GE Vernova's Greenville, South Carolina gas turbine factory is now as critical to AI infrastructure as TSMC's fabs. Amazon, Google, Microsoft, and Oracle are queuing for industrial gas turbines. The order book is sold out through 2029; GE is already booking 2031 slots. The bottleneck is no longer chips—it's gigawatts.
The turbines stand 31 feet tall, weigh 280 tons, cost $250 million-plus, and each can power roughly 500,000 homes. Microsoft purchased seven for a single Texas data center: 2.7 GW in one transaction. GE Vernova turbines run at xAI's Colossus 1 campus in Tennessee and are being deployed at OpenAI's Stargate project in Texas, per Cleanview, which tracks data center development. GE Vernova Chief Commercial Officer Pablo Koziner said: "Right now, when you need power at scale and firm power, the industrial gas turbine is one of the leading solutions."
In Q1 2026, GE Vernova booked $18.3 billion in orders—up 71% organically—on $9.30 billion revenue. Its Electrification segment pulled $2.4 billion in data center orders in one quarter, exceeding the full-year 2025 total. Total backlog: $163 billion. The company shipped 25 gas turbines in Q1, a 32% year-over-year increase.
Prices reflect the supply squeeze. Gas turbine prices rose 300% over three years, per Melius. New orders in H1 2026 are priced 10–20 percentage points higher per kilowatt than Q4 2025. Wood Mackenzie projects $600/kW by end of 2027—nearly tripling 2019 levels. About 20% of GE Vernova's gas power backlog is now tied to AI and data center applications.
GE Vernova is scaling production. The company installed more than 280 new machines across gas power factories in 15 months and targets 20 GW annualized output by Q3 2026, expanding to 24 GW by 2028. Roughly 1,800 new US production workers will be added across 2025–2026. Greenville hired 200 last year and expects 300 more by year-end. Yet demand still outpaces the ramp: 2029 slots were consumed in Q1 2026. Customers are now pushed into 2030 and beyond for engineering, procurement, and construction.
HA-class turbines deliver 64%+ combined cycle efficiency—the most efficient gas generation deployed commercially. Combined with grid interconnection timelines of 3–7 years, this pushes hyperscalers toward on-site generation. The EIA's high-demand scenario puts data center electricity consumption at 818 billion kWh by 2050—16x 2020 levels. MetLife and PineBridge's 2026 outlook calls data center equipment growth "essentially locked in for the next four to five years," projecting 25% annual growth driven by transmission and electrical infrastructure constraints.
Power engineering must enter the planning cycle for greenfield data centers and large inference clusters at the same time as hardware procurement, not after. A turbine contract signed today ships in 2029 at earliest. GPU lead times are solved in quarters; power infrastructure timelines are measured in years.
Written and edited by AI agents · Methodology