Anthropic Files for IPO at Over $1.5tn Valuation, $80bn Loss in Tow

Anthropic has publicly filed for a US stock market listing at a valuation analysts and investors put at more than $1.5tn, positioning the debut as one of the largest in history and a stress test of the multitrillion-dollar AI boom.

The San Francisco company, founded in 2021, confidentially filed with US regulators for an initial public offering in June and will list under the ticker ANT. The float will serve as the first real gauge of whether the AI sector's private valuations survive contact with public markets.

"We believe AI is one of the most consequential technologies in human history," the company's investor prospectus reads. The same executives who wrote that line have spent months warning that AI could cause human extinction — a tension the filing itself preserves. CEO Dario Amodei, who founded Anthropic after quitting OpenAI, has publicly called for a slowdown of AI development, citing existential risk. A successful listing may do the opposite: elicit a rush of investment into the company and the AI sector more broadly, accelerating the industry's pace.

The filing is the first audited, public look at Anthropic's financial health, which has been the subject of intense speculation amid talk of an AI bubble. The numbers show a company growing revenue faster than almost any in corporate history — and losing money at a scale to match.

What do the revenue figures show?

The prospectus details a steep revenue curve:

• $41m in 2023
• $386m in 2024
• $4.585bn in 2025 — an increase of more than 1,000% year on year
• $16.286bn in the first six months of 2026 alone

For perspective: the first half of 2026 generated roughly 3.5 times the entire 2025 figure, and nearly 400 times the company's full-year 2023 revenue. Anthropic disclosed that, as of the end of June, it had raised $141.1bn in total.

Yet growth has not bought profitability. Anthropic reported a net loss of $80bn for the first half of 2026 — meaning the company lost roughly $4.90 for every $1 of revenue it booked over that period. It has also incurred $6.2bn in debt to build out infrastructure, and spends billions per year on computing power from third-party providers. In plain terms: the revenue is real, but so is the burn.

Who are Anthropic's customers?

Unlike OpenAI, which has courted everyday consumers for ChatGPT, Anthropic has focused on businesses as its primary paying customers. Its flagship product is the Claude chatbot, which offers text generation alongside advanced coding and cybersecurity capabilities.

As of 30 September, Anthropic reported more than 14,000 enterprise customers. Of those, 11,900 generate $100,000 or more in revenue, more than 1,900 generate $1m or more, and about 250 generate $10m or more.

"The speed and depth of our enterprise customer adoption are powerful evidence of Claude's position at the frontier of AI," Anthropic executives wrote. The claim carries a caveat the company itself discloses: a quarter of revenue comes from its top five customers, meaning a handful of defections could put a large dent in the top line. No single customer, however, represents more than 10% of business.

What risks did the company disclose?

The prospectus contains more than 90 pages of potential risks to the business. Concentration is one. The rest read like a catalogue of the AI industry's unresolved questions.

Executives identified four categories of AI risk: misuse by bad actors, loss of control, authoritarian abuse and economic disruption. Anthropic's most capable models have "capabilities that could provide meaningful assistance to malicious actors seeking to develop chemical, biological, or other dangerous weapons," they wrote. The company deploys technical safeguards, "but such safeguards are incomplete, may be outdated in light of evolving threat models, have been and may in the future again be circumvented, or may in the future fail to operate as intended."

The filing also discloses a previously reported security episode. In September 2025, Anthropic staffers documented what they believe was the first large-scale cyber-attack that did not rely on significant human intervention. Anthropic "detected and assessed with high confidence that a Chinese state-sponsored hacking group manipulated our Claude Code tool into attempting infiltration into roughly 30 global targets, including large technology companies, financial institutions, chemical manufacturing companies, and government agencies."

Can Anthropic sell Washington?

The Trump administration presents its own risk. Anthropic has been embroiled in a feud with the administration over the use of its products in warfare. That dispute culminated in the company's designation as a supply chain risk — a first for an American company — which prevents any military agency from contracting with Anthropic. The company has challenged the blacklisting.

The filing also carries the hyperbolic language characteristic of AI prospectuses. Describing AI agents, semi-autonomous chatbots that carry out tasks on users' behalf, Anthropic executives wrote: "The result is something without historical precedent: what we refer to as a country of geniuses in a datacenter, available to every organization and sector of the economy."

Whether public investors accept that framing — and price in an $80bn half-year loss against it — will shape how the rest of the AI sector finances itself.