Why Natural Gas Turbines Became the Scarcest Asset in AI Infrastructure, and Why Most Brokers Can't Get You One
9/10/20263 min read
Eighteen months ago, natural gas turbines were a mature, well-understood piece of industrial equipment with predictable lead times. Today, they are the single hardest constraint in bringing new AI and Bitcoin mining power capacity online, with major OEMs quoting order-to-delivery timelines of three to four years, and in some cases longer, for large-frame units capable of supporting a serious campus.
This shift happened faster than almost anyone in the industry anticipated, and it caught most developers, brokers, and even some well-capitalized buyers genuinely flat-footed, having assumed that capital alone would always be sufficient to secure the equipment a project needed on a reasonable timeline.
What Actually Changed
Turbine manufacturers build to order, not to inventory, and their manufacturing capacity was sized for a world of steady, predictable industrial and utility demand. The simultaneous arrival of massive AI data center power demand, sustained Bitcoin mining growth, and continued conventional utility and industrial demand has overwhelmed that manufacturing capacity almost overnight in industry terms, and turbine OEMs cannot simply expand production lines fast enough to catch up, since building the manufacturing capacity itself takes years.
The result is that turbines in the range large-scale AI and mining campuses require, often ranging from tens to hundreds of megawatts of generation capacity per site, now carry price tags of $50 million to well over $100 million per unit, and buyers placing an order today are routinely quoted delivery dates years out, regardless of how much capital they're prepared to commit.
Why Most Brokers and Developers Can't Solve This
Capital is not the constraint for most serious buyers in this market. Manufacturing slot availability is, and slot availability cannot be purchased on demand, it can only be accessed through an order that was already placed, years in advance, in the position it currently holds in the manufacturer's production queue. A broker or developer without an existing order in the pipeline is, functionally, in the same position as the buyer they're trying to serve: at the back of a multi-year line, regardless of how much relationship-building or negotiation happens after the fact.
This is precisely why the ability to deliver on a large power project now depends less on financial engineering or general industry relationships and more on a very specific, narrow form of foresight: having placed standing equipment orders before the current demand wave made doing so this difficult and this expensive.
What This Means for Buyers Evaluating Options Right Now
Buyers evaluating a large power project today should ask any prospective partner a direct, specific question: is the turbine capacity being offered from an existing, already-placed order, or would a new order need to be placed to fulfill this project. The answer to that single question is often the difference between a project that reaches power-on within a reasonable development timeline and one that sits waiting for equipment for several years, regardless of how the rest of the deal is structured or how well-capitalized the buyer is.
The Ripple Effect on Substations and EPC Capacity Too
Turbine scarcity is the most visible constraint, but it isn't the only one. Substation equipment, particularly large power transformers, faces a similar dynamic, with lead times that have stretched considerably as the same wave of demand hits transformer manufacturers alongside turbine OEMs. EPC firms capable of engineering and building at this scale are similarly stretched, since the number of firms with genuine experience executing gigawatt-class power infrastructure projects hasn't grown nearly as fast as demand for their services has. A buyer solving the turbine problem without also confirming substation equipment availability and EPC capacity is only solving one part of a three-part bottleneck.
Why This Bottleneck Isn't Resolving Anytime Soon
Manufacturing capacity for turbines and large transformers takes years to expand, requiring new production facilities, expanded supply chains, and skilled labor that doesn't materialize quickly even with substantial capital investment. Meanwhile, demand from AI infrastructure shows no clear signs of plateauing, and Bitcoin mining operations continue seeking new sites with favorable power economics regardless of the broader equipment scarcity. The realistic expectation for buyers is that this bottleneck persists for the next several years at minimum, which makes existing equipment allocation, rather than capital alone, the determining factor in which projects actually reach power-on and which remain stuck in a multi-year queue.
Positioning for a Market That Isn't Loosening Soon
Buyers who accept this reality early tend to make better decisions than those who keep assuming the equipment market will loosen up before their project needs to break ground. That means prioritizing partners with demonstrated, verifiable access to existing equipment allocation over partners offering the most attractive price on paper for capacity that still needs to be sourced fresh. In a market this constrained, the ability to actually deliver on a stated timeline has become more valuable than marginal differences in price, since a lower price on equipment that arrives three years late is, in practical terms, worth considerably less than a fair price on equipment that arrives on schedule.
Sigma Technology Consulting, Inc.
25 Years of Experience, Vetting & Procuring Technology Vendors
Contact Us
Support
© 2026. All rights reserved.
