It’s not news that natural gas turbine manufacturers are struggling to keep up with demand. Headlines have called out the enormous projected energy needs of new data centers, especially those supporting artificial intelligence (A.I.), for at least the last year. Utilities and tech companies see natural gas as the best near-term solution for meeting those needs. However, a new report outlines just how quickly market forces are driving up costs for turbine customers.

It’s a tricky situation for suppliers and power developers as well. They are boosting capacity to meet new orders, but these investments have a big potential downside. If reality doesn’t live up to ambitious market projections, both groups could be on the hook, financially.

From green to gas

Just a few years ago, Google, Meta and other internet leaders were touting their renewable energy investments, with many reaching toward running their systems on 100% green electricity. That emphasis has shifted in the two years since OpenAI’s ChatGPT debuted, kicking off a technology arms race, with each advance in chip performance coming with another step up in power needs.

As a result, utilities—and data center owners, themselves—have reduced their wind and solar emphasis in favor of natural gas-driven equipment as a preferred power source. In addition to providing stable output regardless of the weather, turbines also can ramp up and down as needed to address A.I. facilities’ wide demand swings. 

The rapid shift to A.I. has technology companies pouring billions into building sprawling server campuses, some with gigawatt-­level power-demand projections. And with transmission connections also coming with years-long waits, corporate buyers now are building their own on-site power plants, creating even more competition for limited turbine supplies. 

Turbine costs have skyrocketed. An April report from the energy consultancy Wood Mackenzie predicts per-kilowatt (kW) costs at the end of 2027 will be triple those seen in 2019. And, with turbine costs making up an estimated 20% to 30%—or higher—of total power plant development expenses, these higher prices have big implications for utility bills for years to come.

According to the report, “The U.S. Gas Turbine Market: Navigating Manufacturing Scarcity and Demand Growth,” global turbine orders totaled 110 GW at the end of 2025, while manufacturing capacity can only manage 60–70 GW annually. This isn’t a situation in which turbine makers can simply add another shift or two to raise their production.

The process of fabricating turbine components, especially those exposed to the highest temperatures, is a highly specialized, hands-on task, and there’s currently a shortage of craftspeople capable of the job. And the manufacturing operations are expensive, which is one reason why only three companies dominate the U.S. market. All are scrambling to expand. 

GE Vernova is now in the midst of a $160 million plan to increase production from approximately 50 large-frame machines annually to 70–80 units by the end of this year. Siemens Energy, Orlando, Fla., is now running its facilities 24/7 and announced a $1 billion investment plan for even greater growth. And Mitsubishi Heavy Industries, Houston, plans to double its capacity through 2028.

Prediction risks

Even with the market currently booming, turbine companies face a risk of oversupply if today’s bullish data center growth models don’t pan out in the long term. An April Bloomberg report found that between a third and a half of all U.S. data centers planned for 2026 will likely be delayed or canceled. Of the estimated 12–16 GW now on the boards, only 5 GW is currently under construction. 

The turbine shortage itself is a leading factor causing this disparity. But there’s also a feedback loop as chip designers continue to leap-frog ahead in performance and efficiency with each new product generation. This can lead developers to rethink the platforms on which their projects should be based.

Additionally, the sheer cost of A.I. data center construction is beginning to catch up with leaders like OpenAI and Oracle. The companies recently pulled back from plans to expand their signature Stargate effort in Abilene, Texas, to 2 GW of capacity from its current 1.2 GW due to a collapse in financing negotiations with the project’s developer. 

Newer projects could have a similar level of uncertainty underlying ambitious ribbon-cutting announcements. In March, plans were announced for what would be the world’s largest gas-fired generating facility, with a 9.2 GW capacity and costing $33 billion. The plant would be dedicated to supporting a 10-GW data center on U.S. Department of Energy land in Piketon, Ohio. However, developer SB Energy’s initial plans call for an 800-megawatt facility, with further buildout yet to be announced.

For turbine manufacturers, this roller coaster environment is making planning difficult. With new turbine orders now facing 5- to 7-year lead times, they’re having to make long-term capital investment decisions in a tech landscape that’s shifting every 18 months. 

STOCK.ADOBE.COM/ chakawut