Blackstone Tests Appetite for a $36bn Anthropic Financing

Blackstone has held early discussions with investors about a potential large-scale debt package that would finance Anthropic’s use of Google artificial-intelligence accelerator chips, according to reports. The talks are at an early stage, and the size, structure, pricing and timing of any transaction could change. Even so, the initial proposal has drawn attention because it would be one of the largest private-credit deals ever assembled for an AI company.

One early outline for the financing was for at least $36bn, or roughly £26.8bn, the reports said. That figure would exceed the $35bn package that Apollo Global Management and Blackstone arranged about two months ago, also to fund Anthropic’s lease of Google AI chips. The earlier transaction was already described as one of the largest private-credit transactions in history, underscoring how quickly AI infrastructure spending has moved into the private capital markets.

The new discussions come as technology companies strike increasingly complex and often circular deals involving AI chips, data centres, cloud capacity and financing. In this case, the debt would help Anthropic secure access to Google’s tensor processing units, or TPUs, and related computing capacity. Google is both an investor in Anthropic and a major supplier of cloud and chip resources, making the financing structure part of a broader web of commercial relationships.

Key Facts at a Glance

• Blackstone has held early talks with investors about a potential large-scale debt package for Anthropic’s use of Google AI accelerator chips.
• An initial proposal was for at least $36bn, or about £26.8bn.
• The proposed deal would exceed the $35bn package that Apollo Global Management and Blackstone arranged about two months ago for a similar purpose.
• The earlier $35bn transaction was one of the largest private-credit deals in history.
• Google is an early Anthropic investor and has been increasingly backstopping financing deals tied to chips and computing power bought through Google Cloud.
• Google’s infrastructure financing programme for Anthropic is estimated to cover more than $150bn worth of TPU chips.
• A deal announced in April gave Anthropic access to Google AI processing power and involved commitments by Anthropic to pay Google about $200bn over five years.
• Those commitments suggested Anthropic accounted for more than 40% of the cloud revenue backlog Google disclosed at the time.
• Google said it would invest an additional $10bn or more in Anthropic as part of their arrangement.

Why AI Infrastructure Is Moving Into Private Credit

The scale of modern AI computing has created financing demands that traditional venture funding and public markets cannot easily absorb. Training and running frontier models requires enormous quantities of accelerators, high-bandwidth networking, power and data-centre space. The upfront costs are large, the useful life of the equipment is uncertain, and the competitive race is intense. For AI developers, securing compute has become as important as securing talent or capital.

Private-credit funds have stepped into this gap because they can write very large cheques, negotiate bespoke terms and move quickly. Banks might be constrained by capital rules or syndication timelines. Public bond markets might demand ratings and disclosure that are difficult for fast-growing private companies. Private lenders, by contrast, can structure loans around specific assets, contracts or revenue streams. In the Anthropic case, the financing is tied to the company’s use of Google chips, which gives lenders a defined commercial relationship to analyse.

Another driver is supplier financing. Chipmakers and cloud providers have an incentive to help customers pay for capacity because it locks in demand and supports their own revenue forecasts. When a cloud provider is also an investor in the customer, the lines between equity, debt, procurement and revenue become blurred. That can make deals easier to close, but it also makes them harder for outside investors to evaluate.

The Google-Anthropic Financing Web

Google was one of Anthropic’s earliest investors and has become an increasingly important partner in the company’s compute strategy. The relationship extends beyond equity: Google Cloud provides infrastructure, and Google’s TPUs give Anthropic an alternative to Nvidia’s graphics processing units, which have been in heavy demand across the AI industry. TPUs are custom-designed for machine-learning workloads, and access to them can be a strategic advantage for a model developer trying to manage costs and supply constraints.

Google’s infrastructure financing programme for Anthropic has been constructed to enable the AI company to buy access to Google’s own chips and data-centre capacity. According to estimates, the programme encompasses more than $150bn worth of Google TPU chips. That staggering figure reflects both the physical scale of the build-out and the financial engineering required to fund it. The chips are not simply sold in a one-off transaction; they are part of a multi-year relationship involving cloud services, investment and procurement commitments.

The April deal that gave Anthropic access to Google AI processing power included commitments by Anthropic to pay Google about $200bn over the next five years, according to reports. The figures suggested that Anthropic accounted for more than 40% of the cloud revenue backlog Google disclosed at the time. Google also said it would invest an additional $10bn or more into Anthropic as part of the arrangement. Taken together, the commitments show how a single AI developer can become a material customer, investee and strategic partner for a large cloud provider at the same time.

Circular Financing Raises Questions

The structure has drawn scrutiny because it is circular in nature. Google invests in Anthropic; Anthropic commits to buy Google Cloud capacity and TPU access; lenders finance Anthropic’s obligations; and Google’s revenue backlog grows. Each piece can be justified on its own, but the interlocking relationships mean that the health of one party affects the others. If AI demand slows, if Anthropic’s model development lags, or if Google’s chip roadmap slips, the financing could come under pressure.

Circular deals are not inherently improper. Vendor financing, strategic investments and long-term supply contracts are common in capital-intensive industries. They can accelerate adoption and reduce friction for customers that need expensive equipment. But they can also obscure true demand, concentrate risk and create incentives to inflate commitments. For private-credit investors, the key questions are whether the underlying cash flows are genuine, whether the collateral is enforceable and whether the borrower has enough flexibility to survive a downturn.

The reliance on private credit also raises questions about transparency. Private loans are negotiated privately, and their terms are not always visible to public-market investors. That can make it difficult to assess leverage across the AI ecosystem. If several large AI companies borrow against similar assumptions about future compute demand, the risk may be correlated in ways that are not obvious from individual deal documents.

Chip Access as a Strategic Asset

For Anthropic, the ability to secure chips is a strategic imperative. Frontier AI models require massive compute for training and inference, and supply has been constrained by manufacturing capacity, packaging and data-centre power. Access to Google TPUs gives Anthropic a second source of accelerators alongside other providers and helps it plan capacity for years ahead. The debt package, if completed, would effectively convert future compute needs into present financing, allowing Anthropic to commit to long-term capacity without paying the full cost upfront from its own balance sheet.

For Google, the arrangement supports demand for its TPUs and cloud services while deepening its ties to a leading AI developer. The company has invested heavily in custom silicon and data-centre infrastructure, and large anchor customers help justify that spending. The additional $10bn investment commitment shows that Google is willing to put more equity behind the relationship, not just supply hardware. That combination of equity, debt and procurement is becoming a template for how AI infrastructure is financed.

Deal Size and Market Signals

The proposed $36bn package would be a significant test of investor appetite. It would surpass the $35bn transaction arranged by Apollo Global Management and Blackstone about two months ago, which was already one of the largest private-credit deals in history. The fact that Blackstone is reportedly exploring an even larger package suggests that lenders see AI infrastructure as a generational opportunity, even as they recognise the risks.

Private-credit firms have been searching for large, high-quality assets as traditional buyout lending becomes more competitive. AI infrastructure offers scale, long-term contracts and strategic importance. It also offers yields that may be attractive relative to public bonds. But the concentration of exposure to a small number of AI developers and cloud providers is a concern. A single financing of this size could tie up significant capital and reduce diversification for lenders that participate.

The final structure will matter as much as the headline number. Lenders will want to know who bears the risk if Anthropic’s compute needs change, whether Google provides any backstop, how the debt is secured and what covenants protect investors. They will also examine the maturity profile and the pricing of the loan. A $36bn deal would likely be syndicated across multiple investors, reducing individual exposure but increasing complexity.

What to Watch in the Financing Talks

The talks are early, and no final terms have been disclosed. Investors will watch for the size of the package, the identity of participants, the role of Google and the conditions attached to the debt. They will also look for signs that the financing is tied to specific chip deliveries, cloud contracts or revenue milestones. Those details will determine whether the deal is a one-off financial engineering exercise or a repeatable model for funding AI infrastructure.

Another key issue is how the transaction affects Anthropic’s balance sheet and future fundraising. A large debt package could give the company more runway to secure compute, but it could also constrain its flexibility if AI markets shift. The company’s investors will want to see that the debt is matched to realistic demand and that the obligations are manageable alongside other commitments.

For the private-credit market, the deal could set a benchmark for how lenders price and structure AI-related risk. If it succeeds, more AI companies may turn to private credit for chip and data-centre financing. If it struggles, the caution could slow the pace of similar transactions. The outcome will be closely watched across technology, asset management and infrastructure finance as the AI build-out continues to reshape capital markets.

Source: Silicon UK News