Applied Digital turns to 1800s steam turbine technology to solve AI power bottleneck
When modern turbines are sold out until 2032, AI infrastructure builders are reviving century-old technology to avoid multi-year delays.
Gas turbine shortage forces alternative approach
Applied Digital, a data center operator focused on AI workloads, reported on January 11, 2026, that it has partnered with Babcock & Wilcox to deploy natural gas-fired power plants using boilers and steam turbines rather than conventional gas turbines. The decision came after the company discovered that natural gas turbines are sold out across the industry until around 2032.
The company signed a limited notice to proceed with B&W in November to deliver 1 gigawatt of power using technology that has been around for more than a century. Each installation will feature 300-megawatt power plants equipped with boilers and steam turbines.
Three to four year time advantage
Applied Digital aims to increase its capacity to 5 gigawatts over the next five years. By turning to steam turbine technology, the company expects the new power generation capacity to come online in 2028, years before standard gas turbines could be installed.
The approach allows Applied Digital to deliver AI computing capacity three or four years earlier than waiting for modern turbine availability. The company delivered its first 100 megawatts of AI computing at its Polaris Forge 1 campus and signed a 15-year lease with a U.S. hyperscaler for 200 megawatts at Polaris Forge 2.
Power generation, not GPUs, becomes the constraint
The move highlights a fundamental shift in AI infrastructure bottlenecks. While demand for AI data centers continues to grow, the limiting factor is no longer GPU availability or physical buildings, but access to power generation capacity.
Steam turbine technology, though older, offers a practical path forward when supply chains for modern equipment cannot keep pace with demand. The decision reflects how infrastructure constraints are forcing the industry to balance innovation with pragmatic deployment timelines.