AI Data Centers Are Buying Used Jet Engines as New Power Turbines Take Six Years
AI data centers are acquiring used jet engines and refurbished gas turbines to power their facilities due to a six-year waiting list for new equipment, driven by surging demand and manufacturing backlogs.
Intelligence analysis by Gemini 2.5 Flash
The rapid expansion of AI data centers is creating an urgent demand for power generation, leading developers to unconventional solutions like repurposing jet engines and buying second-hand turbines. This workaround addresses the significant delays in obtaining new gas turbines, which are crucial for round-the-clock electricity supply to densely packed computing facilities.
Imagine a super-fast computer brain needs a lot of electricity, like a hungry robot. But the special power generators it needs are like popular toys that take years to make. So, clever builders are buying old airplane engines and fixing them up to make electricity quickly, even though it means they have to think about things like air pollution and getting special permission to run them.
Analysis
The burgeoning demand for AI computing power is creating an unprecedented strain on global energy infrastructure, particularly concerning the supply of large-scale power generation equipment. New gas turbines, essential for providing the continuous, high-capacity electricity required by modern data centers, now face an average delivery time of six years in the United States, with discussions for new orders extending to 2030 and 2031. This timeline is fundamentally incompatible with the AI industry's aggressive growth trajectory and its need to bring enormous computing campuses online much sooner. The global backlog is severe, with orders exceeding 100 gigawatts against an annual manufacturing capacity of only 60 to 70 gigawatts, as reported by Reuters, indicating a systemic bottleneck from major producers like Siemens Energy, GE Vernova, and Mitsubishi Heavy Industries.
Crusoe
To circumvent these extensive delays, AI infrastructure firms are exploring innovative and often unconventional power solutions. One notable example is Crusoe, an AI infrastructure firm, which placed an order for 13 aeroderivative turbines from Missouri-based ProEnergy in April. These units, each rated at 50 megawatts, collectively provide 650 megawatts of electricity, derived from technology adapted from aircraft jet engines. While the contract does not necessarily mean Crusoe is buying engines directly from retired aircraft, it signifies a pivot towards rapidly deployable, high-capacity power generation. The secondary market for used turbines is also experiencing a boom, with equipment brokers noting that good condition used units sell for approximately $1.5 million to $1.7 million per megawatt, a substantial discount compared to the $2 million to $2.5 million for new units. This cost-effectiveness, coupled with faster availability, makes second-hand machinery an attractive, albeit increasingly competitive, option for developers.
Environmental Protection Agency
While these alternative power sources offer a temporary reprieve from supply chain woes, they introduce a new set of challenges, particularly concerning environmental regulations and permitting. Building a power plant directly alongside a data center necessitates careful consideration of fuel requirements, air emissions, and the acquisition of appropriate permits. The Environmental Protection Agency (EPA) explicitly states that stationary turbines and engines used by data centers are subject to federal air-pollution standards. This regulatory hurdle can complicate and potentially delay projects, even if the power equipment itself is readily available. The article notes that OpenAI recently committed to covering electricity and grid costs at its future campuses, a move that could be interpreted as a response to increasing scrutiny and tighter restrictions on data center construction, as seen with recent voter-backed measures. This highlights the growing tension between the urgent need for power and the imperative for sustainable and compliant energy solutions in the rapidly expanding AI sector.
Key points
- AI data centers face a six-year wait for new gas turbines due to global manufacturing backlogs.
- Companies are turning to used jet engines and refurbished turbines to generate on-site power.
- AI infrastructure firm Crusoe ordered 13 jet-engine-derived units capable of 650 megawatts.
- Second-hand turbines offer significant cost savings but present challenges with supply, maintenance, and environmental permits.
- Global data center electricity consumption is projected to nearly double by 2030, with AI facilities tripling their usage.
The adoption of aeroderivative turbines and the secondary market for power equipment could accelerate the deployment of critical AI infrastructure, enabling faster innovation and economic growth in the tech sector. This creative problem-solving might also foster new industries focused on repurposing and maintaining such specialized machinery.
Relying on used jet engines and on-site power generation could lead to increased environmental concerns due to air emissions and complicate permitting processes, potentially slowing down data center expansion despite the initial speed advantage. The rising competition for second-hand equipment also risks driving up costs and creating new supply chain bottlenecks.
