Anthropic Deepens Cloud Ties as AI Distribution Costs Surge
Anthropic PBC is becoming more reliant on Alphabet Inc.’s Google, Amazon.com Inc., and Microsoft Corp. for the dissemination of its artificial intelligence technology to clients, despite the fact that these companies are competitors and major suppliers of computing resources. Customers of Amazon, Google, and Microsoft have the opportunity to acquire access to Anthropic’s Claude AI models through their existing cloud accounts. The cloud providers issue invoices to the customer, with a portion retained as a fee, while the remainder is transferred to Anthropic. That business accounted for 48% of the AI startup’s revenue in the first half of 2026, a notable rise from 11% in 2023, as detailed in Anthropic’s initial public offering filing released on Monday. The arrangement has facilitated Anthropic’s swift growth. In the second quarter, sales amounted to $11.6 billion, with the company recording an operating profit of $974 million. However, the expenses incurred by Anthropic in compensating the three cloud service providers accumulate considerably. The AI lab’s sales, marketing, and partnership expenditures surged to $3.2 billion in the first half of the year, a significant rise from $120 million in the same period last year. This escalation was primarily fuelled by a $2.7 billion increase in fees allocated to channel partners, which encompasses the cloud providers.
At the same time, the S-1 filing warns of risks. Anthropic describes the cloud providers as “investors, customers, distribution partners and competitors” all at once. It cautions that the companies have “visibility into our pricing and commercial terms.” The filing also says the cloud companies could favor their own AI models, and that Anthropic’s ties to them have attracted antitrust scrutiny and could draw more. Anthropic’s IPO will serve as a significant indicator of AI sentiment within the realm of public market investors, who appear increasingly aware of the considerable costs linked to AI development. Reports of Anthropic’s third-quarter revenue growth slowing compounded a broader tech selloff earlier this month amid concerns that AI spending may be outpacing demand. Anthropic has encountered scrutiny for participating in “circular deals” within the technology sector, wherein the same firms provide capital to its balance sheet while concurrently retracting funds. Anthropic is a significant client of Amazon Web Services, employing Amazon data centers and chips for the training and functioning of its Claude models, while Amazon facilitates access to Claude through its Bedrock platform. Similarly, Anthropic employs Google’s custom AI chips, while Google promotes Claude through its own cloud platform.
Anthropic depends on the same companies for its computational resources. It has committed to designating at least $114.5 billion to Google by 2033, $110 billion to Amazon by 2036, and $44.9 billion to Microsoft, irrespective of capacity utilisation. The figures contribute to the intricate dynamics of the relationships among the parties involved. When Anthropic secured its agreement with Amazon in April, it designated AWS as its “primary training and cloud provider for mission-critical workloads.” However, its minimum commitment to Google is greater and spans a more limited timeframe. Anthropic also estimates it will incur an additional expenditure of $125 billion to $150 billion for supplementary Tensor Processing Unit computing capacity from Google and other providers, scheduled for delivery between 2028 and 2031. Amazon has formed a substantial investment and commercial alliance with Anthropic, subsequent to its pledge in 2023 to invest as much as $4 billion. It has invested approximately $18 billion in Anthropic and has pledged up to an additional $15 billion, contingent upon specific conditions, as stated in the filing.
It currently holds an economic stake of approximately 9% on an as-converted, fully diluted basis, based on calculations. It also holds approximately $6.2 billion of outstanding convertible notes issued under financing agreements signed in 2023 and 2024, which could convert into 1.84 billion shares – representing an additional 11% of Anthropic. The company’s ownership stake is limited to 12%, although Amazon maintains the option to forgo this restriction. Google has committed around $23 billion to Anthropic through a series of agreements and has promised to invest an extra $20 billion, dependent on the achievement of specific performance metrics. It currently holds an economic stake of at least 6.6% on an as-converted, fully diluted basis, according to calculations, which exclude a separate $300 million investment by the company in 2002 that was reported by the Financial Times and isn’t detailed in Monday’s filing. It also possesses 20 billion in remaining financing capacity under a 2026 agreement, which could lead to further equity or convertible-note issuance if utilised.
Its stake is capped at 14.99% under its latest financing deal; however, Google retains the option to waive that limit. Additionally, Google has the opportunity to invest up to $10 billion in stock via a private placement that aligns with Anthropic’s IPO at the public offering price. Any amount invested through that option would reduce the remaining capacity under the financing agreement. The filing delineates the manner in which Anthropic has interwoven its own success with the cloud companies. According to the company, shifting its workloads to alternative providers would be described as “costly, time-consuming” and could negatively affect its compute efficiency. Anthropic’s reliance extends beyond the three major cloud providers, however. It also depends on alternative suppliers. It possesses 161.2 billion in equipment leases with Broadcom Inc. and has committed to allocate up to 120.5 billion for computing power from SpaceXAI. The firm also indicates that chipmaker Taiwan Semiconductor Manufacturing Co. has at times faced challenges in meeting demand.
