Opinion | How jet engines can help power the global AI economy
Using old engines to provide energy for data centres runs counter to climate commitments, but the proven technology is easier to scale than green energy
Artificial intelligence (AI) runs on electricity. Sometimes that electricity might come from a jet engine. FTAI Aviation, a New York-based aircraft engine lessor and maintenance provider valued at more than US$25 billion, plans to convert roughly 100 retired CFM56 engines a year into turbines producing about 25 megawatts each.
The CFM56 is one of the most widely used jet engines, with more than 30,000 delivered to date, powering the Boeing 737 and Airbus A320 families. Having exhausted their aerospace economic life, those engines would otherwise head for the scrapyard. The company’s share price rose after it unveiled the FTAI Power venture in late 2025.
The proposal is not as eccentric as it sounds. Global data centre electricity demand is projected to approach 1,000 terawatt-hours per year by the end of the decade, roughly comparable to Japan’s annual power consumption. AI workloads are expanding rapidly, while electricity grid expansion lags behind demand. Training modern AI systems and running the specialised chips behind them require vast amounts of electricity.
Advertisement
In many developed markets, grid connections now face years-long waiting lists. This constraint is increasingly felt across the Asia-Pacific, where Singapore imposed a moratorium on new data centre construction in 2019 and now tightly rations new capacity. The gap is structural, and capital is now moving to exploit it.
In that context, ageing aviation hardware acquires unexpected value. FTAI has found a way to reclassify industrial assets nearing the end of their useful life as critical AI infrastructure. Similar efforts are under way elsewhere. ProEnergy, a Missouri-based global power generation company, has been converting retired widebody aircraft engines such as the General Electric CF6 into ground turbines for data centres. These engines have already sold more than 1 gigawatt of generation capacity to data centre projects. These early moves suggest a pattern rather than isolated opportunism.
Advertisement
The logic is simple, though not entirely virtuous. Renewable energy and new nuclear capacity require many years to develop and scale. A jet engine, a proven and modular technology, can be converted in a matter of weeks. It can be installed in clusters, ramped up rapidly to meet sudden spikes in demand and positioned close to data centres where grid capacity is scarce.
China trials underwater data centres to solve energy woes
China trials underwater data centres to solve energy woes
