Invest1 publisherNot yet confirmed elsewhere2 min readPublished
GE Vernova's 116 GW backlog outruns global turbine output, with slots quoted into 2031
The gas power book grew 16 GW in a single quarter. Annualised, one supplier's order intake matches what every turbine plant on earth can build in a year.
The Investor · Invest desk
What happened
- GE Vernova's gas power equipment backlog reached 116 GW in Q2 2026, up from 100 GW the quarter before.
- The company is quoting turbine delivery slots as far out as 2031.
- Total global turbine manufacturing capacity is 60 to 70 GW a year.
- Siemens Energy sold 100 gas turbines in 2024 and 194 in 2025, with about 60% of recent orders tied to data center projects.
- PJM's July 2026 capacity auction fell 6,831 MW short of its reliability target, the third consecutive miss.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Behind-the-meter generation does not escape the queue. Private plants draw on the same turbine order book, so going around the grid buys siting freedom, not an earlier date.
- cost Doubled combined-cycle capex is now embedded in power contracts signed today for compute sold at the end of the decade, and the buyer of that compute pays it.
- decision Anyone holding a 2028 in-service date has to choose: pay to reserve a slot that lands years later, or redesign the site around generation and interconnection that already exists.
- contradiction An industry order total below the backlog of a single manufacturer means the published aggregate understates the queue, and capacity plans benchmarked to it are anchored to the wrong number.
Annualise the quarter and the intake runs at roughly 64 GW a year [17], against a global turbine industry able to build 60 to 70 GW [11]. One manufacturer, in three months, booked orders at the pace the whole world produces. The stock number is worse than the flow: a 116 GW book is between 1.7 and 1.9 times global annual output [18].
The reported industry aggregate does not close. The cryptobriefing account puts total orders above 110 GW [12], which is smaller than GE Vernova's book on its own [2]. Either that figure counts a narrower slice or it is stale. Treat any industry-wide order total in this market as a floor.
The demand side has the same arithmetic problem. Goldman Sachs has US data center power demand going from 31 GW in 2025 to about 66 GW by 2027 [4], and separately forecasts US capacity additions of 13.6 GW in 2026 and 36.3 GW in 2027 [5]. Those additions sum to 49.9 GW, about 15 GW more than the demand growth implied by the first pair [8]. Take the additions literally and 2027 alone would call for between 52% and 61% of world turbine output [19], in a market where Siemens Energy traces about 60% of its recent orders to data centers [6] and Gulf buyers are queued alongside [13].
Not every data center gigawatt is fed by a new turbine, which is the only reason those numbers are survivable. But the alternatives are thinnest where the grid is tightest. PJM's July 2026 auction cleared 6,831 MW below its reliability target [1], a gap equal to roughly half of Goldman's entire 2026 US build [9].
Price is behaving the way it does when the scarce asset is a position rather than a product. Wood Mackenzie puts turbine cost inflation above 195% since 2019 [7], close to three times the 2019 price [10], and combined-cycle plant capex has roughly doubled [16]. Manufacturers now charge reservation fees simply to hold a place in line [15]. That is the part worth pricing. What is being sold is queue position, years ahead of any concrete, and a quoted slot buys delivery of a machine rather than a commissioned plant behind it [3]. It rations by balance sheet. Whoever can write a fee cheque in 2026 against equipment arriving in 2031 has real 2029 capacity; whoever cannot has a slide.
What to watch
- Whether GE Vernova's next quarterly report adds another 16 GW to the book or shows intake slowing.
- Whether PJM's next capacity auction closes the 6,831 MW gap or misses for a fourth consecutive year.
- Whether Goldman revises the 36.3 GW 2027 US addition figure down toward what turbine output can physically support.