Power Hungry Tech Giants Turn to Jet Engines Amid AI Energy Crunch

Soaring demand for artificial intelligence is dramatically reshaping the energy landscape in the United States, with leading technology companies forced to seek unconventional ways to feed the relentless appetite of data centres. As the scale and intensity of AI workloads grow, the demand for electricity is outpacing the capacity of the country’s ageing and congested power grid. The solution for many Silicon Valley giants has been a shift to on-site generation powered by gas turbines, with some even resorting to repurposed jet engines.

Recent estimates from the International Energy Agency illustrate the magnitude of this shift. AI data centres in the US consumed just under 200 terawatt-hours of electricity last year, a figure projected to triple to 600 terawatt-hours by 2035. That jump will take the sector’s share of total American electricity consumption from 4 percent to more than 10 percent. Conventional grid upgrades cannot accommodate this surge swiftly enough, compelling firms such as xAI, OpenAI, Google, and Amazon to secure alternative power sources that can be deployed quickly, often off-grid.

This urgency has precipitated a scramble for gas turbines. However, with manufacturing capacity constrained—two-thirds of the world’s gas turbines are produced by just three major companies, GE Vernova, Siemens Energy, and Mitsubishi—orders have reached their highest levels since 2011, with backlogs stretching five years or more. Faced with these bottlenecks, technology companies have increasingly turned to aeroderivative turbines, essentially modified jet engines adapted for stationary power generation.

For example, OpenAI has procured 29 GE Vernova LM2500XPRESS aeroderivative turbines for its imminent Stargate data centre in Texas, amounting to nearly one gigawatt of capacity, a figure that rivals a typical nuclear reactor. Boom, the supersonic jet developer, has also entered this market, unveiling a 42 megawatt Symphony engine and securing a £940 million order backlog for data centre installations. The company acknowledges heavy demand and has plans to construct a high-output factory to increase turbine production.

While quick to deploy, these jet-based turbines fall short on efficiency. According to GE Vernova, aeroderivative turbines operate at around 40 percent efficiency, compared to the 64 percent achieved by modern combined-cycle gas turbines which recycle exhaust heat for greater output. This efficiency gap carries financial and environmental penalties, driving up fuel costs and carbon emissions.

Concerns about pollution and regulatory compliance have already escalated tensions in communities hosting such facilities. In South Memphis, Tennessee, a legal challenge is underway against xAI’s Colossus data centre over air quality issues and the alleged lack of proper permits. The site, currently equipped with 15 gas turbines providing continuous power, is targeted for expansion to one gigawatt and is expected to draw partly on solar energy. Critics contend that such rapid infrastructure growth poses health risks, with emissions contributing to respiratory and cardiac illnesses. Despite the controversy, technology leaders appear resolved to press ahead, prioritising a continuous and scalable power supply to sustain their AI ambitions.