Anthropic's S-1 Is Public: $16.3B Revenue, 900% Retention

Anthropic filed its S-1 prospectus with the SEC on Monday, reporting $16.3 billion in first-half revenue, an $80.7 billion net loss, and about 900% net revenue retention.

The company plans to list on Nasdaq under the ticker "ANT." It filed confidentially in June; this is the first public document detailing the finances, the compute commitments, the governance structure and the risk factors behind Claude.

Two loss numbers, and only one of them is operations

Revenue of $16.3 billion for the first half of 2026 is up from $1.2 billion in the same period last year. The headline $80.7 billion net loss sitting next to it is not cash burn from running the business. Per the filing, it was largely driven by noncash accounting adjustments tied to convertible debt — consistent with what the Wall Street Journal had reported from leaked versions of the document.

The operating figure is the one to read for unit economics: an operating loss of $693 million for the first half of 2026, improved from $2.1 billion in operating losses in the first half of 2025. The company still states plainly that it "may not achieve or sustain profitability in the future."

The compute bill is the business model

The filing lays out commitments that dwarf current revenue. Anthropic has agreed to pay about $3.1 billion per month to SpaceX for access to Nvidia GPUs once all planned capacity is delivered, totaling as much as $120.5 billion by 2029. Alongside that sit $44.9 billion with Microsoft through 2033, at least $114.5 billion with Google through 2033, and $110 billion with Amazon over the next ten years.

Those same cloud partners are also distribution. Under what the filing calls a "joint go-to-market approach," Anthropic holds the direct contractual relationship with customers and pays each cloud provider a fee, which the company says lets it "avoid channel conflict" and "set pricing." The filing also flags the obvious tension: the partners increasingly see Anthropic as a competitor, and Amazon, Google and Microsoft may prioritize their own requirements or rivals they have invested in — which could affect Anthropic's ability to secure compute from them.

Developer tools are the cross-sell engine

The enterprise section is where the filing is most specific about how adoption spreads inside a company. More than half of customers spending at least $100,000 a year now use both the Claude platform and a first-party app such as Claude Code or Cowork, up from about 30% at the end of last year.

As of September 30, the filing counts more than 11,900 customers spending at least $100,000 annually, more than 1,900 spending upwards of $1 million, and about 250 spending $10 million or more. Net revenue retention sat at about 900% as of the end of the third quarter, which Anthropic attributes to "significant usage expansion as customers realize the value of deploying Claude across their organizations."

The sharpest data point for anyone building on the API: in the third quarter, the highest-spending 5% of Claude Enterprise customers who used Claude Code spent roughly 20 times as much per month as the average enterprise user who did not use the coding tool. The filing's own framing is that a business starting with the API can deploy the coding tool to its engineers and then Cowork to the wider workforce, without finding a new buyer each time.

Anthropic pitches this as an "integrated approach" spanning models, platform and apps, in which "an advance at any layer compounds across all three layers." Read as a vendor argument rather than a measurement, that is a claim about lock-in — the ecosystem shape that made AWS hard to leave, which is the comparison TechCrunch draws in its reading of the filing.

Two distribution mechanics sit underneath it. An internal Applied AI team works with startups and enterprises to place Claude inside existing technology stacks, and a separate venture with Blackstone and other private equity firms, called Ode with Anthropic, embeds forward deployed engineers into businesses to build bespoke tools.

Seven founders keep control

The governance structure is unusual enough that it will likely outlast the revenue numbers as a talking point. Anthropic's seven co-founders — Dario Amodei, Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Christopher Olah — will form a "Founder LLC" that collectively controls 50.1% of shareholder voting power after the company goes public. The Long-Term Benefit Trust, established in 2023, will have authority to appoint most of the board.

Anthropic says the arrangement is designed to preserve its AI safety mission. The structural consequence is that voting power is not proportional to economic ownership, so public shareholders will have little practical control over a company that, by its own admission in the filing, is building technology posing "catastrophic or existential risks to humanity."

On pay, the filing puts Dario Amodei's 2025 compensation at nearly $18 million, most of it stock and option awards, with Daniela Amodei second at $16.4 million.

The risk factors are unusually blunt

Alongside standard risks, the prospectus lists items that do not appear in a typical technology S-1: the prospect of its models developing "forms of sentience," the possibility of "AI-specific taxes," and unfavorable resolutions to the copyright disputes over how its models are trained.

It also warns about "catastrophic" AI-related incidents and says it is likely that no single insurer — "and potentially the insurance industry in the aggregate" — could underwrite such risk. Customer concentration gets its own paragraph: losing one or more of the largest customers, if not offset elsewhere, "could adversely affect our business, financial condition, and results of operations."

In its own words, the company tells investors to expect behavior that differs from a purely profit-maximizing firm, and frames the pitch around four things: the potential of AI, the risks, its approach to building a company, and the ways its behavior may diverge.

What it changes for people building on Claude

Nothing here moves an API price today. Anthropic has not disclosed how much it hopes to raise or a proposed share price, so the valuation debate is still ahead.

What changes is the planning horizon. A company carrying monthly compute commitments at this scale, reporting 900% net revenue retention, and stating an intent to release "new classes of products whose aim is to dwarf the economic value of our existing ones" is explicitly telling builders it will devote less effort to entrenching current products and more to defining new ones. That is a poor surface for a thin wrapper and a good one for workloads that get cheaper as models improve.

Three things to watch over the next two quarters: whether pricing pages and deprecation notices tighten as public-market margin discipline arrives; whether the cloud partners' competitive tension shows up as compute constraints; and whether that 900% retention figure holds once the cross-sell from API to first-party apps has run through the existing customer base.