The Data Center Industry Is Now Chasing Power, and It Will Locate Facilities in Suboptimal Locations

Unlike latency-sensitive workloads or ones beholden to regulation, crypto doesn’t care where it is mined. This meant that the sector ignored most other location factors, instead giving primacy to power

Data center company Crusoe Energy Systems is expanding its footprint rapidly, adding new AI training data center capacity in response to surging demand.

The company launched in 2018 as a cryptominer deploying containerized data centers to oil wells to harness natural gas that would otherwise be "flared off" and wasted.

Now the company is applying its techniques to support other latency-insensitive workloads, such as AI training and inference.

"We're really doing this in an energy-first way where we're going to places that have a lot of overdeveloped or excess, otherwise curtailed, power,” said Crusoe Energy VP of Business Development Chase Cavness. “And sometimes these are in remote areas. We're taking that same ethos into the AI development space."

Cruesone still has some AI capacity being powered by its flared gas operations - such as in Montana, the Bakken oil field - and is quickly trying to bring on as much capacity as it can.