Anthropic Locks in 1 GW Compute as AI Model Arms Race Intensifies

For AI researchers and engineers, the news is about more than just real estate—it’s about the compute foundation needed for next-generation models like a potential Claude 4 or beyond. By securing a dedicated 1 GW footprint, Anthropic is signaling that the next frontier in AI requires bespoke, at-scale infrastructure that public clouds may not efficiently deliver. This move underscores a pivotal shift in AI development: from renting compute to building the factory.

Anthropic, the AI startup behind the Claude family of large language models, is making a monumental infrastructure play that underscores the enormous compute demands of the next generation of artificial intelligence. According to a June 11 report by The Information, the company has signed more than a dozen preliminary agreements to lease data centers across the United States, with a combined power capacity exceeding one gigawatt. This is not a speculative land grab — it is a strategic move to secure the physical foundation required to train and run increasingly complex AI models, reduce dependence on public cloud providers, and prepare for life as a public company. The scale is breathtaking: one gigawatt is roughly the output of a large nuclear power plant, enough to supply hundreds of thousands of high-end server chips simultaneously. By comparison, typical hyperscale data center campuses operate in the 100–300 megawatt range. The sheer ambition of Anthropic’s plans signals that the company is not content to merely compete — it aims to set the pace in the AI arms race.

“

In late May 2026, Anthropic raised $65 billion at a post-money valuation of $965 billion, according to Reuters, vaulting it ahead of rival OpenAI in terms of private market worth.

The backdrop to this expansion is a staggering influx of capital. In late May 2026, Anthropic raised $65 billion at a post-money valuation of $965 billion, according to Reuters, vaulting it ahead of rival OpenAI in terms of private market worth. Earlier in April, Alphabet had announced plans to invest up to $40 billion in Anthropic, deepening the ties between the search giant and the AI startup. Google already co-designs custom server chips that could be used in Anthropic’s new facilities, and the report indicates that executives have discussed a potential arrangement where Google would provide financial guarantees for the lease payments. Such a guarantee would dramatically reduce the risk for lease landlords and probably lower Anthropic’s cost of financing this massive buildout. The Google connection is pivotal: it ties Anthropic’s infrastructure ambitions to one of the world’s largest cloud providers, yet simultaneously suggests that even Google’s cloud may not be able to accommodate the scale Anthropic anticipates. In effect, Anthropic is preparing to become its own cloud.

This shift has profound implications for the AI industry and its supporting infrastructure. For years, AI model companies have largely rented compute from cloud providers like AWS, Azure, and Google Cloud. But as frontier models grow in size and complexity, the economics and logistics of cloud rental are being stretched. Owning or leasing dedicated infrastructure grants more control over hardware selection, networking design, and operational reliability — all of which directly affect the pace of research and the efficiency of training runs. Anthropic’s move parallels similar strategies by competitors: NVIDIA itself has pushed for AI factories, and Meta has built large AI research superclusters. The trend is clear: the AI firms with the deepest pockets are vertically integrating into infrastructure.

The timing of the leak is no coincidence. Anthropic confidentially filed for an initial public offering in the United States earlier in June 2026. The company has not disclosed terms, but the IPO will almost certainly rank among the largest tech offerings in history. By solidifying its compute strategy now, Anthropic is de-risking its growth narrative for potential investors. A compelling data center story — with guaranteed power capacity, financial backing from Google, and a clear path to independence — could be the linchpin of its public market pitch.