Field notes The Energy Transition for the Rest of Us

Catalyst N° 065 of 125 5 Jun 2025

The gas turbine crunch

with Anthony (Tony) Brough, president, Dora Partners

In this note
  1. 01The question
  2. 02The answer
  3. 03The argument
  4. 04What you need to know first
  5. 05Details worth keeping
  6. 06Claims worth citing
  7. 07Where it’s contested

The question

Gas turbines are suddenly impossible to get and increasingly expensive. What is actually causing the bottleneck?

The answer

Mostly not the power sector. Roughly half of all industrial gas turbines are sold into oil and gas rather than electricity, and the raw materials and cast components come from the same suppliers feeding an aerospace industry with tens of thousands of aircraft on backlog. The electricity crunch is real, but it is being squeezed by neighbors in industries most energy people never look at.

03The argument

Turbines have always been a boom-and-bust business, and the most recent bust left scars. The 1998-2001 bubble was driven substantially by Enron’s artificial market signals arriving alongside deregulation, and by Brough’s estimate at least half that order volume was never real. Orders were cancelled, including by large regulated utilities, and everyone involved got burned. So the natural question is whether manufacturers are now too gun-shy to expand. The answer he gives is “guarded optimism”: they are investing, carefully.

What makes them willing is that this boom rests on five demand drivers at once rather than one. Grid-scale battery storage, coal plant retirements, grid-scale renewables, data centers and AI, and cheap abundant natural gas are all pulling simultaneously. Brough contrasts this directly with his own experience as an executive at a major manufacturer twenty years ago, when the strategy depended on a single driver that failed to materialize. Diversified demand is what converts this from a bubble they should fear into a market they can build toward.

The supply chain explanation is the part worth carrying. He splits it into four levels: raw materials such as nickel-based superalloys, chromium and titanium; cast and forged components like blades and vanes; the manufacturer assembling the turbine itself; and final packaging, installation and commissioning. The first two levels are shared with aerospace, where roughly 40,000 aircraft sit in backlog, which represents at least 80,000 jet engines drawing on the same foundries and casting shops. Add that about half of industrial gas turbines go to oil and gas, and the conclusion follows: you cannot understand the power sector’s turbine problem by studying the power sector.

The last move is to stop treating “gas turbines” as one market. Small units under 20 megawatts, mid-size 20 to 100 megawatt machines, and jumbo units of 150 to 250 megawatts and up respond to the same driver in opposite directions. Grid-scale batteries are the clearest case. Intuitively, storage should destroy demand for gas, and for jumbo units it does. But for mid-size units it is a positive, because developers use gas to recharge batteries in hybrid configurations rather than relying on renewables alone. The same inversion shows up with renewables expansion, which hurts jumbo units while creating demand for peaking and mobile machines to cover when the sun sets.

04What you need to know first

OEM
The original equipment manufacturer, meaning the company that actually builds the turbine. Three dominate: Mitsubishi Heavy Industries, Siemens, and GE Vernova, with Caterpillar’s Solar Turbines significant in small units.
Combined cycle
A plant that captures waste heat from the gas turbine to drive a second steam cycle, which is why efficiency numbers near 60% are possible at all.
Levelized cost of electricity
Lifetime cost divided by lifetime output. Fuel is a large share for gas, which is why small efficiency gains translate into real competitive advantage between manufacturers.
Bridge power
Temporary on-site generation covering the years between when a data center wants to operate and when the grid connection actually arrives.

05Details worth keeping

  • Mobile turbines are not only a data center bridge-power story. Utilities buy them, station them in grid-constrained areas for five to ten years while upgrading infrastructure, then relocate them to the next constrained spot.
  • Efficiency has been a slow grind with real payoff: combined cycle plants moved from roughly 55% average efficiency to pushing 62%, with progress stalling near 60% until manufacturers started optimizing the whole plant rather than the turbine alone.
  • Every manufacturer is developing hydrogen-capable combustion and all have demonstrated some capability. Brough is openly doubtful that this converts into meaningful purchases, for the unglamorous reason that nobody can say where the hydrogen for a jumbo unit would come from. Kann characterizes the strategy as build it and they will come.
  • Raw material producer price indices show a marked increase over the three years prior, before tariffs entered the picture.
  • Oil and gas turbine demand is steadier than oil prices suggest, because midstream operators earn on throughput rather than commodity price, and majors plan on five-to-ten year strategies through roughly seven-year cycles.

06Claims worth citing

All as stated on 2025-06-05. Lead times and prices in this market move fast.

  • Lead times of 36 to 48 months typical, with some manufacturers quoting up to 60. Brough does not expect them to get much worse. Brough
  • Turbine prices up roughly 30-35% over five years. Brough
  • Customers putting down non-refundable deposits of 15% to 25%. Brough
  • About 50% of industrial gas turbines delivered in a given year go to oil and gas rather than electric power. Brough
  • Roughly 40,000 aircraft in global backlog, implying at least 80,000 turbines competing for the same upstream supply chain. Brough
  • Roughly 11,000 data centers worldwide serving digital commerce and AI, averaging about 4 megawatts of load; around 1,400 new ones planned in the US, over 1,000 of them large scale. Brough
  • Grid-scale renewable deployment expected to double over five years. Brough
  • At least half the 1998-2001 order volume was artificial. Brough

07Where it’s contested

  • The widely-quoted price figure gets pushed back on. Kann cites a NextEra number of roughly $750 per kilowatt a decade ago against about $2,500 today. Brough says the $750 referred to a full combined cycle plant and calls $2,500 “a bit aggressive,” offering 30-35% over five years instead. A useful caution: this is a number in wide circulation that the specialist partly disputes.
  • Whether lead times have peaked. Brough says manufacturers are working hard to flatten them and he sees signs of it. That is a forecast from someone serving this market, and the customer-patience argument behind it is reasonable rather than proven.
  • How much of data center demand is real. Kann states flatly that he does not believe 1,000 new hyperscale data centers will be built in the US anytime soon, and frames it as two things being true at once: genuine turbine demand alongside speculative development, with “cowboys” in the market. Brough is more sanguine and emphasizes that AI and digital commerce are not going away. The non-refundable deposit is the mechanism for telling real buyers from cowboys.
  • Storage as a gas killer. The intuitive answer is wrong for mid-size units, which benefit from hybrid recharging configurations. Worth remembering whenever someone asserts a technology straightforwardly displaces another.

Cite as: “The gas turbine crunch,” The Energy Transition for the Rest of Us, note on Catalyst with Shayle Kann, June 5, 2025. CC BY 4.0. View the Markdown