The bottleneck for AI is no longer algorithms, it’s energy. Right now, we are seeing a global scramble for capacity to feed power-hungry data centers. One of the biggest roadblocks? A gas turbine supply crunch. Natural gas is being promoted as the reliable, dispatchable answer to our dwindling power supply. But turbine manufacturers can’t seem to keep up with demand. With wait times stretching years in some cases, a supply chain shortage is threatening to both slow AI’s growth and risk blackouts across the country.
Gigawatts of Turbine Demand
The shale boom of the early 2000s was the last time the U.S. saw a spike in gas-fired power plant builds. After a rise in renewables and lull in power demand growth, new construction essentially stopped. As a result, turbine manufacturers slowed production expansion. This was fine for the time, but has now created a gap between turbine supply and demand as electricity needs skyrocket.
Training and running large AI models requires a constant and reliable power supply. Solar and wind, while great for greening the grid, are intermittent. Batteries can store some power but cannot yet provide days or weeks of continuous output. This leaves gas turbines as the default and logical solution for developers who need to get projects online within a few years, not a decade or more. Due to this, we’re seeing a manufacturing backlog that has pushed turbine lead times to sometimes as much as seven years.
According to McKinsey, 120 new gas-fired plants will be constructed by 2030, accounting for 80 GW of capacity. This does not include small projects and those without completion dates, which actually brings the total capacity up to 100 GW. Major manufacturer Siemens Energy North America alone has seen 14 GW worth of turbine orders so far this fiscal year, with 65% of that coming from data centers.
What are manufacturers doing?
Today, three companies are in charge of over 70% of the world’s turbine manufacturing capacity: Mitsubishi Heavy Industries, GE Vernova, and Siemens Energy. These manufacturers are pushing hard to meet demand in light of recent growth. Mitsubishi Heavy Industries is doubling its manufacturing capacity for key components over the next two years. GE Vernova is investing $41 million into its production lines for generators used in its efficient HA-class turbines. And Siemens Energy is expanding a key US-based blade and vanes facility by 61,000 square feet to increase output.
These efforts, however, are doing little to truly help. Turbine builds require specialized precision, which limits just how fast we can expand production. Right now, these companies can really only support dozens of turbines per year. According to experts from the Electric Power Research Institute (EPRI), even if one manufacturer doubles its output, the net increase in global supply may only be 15-40%. This new capacity will be quickly absorbed by a seemingly bottomless demand pipeline, doing little to reduce immediate wait times.
Navigating Power Today
How can we best address this turbine shortage? While increasing supply is necessary, there are a few other paths that can and should be pursued in tandem:
- For Investors and Developers: Multi-year component lead times drastically increase the risk of new power generation projects. We are now seeing developers taking equity stakes in generation assets or striking exclusive deals with turbine manufacturers to ensure that their needs are met. Existing, operational plants with available capacity have become incredibly valuable, making them prime acquisition targets.
- For Policymakers: While we still need to decarbonize, there is an immediate urgency for firm capacity in order to keep the lights on. Policies that are purely restrictive on fossil fuels risk energy shortages and stalling digital economic growth. The focus must shift to enabling a bridge, perhaps by fast-tracking permits for turbines that are “hydrogen-ready”.
- For the Tech Sector: AI companies will need to directly engage with the energy sector to secure necessary gigawatts. We are now seeing many invest in advanced nuclear (SMRs) and energy storage to meet their long-term goals.
The path forward requires a new level of collaboration between the tech and energy sectors. We need to ramp up turbine supply as much as possible, but must also simultaneously pursue grid efficiency and advanced energy solutions (nuclear and storage). Meeting electricity demand will take innovation on all fronts.
