Anthropic secures 1GW+ data center pipeline ahead of $965B IPO

In the high-stakes world of frontier AI, compute is the ultimate moat. Anthropic’s decision to lease more than a gigawatt of its own data center capacity—and the revelation that Google will co-design chips for those facilities—shows the company is building a vertically integrated hardware-software stack to train and serve larger, more capable models, all while preparing for a public listing that could redefine AI market capitalizations.

Anthropic's reported pursuit of leasing and operating its own data centers, with Google potentially providing financial guarantees, marks a pivotal shift in the AI industry's infrastructure landscape. The Information revealed on June 11, 2026, that the Claude developer has signed more than a dozen preliminary agreements across the United States, targeting a combined capacity exceeding 1 gigawatt—a scale typically associated with hyperscale cloud providers. This move goes beyond relying on third-party clouds and signals Anthropic's intent to control its compute destiny as it prepares for a public listing.

“

In late May 2026, Anthropic raised $65 billion at a staggering $965 billion post-money valuation, according to Reuters, leapfrogging OpenAI as the most valuable AI startup.

The timing is no coincidence. In late May 2026, Anthropic raised $65 billion at a staggering $965 billion post-money valuation, according to Reuters, leapfrogging OpenAI as the most valuable AI startup. That round came after Alphabet had committed up to $40 billion in April, underscoring the immense capital pouring into the sector. With the IPO confidentially filed earlier this month, Anthropic is now racing to lock in the physical infrastructure necessary to train and serve the next generation of Claude models, which are seeing surging demand from enterprise and developer users.

Industry context: The AI arms race has made compute the scarcest resource. OpenAI, backed by Microsoft, has been building massive data center clusters; xAI assembled a 100,000-GPU supercomputer; and Amazon and Meta are each spending tens of billions on AI infrastructure. Anthropic's 1 GW aspiration would place it among these titans, moving from a software-centric research lab to a full-stack AI platform operator. Owning or leasing data centers offers cost efficiency, performance optimization (especially with custom server chips co-designed with Google), and resilience against supply chain bottlenecks that have plagued the industry.

Google's deep involvement is a linchpin. Beyond being a major investor, Google is discussed as providing financial guarantees for the lease payments, a structure that reduces Anthropic's upfront risk while binding the two companies more tightly. Google also co-designs the chips slated for those data centers, likely its Tensor Processing Units or custom silicon derivatives, which would give Google Cloud a competitive edge if Anthropic continues to use its infrastructure. However, this entanglement could draw antitrust scrutiny, especially if Anthropic's IPO succeeds and Google's role as both investor and guarantor is seen as anticompetitive.