce its reliance on Claude. The IPO will price not just Anthropic's growth but the market's belief that a frontier lab can hold pricing power against commoditization.
What to Watch
The next few weeks should bring Anthropic's public S-1, which will replace leaked run-rate figures with audited quarterly revenue, margins and the true scale of its compute commitments. Watch whether the company ships a new Claude model before the roadshow and how it squares that with Amodei's essay; whether the Ramp and OpenRouter share data keep tilting toward OpenAI through October; and whether the listing lands before or after the November 3 midterms. Any further slip past November would suggest investors are pushing back on either the $2 trillion figure or the growth story behind it.
"You don't need to rent a Ferrari to go grocery shopping."— Eric Glyman, Co-founder, Ramp