What’s the big deal
Salesforce took 20 years to reach $30 billion in annual revenue. Anthropic did it in under three years from its first dollar. The company was essentially pre-revenue in early 2024. By January 2025, it had reached $1 billion in annualised revenue. By the end of 2025, $9 billion. Then it tripled in four months: $14 billion in February 2026, $19 billion in March, $30 billion in April. That last stretch, $14 billion to $30 billion in roughly eight weeks, has no precedent in the history of enterprise software. As of April 2026, Anthropic’s revenue run rate exceeds OpenAI’s for the first time, driven by an enterprise-heavy revenue mix where 80% of income comes from business customers rather than consumers. Over 1,000 companies now spend more than $1 million a year on Claude, a figure that doubled in under two months. By mid-May, run-rate revenue had crossed $47 billion, a fivefold increase from six months earlier, and on 28 May Anthropic closed a $65 billion funding round at a $965 billion valuation, becoming the world’s most valuable AI start-up before filing for an IPO on 1 June. Revenue kept climbing after the raise: $65 billion by the end of July and more than $100 billion by mid-September, more than ten times its pace at the end of 2025. But the growth comes with context: Anthropic reports revenue from cloud resellers on a gross basis, which inflates top-line figures relative to net-reporting peers, and it lost more than $8 billion on an operating basis in 2025.
What it is
This page tracks Anthropic’s financial trajectory: revenue milestones, funding rounds, valuation history, enterprise adoption metrics, and the path to a potential IPO.
Revenue timeline
January 2024 — Approximately $87 million annualised run rate.
January 2025 — $1 billion annualised. Less than two years after Claude’s March 2023 launch.
May 2025 — Claude Code reaches general availability, becomes a major revenue driver.
August 2025 — $5 billion annualised. Anthropic is now one of the fastest-growing technology companies in history.
October 2025 — Claude Code crosses $500 million in annualised revenue with 10x usage growth in three months. 300,000+ business customers.
December 2025 — $9 billion annualised. Claude Code crosses $1 billion in annualised revenue roughly six months after GA.
February 2026 — $14 billion annualised (disclosed at Series G). Claude Code at $2.5 billion, more than doubling since the start of 2026. Over 500 companies spending $1 million+ annually.
March 2026 — $19 billion annualised (Bloomberg).
April 2026 — $30 billion annualised (Bloomberg, confirmed by Anthropic). Over 1,000 companies spending $1 million+ annually, doubling from 500 in under two months. Revenue surpasses OpenAI’s approximately $24 to $25 billion for the first time.
May 2026 — $47 billion annualised, confirmed by Anthropic at the Series H close. Revenue has grown roughly fivefold from $9 billion at the end of 2025. Gross margins reported to have improved to over 70%.
July 2026 — $65 billion annualised, a sevenfold increase from the end of 2025.
September 2026 — More than $100 billion annualised (New York Times), more than ten times the end-of-2025 pace.
Where the money comes from
Approximately 80% of Anthropic’s revenue comes from business customers, with the remainder from consumer subscriptions (Pro, Max). The enterprise revenue mix breaks down across three channels: direct API contracts where companies build Claude into their own products, cloud reseller agreements through AWS Bedrock, Google Vertex AI, and Microsoft Azure Foundry, and subscription plans for teams and enterprises.
Claude Code is the single largest product revenue driver, generating over $2.5 billion in annualised revenue as of February 2026. Enterprise use now represents over half of all Claude Code revenue. Business subscriptions to Claude Code have quadrupled since the start of 2026.
The number of customers spending over $100,000 annually has grown 7x in the past year. Eight of the Fortune 10 are Claude customers. According to Ramp, a corporate expense platform tracking real spending across 50,000+ businesses, 43.8% of US businesses with paid AI subscriptions now pay for Anthropic as of August 2026, up from 16% a year earlier and ahead of OpenAI’s 39.8%. Anthropic overtook OpenAI on this measure for the first time in April and is now the model new AI buyers most often choose.
An important caveat on how to read the numbers. Anthropic reports revenue from cloud resellers on a gross basis, counting total end-customer spend as revenue and booking partner payouts as expenses. This inflates top-line figures relative to companies that report on a net basis. The revenue figures are real, but direct comparisons to net-reporting peers require adjustment. Its IPO prospectus shows 47% of 2025 revenue came through the Amazon and Google cloud marketplaces.
Funding rounds
Series H — May 28, 2026. $65 billion at $965 billion post-money valuation, the second-largest private funding round on record, after OpenAI’s $122 billion raise weeks earlier. Co-led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, with Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN co-leading. Institutional participants included Baillie Gifford, Blackstone, Brookfield, D.E. Shaw Ventures, DST Global, Fidelity, Lightspeed, T. Rowe Price, and Temasek. Strategic memory and compute suppliers Micron, Samsung, and SK Hynix joined, a sign the round doubled as supply-chain alignment for the hardware Anthropic needs to scale. Roughly $15 billion was previously committed hyperscaler money, including $5 billion from Amazon announced in April. The round made Anthropic the world’s most valuable AI start-up, surpassing OpenAI’s $852 billion, and is expected to be its last private raise before going public.
Series G — February 12, 2026. $30 billion at $380 billion post-money valuation. Led by GIC and Coatue. Co-led by D.E. Shaw Ventures, Dragoneer, Founders Fund, ICONIQ, and MGX. Microsoft and Nvidia contributed a portion of the up to $15 billion they had committed in November 2025. Anthropic initially sought $10 billion but increased the target after attracting demand several times larger. A tender offer at the same valuation allowed employees to sell shares.
Series F — September 2, 2025. $13 billion at $183 billion post-money valuation. Led by ICONIQ, co-led by Fidelity and Lightspeed.
Series E — March 3, 2025. $3.5 billion at $61.5 billion post-money valuation. Led by Lightspeed.
Earlier rounds — Amazon invested $8 billion across multiple rounds before 2026, and Google more than $3 billion, including $1 billion in January 2025. In April 2026 both went further: Amazon invested $5 billion with up to $20 billion more tied to commercial milestones, and Google $10 billion with up to $30 billion more tied to performance milestones. Total investors to date: 90+, including 81+ institutional.
The path to profitability and IPO
Anthropic’s confidential IPO prospectus, reported by Reuters in September, shows 2025 revenue of nearly $4.6 billion, up twelvefold, against an operating loss of more than $8 billion. The headline net loss of about $42 billion includes a roughly $34 billion non-cash charge from revaluing financing that could convert into shares. The picture has since shifted: second-quarter 2026 revenue topped $11.5 billion, up from $4.73 billion in the first quarter, and Anthropic reported positive adjusted operating income (a measure that excludes costs such as stock-based compensation) for the first time. It is projecting 2028 revenue of roughly $190 billion to $200 billion. For comparison, OpenAI projects approximately $14 billion in losses for 2026 and does not expect positive free cash flow until 2029 to 2030.
Anthropic’s IPO prospectus puts its infrastructure commitments at $518 billion or more over the next decade with six partners, about 80% of it non-cancellable, led by $111.1 billion with Google, $110 billion with Amazon, and about $161.2 billion of Broadcom-related chip-lease obligations. Broadcom has also agreed to lend Anthropic up to $42 billion to help finance its TPU leases. The April 2026 compute deal with Google and Broadcom secures approximately 5 gigawatts of TPU-based capacity starting in 2027. In May, Anthropic leased 100% of SpaceX’s Colossus 1 data centre, adding over 220,000 Nvidia GPUs immediately. In July, it added a 20-year, $19 billion TeraWulf data-centre lease (401 megawatts in Hawesville, Kentucky, online from H2 2027) and an AMD partnership pairing up to $5 billion of AMD equity investment with up to 2 gigawatts of Instinct MI450 GPU capacity from 2027. In August, it added a six-year, roughly $45 billion deal with neocloud Nscale for capacity at its West Virginia campus, and a $35 billion cloud deal with Nvidia-backed Lambda drawing on Nvidia GPU capacity at Hut 8’s Texas data centre; Nscale has since confirmed its side in its own IPO filing (four agreements worth about $44.6 billion), while the Lambda deal rests on reporting by Bloomberg, Reuters, and the Wall Street Journal. Broadcom’s SEC filing noted the scale depends on Anthropic’s “continued commercial success.”
Anthropic filed a confidential draft S-1 with the SEC on 1 June 2026, the first frontier AI lab to do so, and has since chosen Nasdaq, with Morgan Stanley, Goldman Sachs, and JPMorgan leading the offering. Prospective investors put a fair valuation at $1.8 trillion to $2 trillion, roughly double the Series H mark, and Anthropic expects to match or beat SpaceX’s record $75 billion raise. Formal marketing could start as soon as the week of 9 November, with trading before the Thanksgiving holiday, after an earlier mid-October target slipped; the timeline could still shift. The Series H is expected to be its last private raise before going public.
How it compares to OpenAI
OpenAI’s annualised revenue was approximately $24 to $25 billion as of March 2026 and was nearing $70 billion by late September, per Axios, against Anthropic’s more than $100 billion. OpenAI’s last completed round valued it at $852 billion; Anthropic’s closed Series H at $965 billion surpassed it for the first time.
OpenAI announced $2 billion in monthly revenue and a $122 billion funding raise at an $852 billion valuation in the same week as Anthropic’s $30 billion disclosure. OpenAI has approximately 900 million weekly active users, with free users accounting for roughly 95%, meaning massive inference costs are not covered by revenue. By August, enterprise had overtaken consumer as OpenAI’s larger revenue line, ahead of its own end-of-2026 forecast, CFO Sarah Friar told investors.
The structural difference has been revenue composition. Anthropic’s 80% enterprise mix produces higher retention, lower churn, and contracts that expand over time. OpenAI built on consumers, which brings more users but lower revenue per user and higher churn, and only tipped towards enterprise this year. Anthropic reached $30 billion faster than OpenAI despite having a fraction of the user base, because enterprise contracts are larger and more durable.
Both companies are burning cash at enormous rates. OpenAI has committed over $1 trillion to infrastructure over the next several years and projects a $14 billion loss in 2026. Anthropic’s burn is proportionally large. The investors funding both companies are making a specific bet: that compute costs continue to fall per unit of intelligence, that revenue keeps compounding faster than burn, and that whoever owns the AI infrastructure layer in 2029 will generate returns that make the interim losses irrelevant.
Risks
Pentagon supply-chain risk designation, split rulings. The Pentagon declared Anthropic a supply-chain risk in February 2026, after Anthropic’s usage-policy restrictions cost it a $200 million defence contract, and the administration moved to cut it off from federal work. Anthropic challenged the designation in two courts. In August, a federal judge in California ruled the broader ban unlawful retaliation and vacated it. In September, a DC Circuit appeals panel voted 2-1 to uphold the Defense Department’s own designation under a separate supply-chain law, so Claude remains barred from Pentagon systems, including for contractors’ defence work. Anthropic is considering further review.
FTC investigation. The Federal Trade Commission confirmed on 30 September that it is investigating Anthropic, OpenAI, and other AI developers over risks their products may pose to consumers, and is preparing civil investigative demands for documents and executive testimony.
Copyright litigation. Anthropic agreed to pay $1.5 billion to settle the Bartz authors’ class action over pirated books it downloaded, approved in July 2026 as the largest copyright settlement in US history. It still faces suits from music publishers over song lyrics, from Universal, Concord, and ABKCO and, since August 2026, from Sony Music Publishing and Warner Chappell.
Gross vs net revenue reporting. The revenue figures are reported on a gross basis. Cloud reseller payouts to AWS, Google, and Microsoft are booked as expenses rather than deducted from revenue. This makes direct comparisons to net-reporting companies misleading without adjustment.
Compute cost dependency. The business model depends on compute costs continuing to fall. If they plateau or rise, the margin structure changes fundamentally. The infrastructure commitments, most of them payable regardless of use, are a bet on continued cost deflation.
Cash burn. Despite more than $100 billion in run-rate revenue, Anthropic lost more than $8 billion on an operating basis in 2025, and its first positive quarter is on an adjusted basis that excludes costs such as stock-based compensation. Sustained profit depends on revenue continuing to outgrow compute costs, which is not guaranteed.
Model suspension and regulatory risk. On 12 June, a US government export-control directive forced Anthropic to disable its two most capable models, Fable 5 and Mythos 5, worldwide, the first time a government has pulled a frontier model from the market. The controls were lifted on 30 June after Anthropic agreed to new safeguards and closer government collaboration, and Fable 5 returned on 1 July. But the episode, landing weeks after the S-1 filing, underlines how exposed Anthropic’s frontier roadmap is to regulatory action, and it set a precedent (echoed by OpenAI’s government-gated GPT-5.6 preview) that frontier launches may increasingly require national-security review. That uncertainty now sits in front of prospective IPO investors.
Valuation risk. The targeted IPO valuation of up to $2 trillion is roughly 20 times run-rate revenue. If growth decelerates or the IPO window narrows, the gap between private valuation and public market appetite could become significant. Some early backers chose to wait for the IPO rather than join the Series H.
Availability
Anthropic is a private company, though it filed for an IPO on 1 June 2026 and a Nasdaq listing is targeted for November. There is no way to buy Anthropic stock directly on public markets as of October 2026. Limited pre-IPO exposure exists through secondary market platforms such as Hiive and Forge Global, restricted to accredited investors, where implied valuations have already approached $1 trillion.