Technology news and discussion on TalkFuse.
Analysis:AI power rush sends data centres hunting for used turbines and jet engines
Data centre developers and energy firms are searching globally for used, refurbished and repurposed turbines to power the fast-growing AI sector, as new equipment orders face long delays. The rush shows how demand for AI computing is straining power infrastructure, though second-hand equipment is unlikely to solve much of the supply shortfall.
Global turbine orders exceed 100 gigawatts (GW), while current annual manufacturing capacity is estimated at 60 to 70 GW. US turbine delivery takes six years on average, with some new orders scheduled for 2030 or 2031; meanwhile, used units can cost about $1.5 million to $1.7 million per megawatt. Age, maintenance history and suitability for continuous operation are among the risks, and grid access and permits can also delay projects.
In short
- AI data centre demand is outstripping turbine supply.
- Buyers are turning to used and refurbished machines.
- Secondary equipment cannot resolve the wider power bottleneck.
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Repurposing used turbines and jet engines could bring some power online sooner while manufacturers face years-long queues, helping data centre developers meet demand and potentially easing pressure on constrained grids. Supporters may see this as a pragmatic use of existing equipment, provided its condition, emissions and suitability for continuous operation are carefully assessed.
The case against
A rush to buy ageing equipment could divert attention and capital from durable, lower-emission power and grid improvements, while maintenance risks and permitting delays may limit how much electricity these units actually provide. Critics may also argue that powering more AI capacity this way entrenches energy-intensive growth, and that second-hand turbines are too scarce to make a meaningful difference to the wider shortfall.