Anthropic’s IPO filing reveals explosive growth, huge compute bets and founder control

Anthropic has filed to go public, and its registration statement shows a company growing at breakneck speed while locking in extraordinary spending on compute and preserving unusually strong control for its founders. The S-1, filed with the SEC on Oct. 26, gives the first detailed public look at Anthropic’s finances, customer base, risks and governance as it prepares for a potential IPO.

What the filing says

The most striking figure in the document is revenue. According to TechCrunch, Anthropic brought in $16.3 billion in the first half of 2026, up from $1.2 billion in the first half of 2025. That scale of growth places the company among the fastest-rising names in frontier AI, at least by revenue trajectory.

The company also disclosed a broadening enterprise base. As of Sept. 30, Anthropic said it had more than 11,900 customers spending at least $100,000 annually, more than 1,900 spending at least $1 million annually, and about 250 spending $10 million or more annually, according to TechCrunch.

The usage data suggests the product mix is deepening, not just widening. More than half of customers spending $100,000 or more use both Claude and a first-party Anthropic app, up from about 30% at the end of 2025. TechCrunch also reported that the highest-spending 5% of enterprise Claude Code users spent about 20 times the monthly average of enterprise users without the coding tool.

Why it matters

Anthropic’s filing matters because it shows how expensive the race for advanced AI has become. Revenue growth is only one side of the ledger. The other side is compute: the company says it is committed to enormous infrastructure spending to secure access to Nvidia GPUs through partners including SpaceX, Microsoft, Google and Amazon.

Those commitments are not background details. They are central to the business model. Frontier AI companies need training and inference capacity, and the filings show Anthropic is trying to lock in supply years in advance. That may reassure investors that the company can keep serving customers, but it also raises questions about how much of that growth will turn into durable profit.

The filing also puts governance under a sharper spotlight. Anthropic is preparing for a public market debut while preserving founder influence through a structure that gives seven founders 50.1% of shareholder voting power through Founder LLC on key matters. That will likely appeal to management focused on long-term control, but it may draw scrutiny from investors accustomed to more conventional IPO structures.

How big are Anthropic’s customer numbers?

The customer base is large at the top end and appears to be getting stickier. The filing indicates that Anthropic has more than 11,900 customers spending at least $100,000 per year, and more than 1,900 spending at least $1 million annually. A smaller but important tier — around 250 customers — is spending at least $10 million annually.

That concentration at the enterprise end suggests Anthropic is not just chasing consumer attention. It is building a business around heavy users, especially companies willing to pay for access to Claude and, increasingly, Claude Code. The adoption data also hints that customers are expanding their use of Anthropic across workflows rather than treating it as a single-purpose tool.

What the compute commitments reveal

The filing’s most eye-catching business risk may be the company’s infrastructure obligations. The Verge reported that Anthropic has committed about $3.1 billion per month to SpaceX for Nvidia GPU access once capacity is delivered, a deal that could total as much as $120.5 billion by 2029.

That is not Anthropic’s only giant commitment. The Verge said the company has also committed $44.9 billion with Microsoft through 2033, at least $114.5 billion with Google through 2033, and $110 billion with Amazon over 10 years.

Taken together, those figures show a company trying to secure enough computing power to support current demand and future model development. They also show how dependent advanced AI businesses remain on a small set of infrastructure partners. Anthropic itself warns in the filing, according to TechCrunch, that cloud partners such as Amazon, Google and Microsoft may prioritize their own needs or rivals they support.

How should readers interpret those numbers?

The simplest reading is that Anthropic is betting its future on access to more compute than most companies could ever afford to buy outright. The more complicated reading is that those commitments create both strategic leverage and financial exposure. If demand stays strong, the contracts could support growth. If demand slows, the obligations could become a heavy burden.

Background: how we got here

Anthropic develops the Claude family of AI models and related enterprise tools. It has become one of the most visible rivals to OpenAI and Google in the race to build and sell advanced large language models, while presenting itself as a public benefit corporation that can pursue stated public-interest goals alongside shareholder returns.

The company’s route to the public markets has been watched closely because frontier AI firms face unusual capital intensity. Cloud and GPU capacity are not optional extras; they are core inputs. That makes fundraising, strategic partnerships and governance design unusually important. The S-1 now turns those private realities into public disclosures.

The filing also shows that Anthropic is trying to balance scale with product expansion. The company says it intends to spend less effort entrenching current products and more on defining new ones, a signal that management sees room to broaden beyond its current offerings even as enterprise adoption accelerates.

Founder control and the governance question

One of the clearest signals in the filing is that Anthropic wants to go public without surrendering control. The company said seven founders will control 50.1% of shareholder voting power through Founder LLC after the IPO on key matters, according to The Verge.

That arrangement is likely to protect the company’s long-term direction, especially in a field where product decisions can shape safety policy, model behavior and investment priorities for years. But it also means public shareholders may have limited influence over some of the most important decisions.

For investors, this is the core trade-off: a fast-growing AI company with deep enterprise traction, but one that is explicitly designed to preserve founder dominance. The filing gives the market a first chance to judge whether that is a feature or a flaw.

What the filing says about the Amodei siblings

The compensation disclosures add another layer of scrutiny. The Verge reported that Dario Amodei received nearly $18 million last year, mostly through stock and option awards, while Daniela Amodei received $16.4 million.

Anthropic also disclosed spending tied to personal security and travel. The Verge said the company spent $2.75 million on Dario Amodei’s personal security services, $1.29 million on Daniela Amodei’s protection, home security and transportation, and $97,055 for private-jet access for Dario Amodei’s personal use.

The company says its success is “highly dependent on the continued service” of senior leaders, especially Dario and Daniela Amodei, according to the filing. That kind of language is common in IPO risk disclosures, but here it is unusually direct given how much of Anthropic’s strategy appears to be anchored in a small leadership circle.

“highly dependent on the continued service”

What risks did Anthropic flag?

Anthropic’s risk section is unusually stark, according to The Verge. The company flagged concerns including catastrophic AI harm, possible sentience, and misaligned or autonomous model behavior.

Those disclosures are not unusual for a frontier AI company, but the phrasing matters. Anthropic is telling investors that the business is not only exposed to ordinary commercial risks such as competition and supply constraints, but also to outcomes that remain difficult to measure and even harder to regulate.

The company’s public risk framing is consistent with its public posture over the past several years: build advanced models, but acknowledge that advanced models can behave in ways that create serious safety, legal and reputational challenges.

“catastrophic”

How The Verge and TechCrunch framed the story

The two outlets emphasized different parts of the same filing. TechCrunch focused more on the commercial picture: revenue growth, customer counts, retention and usage patterns. The Verge centered on governance, executive compensation, security spending and the massive infrastructure commitments.

That difference is useful. Taken together, the reports show the full shape of Anthropic’s IPO story. On one side is a rapidly scaling enterprise software business with unusually strong expansion metrics. On the other is a capital-intensive AI lab wrapped in complicated control terms and enormous contractual obligations.

The combination is what makes the filing so revealing. Anthropic is not presenting itself as a conventional software startup anymore. It is presenting itself as an infrastructure-heavy AI company that must keep spending aggressively to hold position in a market dominated by a few giants.

By the numbers

What happens next

The biggest immediate unknown is timing. The filing does not disclose Anthropic’s IPO valuation or share price, and those details will shape how the market reads the company’s growth against its costs and governance structure.

Investors will also be watching how much of the reported revenue growth converts into profit or cash flow. Revenue alone does not answer whether the business can sustainably support the scale of its compute commitments.

Another open question is whether regulators and investors will focus on the founder control structure. Seven founders controlling a majority of voting power is not unprecedented in tech, but it is likely to attract attention at a time when public market buyers are asking tougher questions about accountability.

Finally, the market will be watching whether Anthropic’s dependence on cloud partners and scarce GPU access becomes an advantage or a constraint. The company has clearly secured massive capacity on paper. The challenge is turning those commitments into durable economics while staying ahead of competitors such as OpenAI and Google, both named directly in the filing as rivals.

If the filing is any guide, Anthropic’s public-market pitch will be straightforward: it is growing fast, selling deeply into the enterprise, and willing to spend heavily to stay in the race. Whether investors accept the trade-offs is the next story.