Google to backstop lease payments for Anthropic-linked data centers in a $35 billion AI infrastructure financing

Alphabet’s Google has agreed to backstop lease payments tied to Anthropic’s planned AI infrastructure buildout across five U.S. data centers, according to market reporting that frames the arrangement as part of a roughly $35 billion financing package.

The reported structure is aimed at giving Anthropic’s lenders confidence that the compute assets behind the expansion will be paid for over time. Financial Post reporting described Anthropic as leasing powerful computer hardware at five data centers with the help of Google, whose backstop is characterized as helping secure what amounts to a $35 billion loan. The post also said that Anthropic’s involvement with those data centers as part of the financing had not been previously reported.

Other outlets similarly described the deal as one in which Google’s guarantee-like support helps enable Anthropic to lock in capacity through leases. Tech in Asia reported that Google agreed to backstop lease payments for the five data centers connected with Anthropic, framing the support as enabling a financing arrangement for large-scale AI compute.

The broader financing is described in the market as a private-credit-style package, with Apollo and Blackstone cited in research results as key financiers. A Reuters headline in the research results points to Apollo and Blackstone backing Anthropic’s $35 billion capacity expansion, though the underlying text could not be retrieved in the available research environment. Separately, Crypto Briefing reported on a $35 billion private-credit deal tied to Anthropic’s infrastructure needs, again aligning with the reported quantum and the role of Google-linked backstops and leasing.

For Google, the arrangement fits into a pattern of Big Tech providing capital and commercial support to the AI “infrastructure layer” that powers large language models and other compute-heavy workloads. Alphabet can benefit indirectly through its cloud, accelerator, and networking ecosystem, as well as through the positioning of its custom silicon and related supply chain tied to AI training and inference. Anthropic, for its part, is building out the data center capacity required to run and scale models such as Claude.

The size and complexity of the financing also underline how AI compute is increasingly being treated as a bankable asset class, rather than only a technology spending line item. In the reported setup, the key risk is not only whether the hardware is delivered, but whether lease obligations tied to that hardware are met as demand ramps and as the economics of AI capacity are tested.

Even with multiple reports converging on the same core facts, important specifics were not disclosed in the publicly accessible text available for this review. The reporting cited in the research results did not provide full terms such as the backstop duration, the precise reimbursement mechanics, the interest rate or credit spread on the underlying financing, or how the five sites are contractually structured. The role of any additional parties, including chip suppliers and service-level providers at each data center, also remains unclear from the available excerpts.

What to watch next is whether Alphabet, Anthropic, or the financing syndicate provides additional detail in filings or official communications, particularly around the scope of Google’s commitments and how the arrangement interacts with Anthropic’s hardware leasing and go-forward compute procurement. Investors and customers in the AI supply chain may also look for follow-on announcements about the commissioning timelines for the data centers and any changes in pricing or capacity commitments as the buildout progresses.