Anthropic and OpenAI hunt for smaller data center deals, sources tell CNBC, in race to deploy AI capacity

Anthropic and OpenAI hunt for smaller data center deals, sources tell CNBC, in race to deploy AI capacity. In this articleNVDAFollow your favorite stocksCREATE FREE ACCOUNTAnthropic and OpenAI are hunting for smaller AI data center deals, sources told CNBC, as the race to access the infrastructure needed to deploy workloads ramps up.

What happened

Skip NavigationMarketsBusinessInvestingTechPolitics & PolicyVideoWatchlistInvesting ClubPROLivestreamMenuKey PointsAnthropic and OpenAI are exploring opportunities for smaller data center deals, sources told CNBC. The two AI labs have both inked huge deals for AI data centers in the past year for facilities of multi-hundred-megawatt and gigawatt capacity, but sources have said those companies are now also looking for compute capacity deals for much smaller deployments of 20-30 MW. The sector is also under pressure in much of Europe, where available land and power are in short supply. watch nowVIDEO4:2204:22CoreWeave CEO: AI industry has not done a good job explaining data center impact on communitiesSquawk on the StreetSmaller capacity deals are often attractive because of "speed to usable capacity," Jabez Tan, head of research at Structure Research, told CNBC.

OpenAI had been exploring opportunities for those smaller capacity deployments in the Nordics, two of the sources said. "Anthropic did not comment when approached by CNBC. 'Speed to usable capacity'Both AI labs typically rent compute capacity from data center operators and neoclouds and have sought large-scale, long-term agreements. Anthropic inked a roughly $45 billion cloud deal with Nscale, which will see the AI lab rent around 460 MW of compute capacity at a data center development in West Virginia, two people familiar with the matter told CNBC in August.

Smaller capacity deals are often attractive because of "speed to usable capacity," one analyst told CNBC.

The wider picture

S. company Crusoe, which built a huge data center complex in Texas used by OpenAI, is now investing in smaller data centers, the Wall Street Journal reported on Thursday. Anthropic and OpenAI have announced a flurry of AI infrastructure deals over the past year as demand booms. Deals to secure smaller allocations of compute allow companies to deploy workloads faster amid the AI boom. "For workloads that can operate across separate sites, a collection of smaller deployments can add up to substantial capacity.

"Shift to inferenceTraining AI models requires large amounts of computing power to process huge quantities of data, but deploying those systems day-to-day — a process known as inference — can be done with smaller clusters of chips. The proportion of total data center capacity used for inference workloads is expected to overtake training workloads in 2027, according to a report by real estate company JLL. In February, it was announced that Nvidia would collaborate with several data center stakeholders to study smaller-scale data centers designed for distributed inference.

One source said they were also familiar with talks involving Anthropic and OpenAI about U.

What has been reported

"We're building a diversified compute portfolio to meet growing demand for AI around the world," an OpenAI spokesperson told CNBC. In 2025, inference made up 9% of global workloads in data centers compared to 14% for training, the report said. Anthropic has sounded out agreements within that range across the U. K. and the Nordics, four people familiar with the conversations, who asked to remain anonymous when discussing private business dealings, told CNBC. Both companies have announced a flurry of AI infrastructure deals over the past year as they've looked to train and serve their models to end users.

OpenAI has said it surpassed the original commitment of 10 GW to its Stargate AI infrastructure project in April and has since committed to developing a further 3 GW in Georgia and 8 GW in Ohio. "Many inference workloads can instead serve separate requests across multiple smaller clusters, opening up more locations. The amount of capacity being used to serve inference is therefore expected to rise.

What happens next

By 2030, inference is projected to use 37% of that capacity, compared to just 13% for training. Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news. "Different workloads need different infrastructure, so we have conversations with a range of partners and assess opportunities based on our requirements, performance, reliability, timing and cost," they added. "We don't comment on specific commercial discussions. S. and further afield are increasingly facing pushback from local communities.

"Securing a few megawatts at an existing powered site can be more practical than waiting for a much larger block in one location," he said. "Training a large model typically requires many chips working closely together," Tan said.