Anthropic’s $15 Billion Texas Data Center Is Being Paid For By Google and Morgan Stanley
Anthropic cannot fund its own data center buildout, even on a $965 billion valuation. That is the headline buried inside a Thursday, July 30, 2026, financing package in which a Morgan Stanley-led bank group agreed to lend $15 billion to Nexus Data Centers for a 1.6-gigawatt AI campus in Hubbard, Texas, with Alphabet’s Google backstopping Anthropic’s obligations using its investment-grade credit rating, according to CNBC, Reuters, and The Wall Street Journal.
The structure of the deal is what makes it significant. Anthropic is not the borrower. Nexus Data Centers, a Texas-based privately held developer of large-scale AI-ready data-center campuses, is the borrower, and Google has agreed to guarantee Anthropic’s lease and power-purchase obligations on the four buildings and the on-site natural-gas plant. As Yahoo Finance reported, citing CNBC, the guarantees extend specifically to four Anthropic data-center leases and to the associated power-purchase agreements tied to the 1.6-gigawatt gas plant, which would make it one of the largest behind-the-meter data center power installations in the United States.
Why A $965 Billion Valuation Cannot Pay For A Watt
Anthropic, the developer of Claude Code, is preparing for an October 2026 initial public offering at a valuation of $965 billion, as Forkast reported on Thursday. The market capitalization would make Anthropic one of the most valuable private companies in the world, but the IPO valuation is a paper number. The actual balance sheet does not have the cash to self-fund the kind of multi-gigawatt infrastructure that frontier AI labs now need. A single 1-gigawatt data center campus can cost $12 billion to $15 billion before chips are installed, and Anthropic has signed more than a dozen preliminary agreements to lease US data centers with a combined capacity of more than 1 GW, according to The Information, as reported by Channel News Asia.
That gap between valuation and balance sheet is what Google’s guarantee solves. Google’s backing is limited to the minimum level required by lenders to complete the financing, per Reuters, but it is enough to convert what would otherwise be a speculative AI buildout into a creditworthy loan. The $15 billion deal is structured as a $14 billion bridge loan paired with a revolving credit facility, and the financing is reported to be in advanced talks rather than fully closed.
What The Google Backstop Actually Means
Google’s guarantee is the structural lever that makes the whole arrangement work. The investment-grade rating lowers Anthropic’s effective borrowing cost, allowing Nexus to tap the bank market for a $15 billion check rather than a $5 billion one. It also means that if Anthropic ever fails to make a lease payment, the obligation flows to Alphabet, not to Nexus’s lenders. That is a meaningful credit transfer in a market where every gigawatt of new AI capacity is being treated by investors as a test of whether the buildout will generate enough cash flow to service its own debt.
According to MediaPost, citing the original WSJ report, the consortium of banks led by Morgan Stanley will lend the $15 billion to Nexus for the Hubbard campus, and Google will provide guarantees to back the four Anthropic leases. Anthropic’s executives, The Information reported, are also in discussion with Google about an arrangement under which Google, which already co-designs some of the server chips Anthropic could use in the facilities, would provide a financial guarantee covering part of the lease cost. The end result is a stack of credit that is more like a US utility-project financing than a typical AI capex round.
How The OpenAI Restructure Compares
The Anthropic package is the second large AI infrastructure financing to make headlines in the past two months. In late April 2026, OpenAI and Microsoft amended the terms of their partnership so that OpenAI could court additional investors, including Amazon, ending the exclusivity clause that had bound the two companies. CNBC reported at the time that the change was framed as letting OpenAI raise capital against its own valuation while keeping Microsoft as a core partner. The restructure removed Microsoft’s claim to a large share of OpenAI’s revenue, a significant concession on Microsoft’s part.
Anthropic has chosen a different path, keeping Google as the single anchor backer on the hardware side while diversifying elsewhere. The 2nm custom-chip talks between Anthropic and Samsung, first reported by The Information in early July, are evidence of the same pattern. A frontier lab is hedging across multiple hardware suppliers, with Google providing credit support on the data-center side and Samsung providing foundry capacity on the chip side. Anthropic has not yet decided what the Samsung chip will be optimized for, but the existence of those talks tells lenders that the company is building operational optionality rather than concentrating all of its infrastructure risk in one vendor relationship.
What Changes When The Loans Close
If the $15 billion deal closes on the timeline now being reported, the Hubbard campus would represent a single site consuming more power than the entire city of Austin. That scale of behind-the-meter generation, a 1.6-gigawatt on-site natural gas plant, would also make the data center one of the largest single corporate power purchasers in ERCOT, the Texas grid operator. The market consequence is that other frontier labs lose the ability to claim, as they did in 2024 and 2025, that their compute capacity is constrained only by access to capital. The Morgan Stanley package shows that capital is now available, with the right credit structure, for multi-gigawatt AI builds.
For Alphabet, the backstop is a quiet but significant deepening of the existing Anthropic relationship. Google has already invested billions in Anthropic and co-designs server chips with the company, and the credit guarantee adds a third leg, financing for the physical plant. For Nexus Data Centers, the $15 billion package is a validation of the developer-led model, in which a specialist firm takes construction and power risk in exchange for a long-dated tenant lease. For Anthropic, the deal removes the single biggest near-term constraint on its pretraining roadmap. The company’s October IPO, if it lands at the implied valuation, will refinance these commitments at a lower cost, but the real value of the $15 billion package is that it is being underwritten at all.