Google is moving to finance a massive AI data center campus in Texas for Anthropic to lease, according to a Financial Times report cited by CoinGeek. The initial phase is reported to exceed $5 billion. Construction has reportedly already begun on the 2,800-acre site, known as the Goodnight campus, with approximately 500 megawatts of capacity expected to come online by late 2026.
The financing structure has two layers. Google is providing a loan for the construction phase as part of its broader infrastructure arrangement with Anthropic. Separately, a consortium of banks is reportedly competing to provide additional construction financing by mid-year. Google already holds an equity stake in Anthropic, cross-references place that figure at approximately 14%, following Anthropic’s Series G round, though that figure requires T2 source confirmation before it should be treated as definitive.
This is a reported deal, not an officially announced one. The FT’s sourcing is “people familiar with the matter.” Until Google or Anthropic issue statements, every specific figure carries that qualifier.
Why it matters. The scale here is significant. Five hundred megawatts is a large initial commitment. For context, a typical hyperscale data center campus runs 100-200MW. The Goodnight campus, at 500MW initial and a 2,800-acre footprint, is being built for AI training and inference workloads at a scale that few organizations outside the hyperscalers themselves can finance or fill. The fact that Google is financing infrastructure for a company it also invests in, and whose models run on Google Cloud, illustrates how deeply the frontier AI supply chain is consolidating around a small number of hyperscaler-lab relationships.
There’s a secondary market signal embedded in this story. Reports that Anthropic is internally testing a new model called Mythos reportedly contributed to a decline in cybersecurity stocks, with CNBC reporting identifying CrowdStrike (CRWD) as one of the affected names. The reasoning: a more capable general AI model could reduce demand for specialized cybersecurity software. Whether that thesis holds is debatable, but the market reaction is real and worth tracking.
Context. Google’s infrastructure relationship with Anthropic is layered. There’s an equity stake. There’s a cloud partnership. Now there’s direct construction financing. According to the FT’s reporting, the data center loan is itself part of that broader infrastructure deal, not a standalone transaction. That structure mirrors, in broad outline, how Microsoft has built its infrastructure relationship with OpenAI over the past several years.
What to watch. The bank consortium financing by mid-year is the next concrete milestone. If a financing package closes on schedule, that confirms the deal’s forward momentum and provides a market-verified price signal for AI infrastructure debt. Watch also for any official statement from Google or Anthropic, the absence of one is notable given the scale of reporting. And track whether the Mythos-related cybersecurity stock reaction sustains or reverses as more detail emerges about what the model actually does.
TJS synthesis. The Goodnight campus isn’t just a data center story. It’s a capital structure story. When a frontier AI lab’s physical compute infrastructure is financed by its largest cloud partner and equity holder, the resulting concentration of dependency cuts both ways: Anthropic gets infrastructure it couldn’t self-finance at this scale, and Google deepens a strategic position across equity, cloud, and now physical infrastructure. For enterprise AI buyers evaluating long-term compute access, and for investors watching AI infrastructure, this deal is worth understanding as a structural signal, not just a headline number.