Anthropic is ready for Wall Street.
The company behind Claude filed its public S-1 paperwork on Monday, revealing its plans to list on the Nasdaq under the ticker symbol ANTand setting the stage for an enormous IPO.
Anthropic’s filing pulls back the curtain on the financials that have made it one of the fastest-growing startups in history; the company was valued at $965 billion in May.
“We believe that AI is one of the most consequential technologies in human history,” Anthropic said in its filing. The company said the technology could accelerate scientific innovation, increase economic growth, and cure diseases. “Anthropic aims to realize these benefits while navigating and preventing the risks that come with a technological revolution this fast and this powerful.”
The AI giant disclosed a total net loss of about $42 billion in 2025, with revenue of $4.6 billion. In the first six months of this year ending June 30, Anthropic recorded a staggering net loss of $80.7 billion — nearly double its loss for all of last year. Anthropic attributed much of that loss to an accounting adjustment.
Anthropic’s 5 biggest customers drive ~25% of its revenue — and are spending more than ever
As recently as 2025, Anthropic lagged well behind rival OpenAI. It generated $1.16 billion in revenue over that year’s first six months. Since then, for month after month, Anthropic has won over business customers and bolstered its top line.
From January through June 30 of this year, Anthropic raked in $16.3 billion in revenue. The San Francisco AI lab told investors in mid-October that its third-quarter revenue totaled $16.8 billion, a jump from earlier in the year.
Increased spending from Anthropic’s five biggest customers — who drove around 25% of its overall revenue in the first nine months of the year — is fueling that growth. In the three-month period ending September 30, Anthropic said revenue from its five biggest spenders reached a record high. The company said its net revenue retention rate, which the company uses to measure its ability to retain and extract more revenue from its paying customers, reached nearly 900%.
But the fact that such a large portion of revenue is driven from such a small number of customers could prove to be a double-edged sword, Anthropic said, as they could choose to switch AI providers. Anthropic also cautioned that it expects its revenue growth rate to fluctuate and “decline over time.”
The firm’s growth has stunned the industry and added to a frenzy of excitement over AI’s economic impact. CEO Dario Amodei quipped in May that the growing demand for his products was “too hard to handle”; the company has raced to secure more computing power to serve customers with its chatbot, coding, and workplace tools.
Those commitments are already titanic. Anthropic expects to spend at least $265.6 billion on cloud, computing, and infrastructure over the coming years on fixed or minimum-guaranteed commitments through June 30, according to the filing. The company also has additional compute agreements that could push the total cost up much higher.
Everyday investors — including you — can now see under the hood. Companies going public in the US are required to file an S-1, which includes a company’s vision for its future, recent financial results, major shareholders, and the risks it faces.
Anthropic’s grim risk factors
In its 90-plus-page risk factors — by far the largest section of the S-1 — Anthropic warned that a potential “catastrophic AI-related incident” could be so damaging that the “insurance industry in the aggregate” may not be able to cover it.
“Harmful incidents involving our models, regardless of fault, could also severely and disproportionately damage our reputation and brand as a safety-focused AI company, leading to loss of customer trust and customer attrition, reduced demand for our models, platform, and applications, and difficulty attracting and retaining talent,” Anthropic said.
The company also warned that it may not be able to detect or prevent every harmful use or output from its AI models, leaving the firm vulnerable to reputational damage, regulatory scrutiny, and costly legal claims.
Anthropic said people at risk of suicide or self-harm, as well as people who have committed violence against others, have used its models despite its safeguards.
“AI models developed by others in our industry have been and may in the future be alleged to have played a role in instances of suicide, acts of violence toward others, or serious self-harm, which have resulted and may in the future result in significant public scrutiny, litigation, criminal investigations, regulatory attention, and reputational harm,” the company said.
CEO Dario Amodei’s $18 million pay package included $2.7 million for personal security
Anthropic CEO Dario Amodei received total compensation valued at nearly $18 million in 2025, while president Daniela Amodei — his sister — received about $16.4 million, per the filing. The largest parts of their packages were stock and options, with grant-date equity values of about $13.3 million each.
The filing lists a 2025 salary of $654,167 for each executive. Their annual base salaries rose to $915,000 in January 2026 and then to $1.38 million in July.
Dario Amodei’s package also included about $2.7 million in personal security costs and $97,055 for personal air travel under his security program. Daniela Amodei’s package included about $1.3 million in security and transportation costs.
Krishna Rao, Anthropic’s CFO, received $706,250 in salary in 2025. His annual base salary rose to $1.25 million in February 2026.
Read more Apollo says AI has delivered ‘zero financial payoff’ outside the tech industry
Anthropic wrote that its executive packages reflect its “overall compensation philosophy,” including that it “targets the 90th percentile of the AI development market.”
Compute deals total hundreds of billions
All told, Anthropic disclosed about $637 billion in compute and infrastructure spend with named companies in the coming years. Other deals, including with AMD, do not specify a fixed amount. Anthropic cautioned that the exact amount may vary depending on the amount of compute delivered, while other deals could push the figure higher. The company also left blank several compute-related figures in its publicly released S-1.
Of that staggering amount, the largest individual arrangement is with Broadcom. Anthropic is set to pay Broadcom roughly $161.2 billion under a five-year master lease to deploy its chips in data centers. The figure does not include interest payments.
In addition to Broadcom, Anthropic outlined its deals with Microsoft, Google, SpaceX, Amazon, Nscale, and Lambda. According to the S-1, Anthropic entered some of these arrangements as recently as this month. Anthropic said it originally entered a deal with Microsoft in November 2025, but modified it in February and again this month. Under that arrangement, Anthropic is set to spend at least $44.9 billion between November 2026 and May 2033 “regardless of actual usage.”
The Claude maker’s deal with Google could balloon even higher in what Anthropic described as part of existing deals to secure Google’s Tensor Processing Units, the tech giant’s custom-designed AI chips, to support additional compute between 2028 and 2031. Anthropic estimates those additional payments at between $125 billion and $150 billion, though they would not go exclusively to Google.
Anthropic’s deals with Google, SpaceX, and Microsoft are particularly notable as they underline the degree of interdependence among AI and tech companies even as all three push to develop the most cutting-edge AI models.
Anthropic’s 7 cofounders have voting control
Anthropic’s seven cofounders will retain majority voting control of the company after it goes public. Anthropic will issue one share of Class F common stock to an LLC composed of the seven cofounders, known as the “Founder LLC.”
That single share will carry 50.1% of shareholder votes, meaning the seven cofounders will have more voting power than all other shareholders combined. Should any of the cofounders die or become unable to serve in their role, a successor may be admitted to the LLC until 10 years after the public offering is completed.
“We believe that our success depends in part on establishing a capital structure that enables our co-founders, management, and board of directors to focus on strategic initiatives that create lasting value for our stakeholders, rather than facing pressure to prioritize short-term results,” the filing reads under the section on voting structure.
Each of the LLC members, referred to in the filing as “Continuing Founders,” will get one vote to determine how the Class F common stock will vote. If a majority of the founders can’t agree, the Class F stock will “generally be voted proportionately with the votes of our Class A common stock cast on the matter.”
A fierce rivalry and race toward IPO
A group of OpenAI alums launched Anthropic in 2021 as a public benefit corporation, meaning the company must balance profits with its mission of “developing and maintaining advanced AI for the long-term benefit of humanity.”
The world’s gaze turned toward generative AI companies and tools when OpenAI launched ChatGPT in late 2022. Anthropic quickly followed with its chatbot Claude the following March.
Since then, a rivalry has developed between the two companies over top AI research talent, the best model releases, and customers. The virality of ChatGPT had handed OpenAI a lead.
In 2025, after Anthropic released Claude Code, the race began a new phase. AI’s coding skills improved, and business customers began pouring money into the technology. Each company’s valuation boomed, and this spring, Anthropic’s eclipsed OpenAI’s.
Despite its staggering growth, the narrative that Anthropic has to provide investors to drum up excitement for its IPO is tricky. The AI industry is still in its early days and faces a backlash over data center construction. The technology’s economic implications are still unknown. Anthropic has had a topsy-turvy relationship with the Trump administration. Amodei joined the president for a private dinner in late September after high-profile disagreements.
Plus, Anthropic is contending with cheaper models from Chinese firms, which some developers are using to replace American tools. Leading AI providers are also reckoning with security incidents in which their AI agents have breached outside systems.
Unlike OpenAI, Anthropic is forging ahead with a 2026 IPO as concerns about safety hang over the industry. Amodei has said that frontier AI labs like Anthropic should slow the pace of AI development so that researchers can focus more on safety and alignment.
Anthropic, the first pure-play AI company to publicly file its S-1 in the US, can set expectations with its early stock market performance.
Have a tip? Contact this reporter via email at [email protected], or over text, Signal, Telegram, or WhatsApp at 415-757-8198. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.
About the Author
