Anthropic has agreed to spend at least $111.1 billion on Google's cloud and infrastructure through 2033, according to a prospectus reviewed by GFdaily. The figure, disclosed in the filing, makes Google the AI developer's largest known infrastructure supplier and locks the two companies into a decade-long arrangement.
The commitment covers cloud computing capacity and related services, the prospectus indicates. It's the clearest look yet at how much capital Anthropic has pledged to a single provider as it trains and runs its Claude models.
What the $111.1 billion covers
The prospectus describes the commitment as spanning cloud services and infrastructure through 2033. It doesn't break out year-by-year spending, and it doesn't say how much has already been drawn down.
What it does show is scale. The minimum figure alone is larger than the annual capital budgets of most cloud providers' flagship regions. For Anthropic, it represents a long-term bet that demand for its models will keep growing fast enough to justify the spend.
Why the reliance on a single provider matters
Anthropic's arrangement with Google isn't unusual in AI. Building and running frontier models takes enormous compute, and a handful of companies — Google, Amazon, Microsoft — control most of the capacity. Startups and even well-funded labs end up renting from them.
But the size of this commitment sharpens the dependency. If Anthropic's revenue grows as planned, the deal is manageable. If demand for AI services cools, or if pricing for compute falls faster than expected, the company could be locked into spending well above market rates.
The prospectus flags that risk directly, noting the commitments could be strained if AI demand or pricing dynamics shift. That's boilerplate language in most filings, but here the numbers give it weight.
The Google relationship, in context
Google is both an investor in Anthropic and a supplier. The prospectus doesn't detail the equity terms, but the dual role is common in AI: cloud providers fund the labs that then rent their capacity. It creates a tight loop of capital and compute.
For Google, the commitment provides years of predictable revenue from one of its most prominent AI customers. For Anthropic, it secures access to the chips and data centers needed to keep training larger models.
What happens if the AI market shifts
The risk cuts both ways. A slowdown in AI spending would hit Google's cloud revenue, but it would hit Anthropic harder — the company would still owe the minimum commitment while its own sales lag.
The prospectus doesn't specify penalties or renegotiation terms. It also doesn't say whether the $111.1 billion is a floor or a midpoint. Those details will matter to anyone assessing Anthropic's balance sheet over the next decade.
For now, the number stands as one of the largest disclosed cloud commitments by an AI developer. Anthropic hasn't said how it plans to fund the spending, and the prospectus offers no timeline beyond the 2033 end date.
