The financing for a new AI data center campus in Texas does not look like a typical technology loan. It looks like a real estate deal, and that is exactly the point.
A large data center project tied to Anthropic has secured $1.3 billion in private credit financing, led by Eagle Point, according to Bloomberg. The money will fund the first phase of construction at the Nexus campus in Texas. The project is one of the clearest examples yet of how AI companies are financing their physical expansion: they do not build data centers themselves, they sign long-term leases, and lenders back those leases with credit.
The structure reflects the economics of the AI build-out. Anthropic, like other frontier labs, needs enormous computing capacity, but it does not want to own the real estate, the power infrastructure or the construction risk that comes with data centers. Instead, it agrees to lease capacity for years, and that lease becomes the collateral that lets a developer borrow against future rent. The private credit market, which has grown into a major source of financing for the industry, provides the money.
Eagle Point’s role is emblematic of the shift. Private credit funds, which lend outside the traditional banking system, have become the dominant financiers of AI infrastructure, stepping in where banks are reluctant to concentrate exposure. The loans are structured around the creditworthiness of the tenant, the quality of the power supply and the likelihood that the facility stays leased. For lenders, AI companies’ long-term capacity commitments have become a dependable revenue stream.
The Texas location is no accident. The state has become a center of the AI data center boom, offering cheap land, abundant power and a permissive regulatory environment. The Nexus campus is part of a wave of construction across Texas that has drawn investment from the largest technology companies and the funds that finance them. Power availability, not land, has become the binding constraint, and developers are racing to secure electricity alongside financing.
The $1.3 billion figure covers the first phase only. Data center campuses of this scale are typically built in stages, with each phase adding capacity as tenants commit. The total investment in the Nexus project, and in Anthropic’s broader compute plans, is likely to be several times larger once subsequent phases are funded. The company has said it needs far more computing than it currently has, and its capacity agreements with partners are among the largest in the industry.
The private credit market’s appetite for AI deals has grown alongside the build-out. Funds have raised tens of billions of dollars specifically for infrastructure lending, and competition for deals has pushed down rates and loosened terms. That dynamic has made AI data centers one of the most active corners of private credit, drawing traditional asset managers and dedicated funds alike. The risk, regulators and analysts note, is that the sector is lending against projections of demand that have never been tested.
The structure also separates the risks. Anthropic’s credit is not on its own books; the debt sits with the project developer, backed by the lease. If AI demand disappoints, the developer, not the lab, carries the exposure. That division is one reason AI companies have been willing to commit to enormous capacity: the construction risk has been moved off their balance sheets.
The financing follows a familiar pattern in the AI industry. OpenAI, Microsoft, Amazon and Google have all relied on similar arrangements, signing leases with developers who then borrow against them. The difference now is scale: as AI computing demand has multiplied, so have the loans, and the private credit market has become the industry’s primary funding channel for physical infrastructure.
For investors, the deal illustrates where the money is being made in AI. The labs capture the headlines and the valuations, but the lenders, landlords and power providers collect steady returns on the capital that makes the whole system run. Private credit funds have become the quiet beneficiaries of the AI build-out, earning yields that equity investors in the labs can only dream of. The Nexus project, with its $1.3 billion first phase, is another installment in that story, a reminder that the AI economy runs on electricity, concrete and credit as much as on algorithms.
The lending terms reflect the market’s confidence in AI’s durability. Private credit funds have priced these deals on the assumption that demand for computing will keep growing for years, and that tenants will honor leases even if their own businesses hit turbulence. The yield on infrastructure loans of this kind has been attractive enough to draw steady inflows from pension funds and insurers, which see AI data centers as a rare source of long-duration cash flow in a low-yield world. That money, in turn, has made it possible for developers to break ground on projects they could not finance a few years ago, accelerating the pace of the build-out. The circular logic worries some observers: the credit market’s willingness to lend is itself a driver of the construction boom, and if the lending slows, so does the building. For now, the money keeps flowing, and the Nexus campus is one of dozens of projects across the country funded on the same model.
