The past few years have reshaped the global gas turbine market in ways few expected. As one of the pillars of global power generation, gas remains, alongside wind and solar, the only technology of meaningful scale still growing worldwide. What began as a steady recovery has become one of the most dramatic backlog extensions the industry has ever seen. Lead times for large turbines stretch into the 2030s, and pricing for the largest machines has increased by close to 300 percent. AI driven load growth, rising baseload requirements in the Middle East and Asia, and Europe’s ongoing need for dispatchable power have created a demand environment that overwhelms supply.
Today, the industry can supply roughly 50-60GW of gas turbines per year, while orders last year approached 88GW. Capacity expansions have been announced but may only take supply to around 100GW by 2030, or 120-125GW if we include aeroderivative turbines, gas engines and fuel cells. These estimates are uncertain, but the conclusion is clear; demand for the largest and most efficient machines cannot be met. As a result, many data centre developers unable to secure large turbines are installing less efficient reciprocating engines or small turbine systems as interim or stop-gap solutions.
A rapid build out is underway in these smaller turbine and engine segments to meet this demand. Anyone with a credible power solution is targeting the data centre market, especially in the US. This contrasts sharply with the market for large H-class turbines (300-400MW and above), where the supplier base is highly concentrated, visibility is better, and undersupply is likely to persist. Both markets cater to the same underlying demand, but they are fundamentally different in terms of product, competitive dynamics and ultimately where pricing power resides.
Understanding where pricing power sits is central to our investment process. We start with our top-down industry analysis, focusing on supply and demand and bottlenecks, before turning to bottom-up company analysis. We believe bottom-up analysis alone is rarely sufficient. Directional demand offers limited insight into cycle turning points within specific product categories unless the supply side is properly understood. This framework underpins our conviction in companies like Siemens Energy (ENR), even after nearly a 300% share price appreciation since our initial purchase.
The turbine market spans a wide spectrum. At one end are the large F and H class machines that in combined-cycle configuration can produce more than 800MW using waste heat via a steam turbine. These sit at the top of the technology stack and are dominated by three companies: Siemens Energy, GE Vernova and Mitsubishi. All three are expanding capacity, but at a measured pace and on increasingly attractive pricing, at least for now.
At the other end is a fast-growing ecosystem of smaller power solutions: small gas turbines, reciprocating engines, fuel cells, aero derivatives, hybrids and mobile generator sets. These are increasingly deployed “behind the meter” (BTM) by data centres waiting for grid connections, a topic we discussed in our October 2025 report “Bring your own power, how the data centre is rewiring the grid”. In this segment, barriers to entry are lower, engineering cycles shorter and systems modular enough to scale quickly.
Some examples, but the list is far from exhaustive:
Taken together, this ecosystem is becoming crowded, with existing players and new entrants all scaling aggressively to target the same BTM data centre market.
CAN DEMAND ABSORB ALL THIS SUPPLY?
Estimates for installed US data centre demand in 2030 vary widely from around 75GW (BNEF) to 130GW (BCG), up from roughly 45GW of installed capacity today. The more conservative estimates focus on interconnection queues and assume BTM solutions will remain a niche, given the complexity and cost of operating a parallel power system 24/7. We recognise these challenges, but given hyperscalers’ substantial capex commitments and the high value they place on time to power, we find it likely that BTM solutions will play a role, at least as interim capacity until large gas turbines and grid connections are available. We therefore run different scenarios for the share of new data centres going BTM and track developments closely, as this has important implications for the evolving supply picture.
For illustration, assume 15GW of BTM data centre capacity is added by 2030. Including 30% redundancy, this implies a total addressable market for smaller turbines, engines and fuel cells of roughly 4GW per year. Bloom Energy (BE) alone could soon cover half of that, or more. Even if we double our BTM demand estimate, the capacity additions outlined above suggest that supply is likely to overwhelm demand for smaller power units, whether they are turbine or engines, by 2030.
It remains unclear whether these smaller solutions will become permanent power sources or primarily serve as stop gap measures until grid connections are secured. Our base case is that grid connection remains the preferred option wherever possible, which makes us cautious about taking too much exposure to this segment of the “Powering AI” market.
Instead, we prefer Siemens Energy (ENR) and its larger turbines, which are essential for grid build-out. The end markets are more diversified, in fact in 2025, only about 20-25% of large turbine orders were related to data centers. With three players controlling an estimated 80-90% of this market, we foresee a longer-than-normal cycle for this highly profitable sub-segment of the gas turbine industry. We also like EPCs such as Mastec (MTZ), which should benefit from both grid expansion and from connecting BTM data centres to the gas network.
In our September-2025 report, “Are we in an AI bubble”, we wrote that “overbuilt infrastructure is historically rather a feature, not a bug of transformative infrastructure spending cycles.” Timing such a bubble is incredibly difficult but understanding the underlying cycles is less so. That is why our investment process begins where it does; with understanding the cycle, the bottlenecks and the industry structure. This philosophy has guided our strategy for more than a decade. While we still prefer exposure to the larger turbines and believe this cycle has further to run, we are acutely aware that conditions in cyclical industries can change quickly, and we are continuously monitoring this dynamic.