The gas turbine supercycle — and why it makes every installed megawatt precious

Demand shock: the numbers

The natural gas generation business is being reshaped by a demand shock with no modern precedent. The U.S. EIA nearly tripled its projection of new gas-fired capacity through 2030 — from 23 GW to 66 GW — based on what generator owners are actually reporting. Industry trackers count over 140 gas-fired projects worth roughly $136 billion in the U.S. pipeline. S&P Global projects U.S. data-center demand rising from 76 GW in 2026 to 134 GW by 2030; Bank of America forecasts more than 230 GW of new generating capacity needed in five years against only ~93 GW of planned utility additions — a gap of over 100 GW that gas is being asked to fill.

Unlike the dot-com turbine boom, this demand is broad and durable: AI and data-center load, industrial electrification, coal retirements and grid-firming for renewables, all pulling simultaneously. Manufacturers have responded by raising their view of the sustained annual turbine market from ~90–100 GW to 110–120 GW per year — and they still cannot build fast enough.

Supply reality: sold out for years

Three manufacturers — GE Vernova, Siemens Energy, Mitsubishi Power — build effectively all large-frame turbines, and all three are booked into the next decade. GE Vernova ended Q2 2026 with a 116 GW backlog and is contracting 2031 delivery slots now. Siemens Energy holds 69 GW firm. Large-frame lead times run four to seven years; aeroderivatives — historically the fast option — are out 18–36 months. Prices have risen ~50% in months, and all-in new combined-cycle costs have more than doubled since 2020. Developers are responding by buying existing plants at roughly half of replacement cost, deploying reciprocating-engine fleets they can actually get, and building behind-the-meter generation — over 100 GW of on-site gas capacity has been announced by data-center developers alone, with 7.5 GW already under construction.

What a supply-constrained market means

In a demand-constrained market, an inefficient machine is a fuel-cost problem. In a supply-constrained market, it is a capacity problem — and capacity is the scarcest commodity in the industry. Three consequences follow directly:

• Every installed machine must produce everything it can. The existing gas fleet is running at higher capacity factors, covering retiring coal and unserved data-center load. Conservative fleet calibrations, stale ambient corrections and unmanaged degradation now cost megawatts the market cannot replace at any price before 2030.
• Every arriving machine will be commissioned fast, somewhere hard. The buildout is pushing turbines behind the meter at data centers, onto fast-tracked brownfield sites, into extreme climates and onto whatever gas the local system delivers — including rich shale supplies, RNG blends and field gas. These are precisely the conditions under which a generic fleet calibration gives up the most.
• Every point of efficiency is found capacity. One absolute point of efficiency is ~3% less fuel per kWh — or equivalently, ~3% more energy from the same fuel supply. Across a fleet, calibration recovers megawatts with no interconnection queue, no EPC contract and no 2031 delivery slot.

The case for site-specific calibration, in one paragraph

The machines are unobtainable, the replacements cost $3,000/kW, the fuel is increasingly variable, the sites are increasingly extreme, and the duty cycles look nothing like the ISO certification points. Every factor that made factory calibrations "good enough" in a buyer's market has inverted. Characterizing each installation's fuel chemistry and atmospheric envelope, and calibrating the control strategy to that reality — the discipline we practice — is now the highest-leverage, fastest-payback intervention available on a gas generation asset.

Market signal (2026)	Number	Calibration consequence
GE Vernova turbine backlog	116 GW → 2031	Installed efficiency is the only capacity available now
Siemens Energy firm backlog	69 GW	No relief from the second supplier either
New turbine pricing (EPRI)	~$3,000/kW	Each recovered kW is worth ~$3,000 of avoided capex
EIA U.S. gas buildout to 2030	66 GW (3× prior)	A wave of new sites needing commissioning-grade calibration
Data-center demand (S&P)	76 → 134 GW by 2030	24/7 loads with zero tolerance for derates and trips
Behind-the-meter gas announced	100+ GW	Non-ideal sites, variable gas, generic calibrations — our exact problem statement

Sources: GE Vernova and Siemens Energy quarterly reporting, EPRI, U.S. EIA, S&P Global, Bank of America research, industry press — mid-2026.