Anthropic reveals dependence on Google, Amazon, and Microsoft ahead of IPO

Anthropic is increasingly dependent on Google, Amazon, and Microsoft: sales through their cloud platforms accounted for almost half of the Claude developer’s revenue in the first half of 2026. Meanwhile, the company pays its cloud partners for distribution and has committed to spending hundreds of billions of dollars on their computing capacity.

For Anthropic, the three largest cloud providers represent its investors, computing suppliers, sales channels, and competitors all at once. The company explicitly identifies this combination of roles as a source of risk in its prospectus.

Sales through Google, Amazon, and Microsoft’s cloud marketplaces accounted for 48% of Anthropic’s revenue in the first half of 2026, versus 11% in 2023, according to the company’s prospectus, published on Monday. The cloud providers bill Anthropic’s customers and pass along their payments after deducting fees. In the first half, fees paid to channel partners, a group that includes the cloud providers, rose $2.7 billion year over year.

Cloud sales have helped fuel Anthropic’s rapid growth, Bloomberg notes. Second-quarter revenue reached $11.6 billion, while operating profit totaled $974 million. At the same time, Anthropic disclosed massive computing purchase commitments in its prospectus: at least $114.5 billion to Google through 2033, $110 billion to Amazon through 2036, and $44.9 billion to Microsoft. In addition, the management estimates that payments for additional computing capacity based on Google’s Tensor Processing Units (TPUs), supplied by Google and other providers and scheduled for delivery in 2028-2031, will total $125-150 billion.

In its IPO filing, Anthropic describes the cloud providers as “investors, customers, distribution partners, and competitors” simultaneously. Anthropic warns that these companies have access to information about its pricing and commercial terms and could favor their own AI models. Its relationships with them have also attracted antitrust scrutiny and could draw more in the future. Moving workloads to other providers would be “costly, time-consuming” and could reduce its compute efficiency, Anthropic acknowledged.

Anthropic has previously been criticized for participating in “circular deals” in the tech ecosystem, with the same companies both putting money into its balance sheet and taking it out, Bloomberg writes. For example, Anthropic is a major Amazon Web Services customer, using Amazon’s data centers and chips to train and run its Claude models, while Amazon offers Claude through its Bedrock platform. Similarly, Anthropic uses Google’s custom AI chips, while Google sells Claude through its own cloud platform.

Amazon and Google are among Anthropic’s largest shareholders. After the conversion of preferred shares and certain debt instruments in connection with the IPO, each company will own more than 5% of Anthropic, according to the prospectus. Their shares, however, carry no voting rights.

Amazon has invested about $18 billion in Anthropic since 2023 and committed up to another $15 billion, subject to certain conditions. Convertible debt instruments could also give Amazon an additional stake of up to 10.9% on a fully diluted basis as of June, excluding the shares it already owns. Amazon’s stake is capped at 12% unless it waives that limit.

Google has invested about $23 billion in Anthropic and committed another $20 billion tied to performance targets. Its stake is capped at 14.99%, though Google can also waive that limit.

Anthropic’s IPO will test public market investors’ appetite for an AI business amid growing concerns about massive infrastructure spending, Bloomberg writes. The company’s third-quarter revenue growth disappointed Wall Street and intensified the tech selloff amid concerns that AI investment is outpacing demand.