Anthropic has moved its planned initial public offering into the open, targeting an October Nasdaq listing that could value the Claude maker at $2 trillion or more and become the largest IPO in history.
According to reports from the Financial Times and Benzinga, the company had been expected to release its prospectus publicly last week but instead shared documents with a small group of investors first, fielding questions before making the filing public. The listing would make Anthropic eligible for the Nasdaq 100 index and would vault past SpaceX’s June debut, which valued Elon Musk’s rocket company at $1.75 trillion and was itself the largest public offering on record.
The numbers in the filing
Anthropic’s growth figures are the reason investors are treating a $2 trillion valuation as plausible rather than absurd. Second-quarter revenue rose roughly 14-fold year over year to $11.5 billion, and annualized revenue reached $65 billion by the end of July, up from $9 billion at the end of 2025. The company told investors it expects a second consecutive quarter of positive adjusted operating income, with gross margins above 80 percent before revenue shared with distribution partners and before the cost of training its models.
Analysts quoted by the FT project $120 billion in annualized revenue by the end of 2026 and close to triple that figure by the end of 2027. The company counts roughly 6,000 customers spending at least $100,000 a year. Those numbers explain why secondary-market transactions have already pushed implied valuations between $1.05 trillion and $1.5 trillion, well above the $965 billion post-money valuation from its $65 billion funding round in May.
The valuation arc is worth pausing on. Anthropic raised $30 billion in February at a $380 billion valuation. Three months later the May round nearly tripled that to $965 billion. A public listing at $2 trillion would mean the company multiplied its worth more than five times over in a single year, a pace that has no precedent at this scale in public markets.
| Metric | Value |
|---|---|
| Q2 2026 revenue | $11.5 billion, up about 14x year over year |
| Annualized revenue, July 2026 | $65 billion |
| Valuation, February round | $380 billion |
| Last private valuation (May) | $965 billion |
| Target IPO valuation | $2 trillion or more |
| Raise under discussion | up to $100 billion |
Nvidia as anchor investor
Reuters reported last week that Nvidia is in talks to invest as much as $10 billion as an anchor investor in the offering. The chipmaker’s participation would deepen an already close relationship, since Anthropic relies heavily on Nvidia GPUs for training and inference. It would also mirror the pattern from Anthropic’s private rounds, where Amazon and Google invested billions while simultaneously selling it compute, an arrangement critics describe as circular but which has become standard across the AI buildout.
The compute relationships are enormous in both directions. Anthropic has committed more than $100 billion over a decade to Amazon Web Services and uses over a million of Amazon’s Trainium2 chips, alongside multi-gigawatt TPU deals with Google and Broadcom. It recently signed a $35 billion cloud deal with Nvidia-backed Lambda. It is also building an in-house team to design custom chips for Claude, an effort to gain leverage over its hardware costs as spending scales.
The slowdown question hanging over the listing
The IPO arrives at a strange moment for the company’s public messaging. CEO Dario Amodei has spent the past week urging the industry to slow frontier model development, proposing independent evaluators, joint safety standards and international coordination. Jensen Huang of Nvidia publicly disagreed at Salesforce’s Dreamforce conference, saying the industry needs no new AI laws. OpenAI’s policy chief, meanwhile, confirmed the three labs have been quietly coordinating on safety standards for weeks, and Anthropic’s own policy head told CNBC that AI companies cannot be expected to operate on an honor code.
Investors quoted in the FT coverage are split on whether the slowdown talk hurts the offering. Menlo Ventures’ Matt Murphy, an Anthropic investor, described the company’s growth as off the charts and saw no reason for it to slow. Altimeter’s Brad Gerstner argued markets already know how to price this kind of risk, pointing to SpaceX’s debut as proof of appetite for AI leaders. Others are less charitable. One analyst called the safety push something that feels more like a ladder pull, competitive positioning dressed as caution.
There is also a product-side wrinkle. Data from Ramp’s corporate spending tracker shows Anthropic’s most expensive model, Fable 5, stalling at about 11 percent of sales despite being available for more than two months, while the cheaper Opus 5 overtook it in business usage shortly after its late-July debut. If enterprise buyers keep choosing cheaper models that get the job done, the industry’s core assumption, that customers always pay a premium for the smartest model, takes a hit, and so does the revenue curve the IPO is priced against.
Timing and competition
The company aims to complete the listing before the US midterm elections in November. Filing first gives Anthropic the narrative advantage over OpenAI, which is not expected to go public until at least 2027. Sam Altman said in a CNBC interview after the confidential filing became known that OpenAI would go public when it makes sense, not on a rival’s schedule. Analysts at IPO research firm IPOX noted that filing shortly after SpaceX lets Anthropic capitalize on strong investor appetite for AI and growth stocks while the window stays favorable.
The scale of the raise matters for markets beyond AI. A $100 billion offering would absorb demand that might otherwise spread across dozens of listings, and it would set the reference price for every private AI company watching from the sidelines. If the debut prices at $2 trillion and trades well, expect a queue of imitators. If it stumbles, the repricing will be just as fast, and the phrase AI bubble will move from op-eds to earnings calls.