Anthropic is reportedly offering investment firm Blackstone a yield of more than 10% to participate in a major financing arrangement for artificial intelligence chips, highlighting the growing cost of building the computing infrastructure required to operate advanced AI systems. According to a Financial Times report cited by Reuters on October 16, 2026, Blackstone-managed funds would invest $9 billion in junior debt backed by Anthropic’s five-year lease payments on the chips. The reported financing forms part of a larger $60 billion package designed to support the acquisition of computing capacity.
The proposed arrangement, known as Project North Star, is expected to finance approximately 1.5 gigawatts of Google’s custom tensor processing units (TPUs). The reported structure includes $42 billion in senior debt and $18 billion in junior debt, with Blackstone expected to provide half of the junior portion. Reuters said it could not independently verify the details. If completed on the reported terms, the transaction would illustrate how AI developers are increasingly turning to structured debt and private-credit investors to fund the infrastructure needed to serve growing demand for AI models.
Key Takeaways
- Reported yield: Anthropic is offering Blackstone a yield exceeding 10% on the proposed investment.
- Blackstone investment: Blackstone-managed funds would provide $9 billion in junior debt.
- Total financing: Project North Star is reported to involve a $60 billion chip-financing package.
- Computing capacity: The deal would support approximately 1.5 gigawatts of Google-designed TPUs.
- Debt structure: The reported package includes $42 billion in senior debt and $18 billion in junior debt.
- Collateral: The junior debt would be backed by Anthropic’s five-year lease payments on the chips.
- Important caveat: The terms were reported by the Financial Times and cited by Reuters, which said it could not independently verify them.
What Is Anthropic’s $60 Billion Chip Financing Deal?
Anthropic develops Claude, a family of AI models used for software development, enterprise productivity, research and other tasks. Operating these systems at scale requires access to large amounts of specialised computing hardware, electricity, data-centre capacity and networking infrastructure.
Project North Star is reported to be a financing arrangement intended to help fund the acquisition of approximately 1.5 gigawatts of Google’s custom TPUs. Rather than relying entirely on cash or conventional corporate borrowing, the structure would use debt linked to lease payments associated with the chips.
The reported $60 billion package is divided into two broad layers of borrowing: senior debt and junior debt. These categories determine the order in which lenders generally have claims on available payments or recoveries if a borrower encounters financial difficulty.
| Financing component | Reported amount | Structure |
|---|---|---|
| Senior debt | $42 billion | Reportedly backed by Broadcom |
| Junior debt | $18 billion | Includes the proposed $9 billion Blackstone investment |
| Total financing | $60 billion | Intended to support Google TPU acquisition |
| Computing capacity | Approximately 1.5 gigawatts | Google-designed TPUs |
Figures are based on Financial Times reporting cited by Reuters and subsequent coverage. The proposed terms should not be treated as independently confirmed or as proof that the entire financing has closed.
The scale of the transaction reflects the substantial investment required to secure computing resources for advanced AI systems. It also shows how the financing of AI infrastructure is becoming a specialised market involving technology companies, private-credit funds, equipment suppliers and institutional investors.
Why Is Anthropic Offering a Yield Above 10%?
The reported yield of more than 10% is one of the most important features of the proposed deal. A yield represents the return investors expect to earn from an investment, subject to the terms of the financing and the borrower meeting its obligations.
Higher yields generally compensate investors for some combination of credit risk, structural complexity, limited liquidity and uncertainty about future cash flows. Junior debt usually ranks below senior debt in the payment hierarchy, which can make it more exposed to losses if the financing structure runs into trouble.
In this case, Blackstone’s reported investment would be backed by five-year lease payments associated with the chips. Those contractual payments provide a basis for the financing, but they do not eliminate every risk. The ultimate outcome depends on the transaction documents, the enforceability of the relevant obligations, the parties’ creditworthiness and the wider financing structure.
A yield above 10% is also significant because it indicates that the proposed borrowing is not being priced like the safest forms of corporate debt. Investors appear to be seeking substantial compensation for participating in a large and specialised AI infrastructure financing arrangement.
The figure should not, however, be interpreted as a guaranteed profit or a risk-free return. Actual investor returns can depend on fees, issuance discounts, repayment timing, defaults, refinancing conditions and other terms that may not be fully disclosed in preliminary reporting.
How Senior and Junior Debt Work
The reported financing structure separates lenders into different levels of priority.
Senior debt generally has a higher priority for repayment than junior debt. If a financing arrangement encounters losses or a default, senior creditors typically have stronger claims on the relevant collateral or cash flows, subject to the contractual and legal structure.
Junior debt sits lower in that hierarchy. It can therefore expose investors to greater losses if cash flows fall short or the underlying assets cannot support the outstanding obligations. In return, junior lenders often demand higher yields.
The reported Project North Star arrangement follows this broad pattern, with $42 billion in senior debt and $18 billion in junior debt. Blackstone-managed funds would provide $9 billion of the junior portion, according to the Financial Times.
| Feature | Senior debt | Junior debt |
|---|---|---|
| Repayment priority | Generally higher | Generally lower |
| Exposure to losses | Usually lower than junior debt in the same structure | Usually higher |
| Expected yield | Often lower | Often higher |
| Reported amount in Project North Star | $42 billion | $18 billion |
This table describes the general characteristics of debt seniority. The precise rights and protections in Project North Star depend on the final financing documents.
The arrangement allows a large infrastructure project to draw on different types of capital. Senior lenders can participate at one level of risk, while investors willing to accept greater exposure can provide junior financing in exchange for higher potential returns.
For Anthropic, this structure could help secure substantial computing resources without funding the entire acquisition from cash reserves or issuing conventional unsecured corporate debt alone.
Why Anthropic Needs So Much Computing Capacity
Advanced AI models require computing resources for both training and inference. Training involves developing and improving a model, while inference is the process of running the model to respond to user requests.
As AI adoption expands, inference can become a major recurring expense. Businesses using AI assistants, developers generating code and organisations automating workflows can generate large volumes of requests that require processing power.
Anthropic’s computing requirements are therefore tied to more than the development of new models. The company must also provide dependable capacity for customers who expect its services to remain available and responsive.
Google TPUs are specialised chips designed to accelerate machine-learning workloads. Securing large amounts of TPU capacity could help Anthropic diversify its computing resources and support future demand.
The reported 1.5-gigawatt figure indicates the scale of the infrastructure involved. Gigawatts measure power capacity, not the number of chips or the amount of computing performance delivered. Actual computing output depends on hardware design, efficiency, utilisation and the configuration of the facilities.
Building out this capacity involves several connected expenses, including chips, data centres, power, cooling, networking and long-term operating commitments. Financing structures can spread the cost over time, but they also create obligations that must be met even if future demand develops differently from expectations.
The Growing Role of Private Credit in AI Infrastructure
Private credit refers broadly to lending arranged outside traditional public bond markets, often through funds and other non-bank financial institutions. These arrangements can be tailored to specific borrowers, assets and repayment structures.
AI infrastructure is creating opportunities for private-credit firms because technology companies need large amounts of capital to secure computing equipment and long-term capacity. Traditional financing alone may not always meet the scale, timing or structural requirements of these projects.
Private-credit investors can structure loans around contracted payments, equipment leases and other identifiable cash flows. For borrowers, such arrangements can provide another source of capital. For lenders, they can create access to deals offering higher yields than some conventional fixed-income investments.
However, the growth of AI-linked private credit also raises questions about concentration and risk. If several financing structures depend on the same assumptions about AI demand, equipment values or a small group of technology companies, a deterioration in those conditions could affect multiple investments.
Investors must assess whether the cash flows supporting the debt are sufficiently reliable, whether the underlying equipment retains value and whether the borrower can meet its commitments under less favourable business conditions.
The reported Anthropic deal illustrates the potential scale of this market, but it does not establish that every AI infrastructure financing arrangement offers comparable returns or risks.
What the Deal Means for Google and Broadcom
The proposed financing is connected to Google’s custom TPU ecosystem and Broadcom’s role in the development and supply chain for specialised chips.
Google’s TPUs are an alternative to graphics processing units (GPUs) commonly used for AI workloads. Access to multiple chip platforms can help AI developers manage supply constraints, performance requirements and the cost of running large models.
For Anthropic, obtaining additional TPU capacity could support its computing needs as demand for Claude expands. For Google, the arrangement would reinforce the commercial role of its AI infrastructure, including chips and cloud-related services.
Broadcom’s reported involvement in the senior-debt component highlights the connection between semiconductor suppliers and the financing of AI infrastructure. Hardware development and supply can involve large upfront investments, while financing structures help determine how those costs are allocated among participants.
The arrangement also illustrates that the AI infrastructure economy extends beyond model developers. It includes chip designers, equipment suppliers, cloud providers, data-centre operators, lenders and institutional investors.
The precise financial exposure of each participant will depend on the final contracts and financing documents. It would be premature to assume that all parties carry the same risks or receive the same benefits.
Anthropic’s Broader Infrastructure Commitments
Project North Star comes amid Anthropic’s wider effort to secure computing capacity over several years.
Reuters reported in September 2026 that Anthropic expected to spend at least $518 billion over a decade on AI infrastructure commitments with six partners, according to a confidential IPO prospectus reviewed by the news agency. The reported commitments include cloud services, computing capacity and equipment leases.
A substantial portion of those obligations was described as non-cancellable or payable regardless of actual usage. Such commitments can help secure access to scarce computing resources, but they also create long-term financial exposure.
The distinction between revenue growth and infrastructure spending is central to Anthropic’s business model. More computing capacity can enable the company to serve more customers and develop more capable products. Yet the company must generate sufficient revenue and cash flow to cover both current operating expenses and future contractual obligations.
If AI demand grows strongly, long-term infrastructure commitments may provide a competitive advantage by securing access to capacity. If demand disappoints, the same commitments could limit flexibility and place pressure on margins.
For investors, the key question is whether the revenue generated by AI services can support the cost of the infrastructure needed to deliver them.
What Investors Should Watch Next
The reported yield above 10% will likely draw attention from investors assessing the economics of AI infrastructure debt. However, the headline rate is only one part of the analysis.
Investors will need to understand the repayment schedule, collateral arrangements, creditor priority, fees and other terms. They will also need to assess Anthropic’s financial strength, the expected utilisation of the chips and the reliability of the lease payments supporting the debt.
Another consideration is refinancing risk. Infrastructure financing can extend over several years, and market conditions may change before debt matures. Borrowers could face higher financing costs if credit markets tighten or if lenders become more cautious about AI investment.
The transaction’s reported size also raises questions about how private-credit funds manage concentration risk. Large commitments to a limited number of AI infrastructure projects can produce meaningful returns, but they can also make portfolios more sensitive to developments in the technology sector.
The deal therefore matters not just because of its size, but because it provides a glimpse into how the financing of AI capacity may evolve as companies scale their businesses.
The Bigger Picture
Anthropic’s reported Blackstone financing proposal reflects a broader shift in the AI industry: computing capacity is becoming a strategic asset that requires increasingly sophisticated funding arrangements. A $60 billion package linked to chip leases would connect model development, semiconductor supply, equipment financing and private credit in a single structure.
The opportunity is substantial, but so are the obligations. AI companies need enough capacity to support future growth, while lenders need confidence that contractual payments will continue and that the underlying financing structure can withstand market changes. The proposed yield above 10% highlights the returns investors may demand when financing large, complex AI infrastructure projects.
Looking Ahead
The next important developments will be confirmation of the financing terms, the allocation of the junior debt and the progress of the planned chip acquisition. Until the parties or definitive documents confirm the arrangement, the reported amounts and yields should be treated as details attributed to the Financial Times rather than completed transaction terms.
Over the longer term, the economics of AI infrastructure will depend on how quickly demand grows, how efficiently chips are used and whether revenue from AI products can cover the cost of computing and financing. Project North Star could become a significant example of private credit supporting AI expansion, but its ultimate success will depend on the cash flows and contractual protections behind the deal.
Frequently Asked Questions
1. Why is Anthropic offering Blackstone a yield above 10%?
The reported yield is intended to compensate Blackstone-managed funds for providing junior debt in a large and specialised chip-financing arrangement. Junior debt generally carries greater risk than senior debt and can therefore demand a higher return.
2. How much is Blackstone expected to invest?
Blackstone-managed funds are reportedly expected to invest $9 billion in junior debt backed by Anthropic’s five-year lease payments on the chips.
3. What is Project North Star?
Project North Star is the reported name of a $60 billion financing package intended to support the acquisition of approximately 1.5 gigawatts of Google’s custom TPUs.
4. What is the difference between senior and junior debt?
Senior debt generally has repayment priority over junior debt. Junior lenders typically face greater exposure to losses if a borrower defaults, which is one reason they may demand higher yields.
5. Why does Anthropic need so much computing power?
Anthropic needs computing resources to develop AI models and serve customer requests. As usage grows, the company must secure enough chip capacity and infrastructure to deliver its services reliably.
6. Is the Blackstone financing deal confirmed?
The Financial Times reported the proposed terms, and Reuters cited that report. Reuters said it could not independently verify the details, so the reported financing should not be described as a fully confirmed or completed transaction.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
