Anthropic and OpenAI Reportedly Seek Smaller Data Center Deals as AI Infrastructure Race Accelerates

Anthropic and OpenAI are reportedly pursuing smaller AI data center deals, sources told CNBC, as both labs look beyond the massive infrastructure projects that have defined the past year of AI expansion.

The two companies have signed major agreements for multi-hundred-megawatt and gigawatt-scale facilities, but sources say they are now also exploring compute capacity deals in the 20-30 MW range.

Anthropic has held discussions about agreements in that range across the U.K. and the Nordics, according to four people familiar with the conversations. OpenAI has been exploring similar smaller deployments in the Nordics, two of those sources said. One source also said they were aware of talks involving both companies about U.S. capacity at that scale.

An OpenAI spokesperson told CNBC the company is "building a diversified compute portfolio to meet growing demand for AI around the world," adding that "different workloads need different infrastructure." Anthropic declined to comment.

Smaller deals offer speed. Jabez Tan of Structure Research told CNBC that securing a few megawatts at an existing powered site can be more practical than waiting for a larger block, and that collections of smaller deployments can add up to substantial capacity.

The shift comes as inference workloads - running AI systems day-to-day rather than training them - are expected to grow. A JLL report projects inference will surpass training in share of data center capacity by 2027, reaching 37% by 2030.

In February, Nvidia announced it would work with data center stakeholders to study smaller-scale facilities designed for distributed inference. Meanwhile, Crusoe is investing in smaller data centers, the Wall Street Journal reported, and announced a $3.9 billion funding round at a $30.9 billion post-money valuation.

Large data center projects are increasingly facing pushback from local communities in the U.S. and pressure in Europe, where land and power are scarce.

Source: CNBC