The wait for grid access and new turbines has stretched so long that data center builders are now pulling retired jet engines out of storage to keep their facilities powered. That is a striking image, and it tells you something uncomfortable: the constraint shaping AI infrastructure right now is not ambition or capital but plain electricity, and the workarounds are getting less efficient and less environmentally friendly by the month. The scramble is understandable, but it is also a warning. As we noted in AI's hidden footprint is clear, but the real challenge is acting on it, the hardest part is no longer getting the numbers. It is deciding what to do with them. Retired jet engines are what happens when nobody decides in time. They burn fuel, they underperform modern turbines, and they represent a bet that speed today is worth a higher cost tomorrow.
What does this mean for the people building and buying data capacity? It means power strategy now belongs in the same conversation as site selection and hardware procurement, not downstream of it. If your roadmap assumes the grid will show up on schedule, stress-test that assumption. The builders turning to discarded engines are not doing it out of preference. They are doing it because the alternative is waiting years. That tradeoff will show up in operating costs, emissions reporting, and community relations, and it will not stay hidden. The way we talk about these facilities matters too, as we explored in What we call data centers shapes how we build the future. A facility running on salvaged jet engines is a harder story to tell a neighborhood than a sleek rendering of a modern campus.
The local politics are already catching up. In Brooklyn, as our reporting on the midterm campaign trail showed, AI and data centers have become dinner-table issues, which means the stopgap solutions builders reach for now will be scrutinized by people who live nearby. A less efficient, less clean power source is exactly the kind of detail that turns a permitting conversation into a fight. The practical takeaway is direct: plan your power supply as rigorously as you plan your compute, and treat grid timelines as a risk to be managed rather than a schedule to be trusted. The builders who win the next few years will be the ones who secured clean, adequate power before they needed it, not the ones who improvised loudest. Watch whether the jet engine workaround stays a curiosity or becomes a pattern. If it becomes a pattern, the efficiency and emissions math for AI infrastructure gets worse at exactly the moment more people are doing it.