Google Nears $5B Deal to Finance Anthropic Texas AI Data Center

Broader Context: Anthropic has separately pledged $50 billion for U.S. data centers, while Google has committed $40 billion to Texas infrastructure through 2027.

Google is close to finalizing a deal worth more than $5 billion to finance a Texas data center for AI computing, the Financial Times reported. A consortium of banks is vying to finance the first phase of the project, with Alphabet's strong credit rating expected to help secure financing at a lower cost. A shift in Google's role from cloud provider to direct infrastructure financier for one of AI's highest-valued startups, the arrangement marks a new chapter in tech-industry capital flows.

Moreover, Anthropic is now valued at $170 billion after a ninefold increase from its previous valuation. Google's move from cloud provider to direct infrastructure financier comes as AI companies race to secure computing capacity amid an industry-wide capital expenditure surge. Separately, Anthropic last November announced a commitment to invest in American computing infrastructure, underscoring how rapidly AI companies are scaling their physical footprints to keep pace with demand for frontier model training.

Nexus Data Centers will operate the Texas facility, with Alphabet serving as financial underwriter rather than constructor. A consortium of banks is competing to provide financing for the initial phase by mid-year, according to the Financial Times. Anthropic signed a lease for the site earlier in March, and construction is already underway, suggesting both companies see urgent demand for new computing capacity.

Furthermore, initial capacity will reach around 500 megawatts, with expansion potential to approximately 7.7 gigawatts. Located near pipelines operated by Enterprise, Energy Transfer, and Atmos, the site enables behind-the-meter power generation via gas turbines, giving Anthropic direct access to energy without relying on the public grid. Behind-the-meter generation allows the facility to produce power on-site rather than drawing from the utility network, reducing both transmission costs and exposure to grid congestion.

Beyond financing, the arrangement also involves hardware. Google holds exclusive rights to supply up to 1 million custom AI chips to Anthropic and already provides TPU infrastructure through Google Cloud for training language models.

As a result, by combining compute supply with direct infrastructure financing, Google is positioning itself as both technology vendor and capital partner. Rather than simply selling compute by the hour, Google is now underwriting the physical infrastructure where that compute runs. For Anthropic, the arrangement secures both financing and a guaranteed chip pipeline; for Google, it creates a structural dependency that ties Anthropic's operations to Alphabet's hardware and capital for years to come.

Anthropic's own infrastructure commitment extends well beyond the Google-financed facility. The company pledged $50 billion for data centers across Texas and New York, as WinBuzzer previously reported. The program is expected to create approximately 800 permanent jobs and 2,400 construction jobs. An Anthropic spokesperson previously confirmed in February 2026 that data centers are being built in Mitchell County, Texas and Niagara County, New York, with an additional facility in southeast Louisiana.

Building on this, according to the same announcement, Anthropic now serves more than 300,000 business customers, while its large accounts generating over $100,000 in annual run-rate revenue grew nearly sevenfold in the past year. For its data center buildout, Anthropic selected Fluidstack as a partner to deliver gigawatts of power quickly. Gary Wu, co-founder and CEO of Fluidstack, described the partnership as a defining opportunity, emphasizing Fluidstack's focus on rapidly deploying infrastructure for frontier AI development.

In turn, with training runs for frontier models requiring ever-larger clusters of GPUs and custom chips, compute capacity has become the binding constraint on AI research progress. Sevenfold growth in large accounts translates directly into compute demand that existing cloud infrastructure cannot absorb, making physical expansion a prerequisite for continued revenue growth.

"We're getting closer to AI that can accelerate scientific discovery and help solve complex problems in ways that weren't possible before. Realizing that potential requires infrastructure that can support continued development at the frontier. These sites will help us build more capable AI systems that can drive those breakthroughs, while creating American jobs."

Amodei's framing positions the spending not as a competitive response but as a prerequisite for frontier AI development, connecting the physical infrastructure expansion to Anthropic's broader research mission.

Google committed $40 billion to Texas through 2027 for cloud and AI infrastructure, according to a company blog post published in November 2025. It represents the company's largest single-state investment ever. According to the same announcement, Google contracted more than 6,200 megawatts of new energy generation and created a $30 million Energy Impact Fund for energy initiatives in the state.

Meanwhile, in Haskell County,the company will build a data center alongside a new solar and battery storage plant. Separately, Google is funding the Electrical Training ALLIANCE to train more than 1,700 apprentices in Texas by 2030, addressing workforce demands tied to data center construction.

In October 2025 Google committed $24 billion to AI infrastructure across the U.S. and India, as WinBuzzer reported. Google's financial backing for Anthropic also fits into a wider competitive strategy against OpenAI and Nvidia that combines model development, partnership deals, and infrastructure control.

By financing Anthropic's infrastructure while simultaneously building its own Texas data center campuses, Google is pursuing a two-track strategy: securing long-term influence over a key AI partner while expanding its own computing footprint in a state that has become the epicenter of AI infrastructure development.

However, Google's deal does not exist in isolation. SoftBank, OpenAI, and Oracle have jointly committed to a multibillion-dollar AI infrastructure venture known as Stargate, with eight data centers planned in Abilene, Texas. Such spending would have seemed implausible just two years ago.

In addition, Meta has committed to heavy U.S. infrastructure spending through the end of 2028, including a sprawling site in Louisiana with multiple gigawatts of power capacity.

Meta's Hyperion data center campus in Richland Parish will be powered by seven new gas-fired plants providing 5.2 gigawatts of total generation capacity. Across the industry, hyperscaler capital expenditure on AI infrastructure has surged as companies race to secure the physical resources needed to train and serve increasingly large models.

Against this backdrop, the Anthropic financing deal represents Google's attempt to lock in a competitive position before rival cloud providers can offer similar arrangements. Amazon has invested $8 billion in Anthropic, giving it material influence over the startup's direction. Google's combination of chip supply, cloud infrastructure, and direct capital creates a layered dependency that goes further than a single equity stake.

Consequently, Anthropic has responded by diversifying its infrastructure partnerships, working with Fluidstack for operations, Nexus for facility management, and both Google and Amazon for cloud compute and financing. How long that balancing act can hold as investment scales into the tens of billions remains an open question for the company's independence.

On the regulatory front, Senators Bernie Sanders and Alexandria Ocasio-Cortez have proposed legislation to ban new data centers with peak power loads exceeding 20 megawatts until Congress enacts comprehensive AI regulation. ERCOT estimates that data center energy demand in Texas could reach 78 gigawatts by 2031, and Texas enacted Senate Bill 6 in June 2025 to address data center interconnection and cost-sharing concerns, signaling that regulatory frameworks are struggling to keep pace with the industry's appetite for power.