Four takeaways from Anthropic’s potentially record-breaking stock market filing

The 92-page document released on Monday talks about Anthropic’s explosive growth and the risks AI has brought to the business. Photograph: Kim Kyung-Hoon/Reuters View image in fullscreen The 92-page document released on Monday talks about Anthropic’s explosive growth and the risks AI has brought to the business. Photograph: Kim Kyung-Hoon/Reuters Anthropic Explainer Four takeaways from Anthropic’s potentially record-breaking stock market filing The AI company’s first public financial disclosures reveal explosive growth, huge costs and risks ranging from cyber-attacks to the limits of safeguards

Prefer the Guardian on Google Anthropic has filed to go public on the US stock market, revealing for the first time the AI research lab’s financials and all the risks – more than 90 pages of them – that might negatively impact its business.

The company’s investor prospectus, released Monday, veers from hyperbolic predictions about AI’s “incredible potential” to existential risks of the technology to the reality of its current business: explosive but incredibly expensive growth. The filing to US financial regulators, known as an S-1, provides the public’s first look at an audited version of Anthropic’s finances.

Here are some of the key takeaways from Anthropic’s lengthy S-1.

Read more 1. Explosive and expensive growth Here’s what you need to know: Anthropic is making a lot of money. It’s also losing a lot of it. The five-year-old firm has been valued between $1.5 and $2tn by some investors but did not list a target valuation in its S-1. Founded in 2021, the company has seen explosive growth every year of its existence.

Anthropic brought in $16.3bn in revenue and lost $80 bn in just the first half of 2026, though much of that was due to a $72.4bn one-off charge. In 2025, the company brought in $4.585bn and lost roughly $33.9bn, per the filing.

Securing computing power accounts for the vast majority of Anthropic’s costs. The company spent nearly $9bn on AI infrastructure in the first half of this year, up from $2.3bn in the same period last year.

Anthropic, unlike OpenAI, has made it clear that its customers are largely other businesses. “Frontier AI is most valuable in real-world settings, as deployment across enterprises is where economic impact is greatest,” the company executives wrote in the filing.

The company said its top 5 customers account for 25% of its business – which makes it risky business.

Anthropic says there is considerable potential for AI to reach levels of intelligence that would be akin to what its executives called “‘a country of geniuses in a datacenter’ – millions of virtual agents with the ability to revolutionize every area of human endeavor: software, physics, math, biology, medicine, finance, economics, commerce, robotics, military and intelligence capabilities”. Noticeably, the company mentioned military capabilities despite being embroiled in a standoff with the Trump administration over its refusal to limit restrictions on the way the Department of Defense could use Claude.

The company highlighted four specific categories of opportunity at the start of the filing: Acceleration of progress in biology and medicine; economic growth; spreading said growth globally; and streamlining government services (such as social services and medical programming).

The company said it’s already working with several pharmaceutical and biotechnology companies and improved the process for clinical trials, which is “cutting the time it takes to write clinical trial reports from months to minutes”.

The company’s executives claim they’ve also made biological discoveries with Claude, including a “molecular machine” that has the “potential to be developed into a novel biotechnology tool”. One purported use case: gene editing.

The company discusses risks in two contexts: the dangers that AI itself poses to society and human beings, and the ways Anthropic’s business could go awry.

Risks that AI poses were broken down into four main categories: misuse by bad actors, loss of control, authoritarian abuse and economic disruption. While Anthropic put a great deal of emphasis on the guardrails it has implemented to block attempts to misuse its models, it also includes one big caveat later in the filing: those safeguards may not work.

Anthropic’s most capable models are capable of providing “meaningful assistance to malicious actors seeking to develop chemical, biological, or other dangerous weapons”, the company’s executives wrote in the filing.

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 fail to operate as intended”.

Already, Anthropic has detected what it believes is the first documented case of a large-scale cyber-attack with little human intervention. The company believes a Chinese state-sponsored hacking group manipulated a Claude code tool “into attempting infiltration into roughly 30 global targets, including large technology companies, financial institutions, chemical manufacturing companies, and government agencies”, per the prospectus.

The company said it recognizes AI may be used by authoritarian governments to “deepen repression” and develop “high-tech surveillance states”. Anthropic presented less of a defense or potential safeguard against those uses and instead said it was committed to working with “free societies to provide a counterweight to authoritarian abuse and to promote freedom and democracy”. The firm seems to suggest that its work with what it describes with scant details as “free societies” will cancel out in some way the AI-enabled repression of authoritarian regimes.

The company had a falling out with the Trump administration after the firm said it would not change its restrictions to allow its models to be used to build domestic surveillance or autonomous lethal weapons.

Anthropic goes into detail in laying out 92 pages of risks to its business. Investors are making sense of a type of company that is fairly new to the public market in a highly competitive space with everything to lose.

The company acknowledges that much of its success is tied to its top executives and whether they stay at the firm, being able to eke out its main competitors Google and OpenAI, navigating the very tumultuous regulatory landscape around AI and datacenters and surviving any public relations fallout over its AI models being used to do things they are not supposed to do, like create bioweapons.

The company’s focus on enterprise customers means that if any of its top buyers leave, the company could take a huge hit.

Anthropic’s structure as a public benefit corporation – which is a designation for a for-profit company that balances financial interests with achieving a stated social good – may make it harder to achieve its short-or medium-term financial goals, the company said. The firm is obliged to report on its progress meeting its public benefit goals in the same way they have to report its financials.

There’s also all the risk related to any AI-enabled catastrophic event – the sort of doomsday scenario Anthropic employees have been warning about. Anthropic CEO Dario Amodei has called for a pause on AI development to ensure firms are developing it responsibly to avoid any such catastrophe. In addition to the public harm and potential fallout over a scenario where Claude went rogue, Anthropic said they worry they’ll be subject to a great deal of liability, especially considering insurance options for AI are still evolving.

“We may be unable to obtain coverage for certain AI-related liabilities on commercially reasonable terms, or at all, leaving us exposed to potentially significant uninsured losses,” the filing reads.

The bedrock of Anthropic’s AI infrastructure is concentrated in the US, making it extremely vulnerable to regulatory changes, power grid constraints, natural disasters and opposition to AI or datacenters. The filing specifically calls out data center moratoriums that have been passed or are being considered as among the big issues they’re concerned about.

“We are expanding our data center capacity internationally as part of our efforts to diversify our infrastructure footprint and reduce concentration risks,” the filing reads.

The company, which says a big part of its value proposition is its talent, also said it worries about US immigration policies affecting its ability to recruit and retain highly qualified employees. The Trump administration has targeted H-1B visas, a tool tech firms have come to rely on to attract and hire from the global pool of tech and engineering talent.