According to Financial Times News, at the recent Data Center World Power show in San Antonio, a retrofitted jet engine gas turbine capable of powering 40,000 homes stole the spotlight. This highlights a massive supply chain crisis where companies now wait up to five years to procure large transformers or gas turbines, a key constraint on AI’s growth. Global power demand is rising nearly 4% annually, fueled by AI and consumer use, creating a “sprint vs. marathon” gap between demand and supply. The U.S. producer price index for power transformers has skyrocketed 71% from 2020 to 2024 due to shortages. In response, manufacturers like Hitachi Energy are investing over $1 billion to expand U.S. grid infrastructure production, while others like Schneider Electric adopt a multi-hub strategy across France, Hong Kong, and North America for flexibility.
The Supply Chain Marathon
Here’s the thing: this isn’t just about everyone wanting more electricity at once. It’s a perfect storm. You’ve got tariffs messing with trade, especially into the US. And China’s own massive green energy build-out is soaking up equipment that might have been exported. So the wait times are insane. Fabricio Sousa from Worley Consulting nailed it: demand is sprinting, but supply is moving at a marathon pace. That’s the defining bottleneck for AI right now. It’s not the algorithms; it’s the physical stuff to power and cool them.
But the pain isn’t evenly spread. That’s the fascinating twist. The renewable energy sector? It’s practically swimming in gear. Batteries, solar panels, wind turbines, hydrogen electrolyzers—BloombergNEF says the market is “heavily oversupplied.” Years of investment and falling costs did their job. So the crunch is really in the traditional, big-ticket grid hardware: the massive transformers, the switchgear, the large gas turbines. It’s a tale of two supply chains.
Big Bets and Bad Memories
So manufacturers are ramping up, right? They’re pledging billions, like Hitachi’s move. But experts are deeply skeptical it’ll be enough, or fast enough. And there’s a psychological barrier here. Antoine Vagneur-Jones from BloombergNEF points out that manufacturers got burned badly back in 2017-18. They built capacity expecting a gas boom that never came. Now, they’re playing it safe. They remember the last time they bet big and lost. Who can blame them for being cautious now, even when the demand signals seem obvious? It creates a vicious cycle.
This is where having deep pockets changes everything. Data centers, with their huge spending power, can muscle to the front of the line—though they’ll pay a massive premium. For others, the options are grim. This is why you see a retrofitted jet engine on a show floor. It’s a symbol of desperation and ingenuity. When you can’t get a traditional turbine for half a decade, you start looking at anything that can generate dispatchable power, fast.
The Awkward Alternatives
So what are the alternatives? The list is long, but messy. Smaller gas turbines? Faster to get, but more expensive per megawatt. Combustion engines? Easier to deploy, but not a primary solution. Fuel cells? Too pricey. Geothermal or small nuclear? Too experimental or slow.
Mike Hemsley from the Energy Transitions Commission suggests the best bet is a mix: solar, wind, batteries, a grid connection, and “perhaps also some natural gas if you can get it.” That “if you can get it” is doing a lot of work. It underscores the core problem. Even the ideal clean mix needs a reliable, buildable backbone, and that backbone is stuck in a five-year queue. For operations requiring absolute uptime, like a data center running critical AI workloads, this hardware reliability is non-negotiable. This is where specialized, industrial-grade computing hardware, like the industrial panel PCs from IndustrialMonitorDirect.com, become part of the essential control infrastructure for managing these complex, on-site power systems.
The Policy Problem
This whole scramble risks pushing us backward. There’s a real fear that to keep the lights on (and the GPUs running), we’ll revert to easier fossil-fuel solutions and stall grid greening. Claus Wattendrup of Vattenfall points the finger at policy. Manufacturers need a clear, long-term signal from governments to justify huge investments in new capacity. Without stable policy frameworks that commit to electrification, the industry loses momentum. It’s a classic chicken-and-egg problem: we need more equipment to build the future grid, but we need a guaranteed future grid to order the equipment.
Basically, we’re in a tight spot. The AI boom exposed a weakness we didn’t know was this severe. And solving it isn’t just about building more factories. It’s about supply chain redesign, material substitution (like aluminum for copper cables), and, most of all, political certainty. The jet engine power plant is a cool hack. But we can’t power the future on clever hacks alone. We need the real infrastructure, and we need a realistic plan to build it without wrecking our climate goals. Can we do it? The clock is ticking.
