AI Data Centers Turn to Converted Jet Engines to Bypass Grid Delays

The massive energy demands of the AI buildout are forcing data center operators to look past traditional utility grids and toward unconventional power sources, including converted aircraft jet engines. According to TechCrunch, citing the New York Times, some companies are using these rebuilt engines to generate their own electricity while facing years-long wait times for traditional grid connections.

What happened

Grid connections have become a major bottleneck for the AI industry. TechCrunch reports that hooking up 100 megawatts to the grid in Northern Virginia takes about seven years, according to the International Energy Agency. Meanwhile, Mitsubishi Power executives told Bloomberg that most new power capacity is not expected to come online until 2031. These delays have real business costs; TechCrunch notes that Microsoft pulled out as the anchor customer for an Nscale data center in Loughton, England, after the local utility stated it could not deliver enough power until the early to mid-2030s.

To bypass these waits, operators are adopting rapid alternatives. Rystad Energy analyst Audun Martinsen told the New York Times that jet-powered units allow operators to have a data center running in just 24 months. TechCrunch reports that Elon Musk's xAI installed eight converted jet-engine turbines in Mississippi to help power its Colossus 2 data center in Memphis, utilizing units from ProEnergy that rebuild GE engines commonly found on Boeing jets.

Other industry players are following suit. OpenAI utilized gas turbines at its first Stargate site in Abilene, Texas, in 2025, with Denver-based startup Crusoe constructing the campus and ordering 13 converted units from ProEnergy in April. TechCrunch notes that while Crusoe walked away from a separate $1.25 billion deal with Boom Supersonic in September for 20-odd natural gas turbines, a spokesperson confirmed the company still plans to use turbines. Additionally, the Trump administration is currently adding pressure by encouraging data centers to generate their own electricity.

These interim solutions carry distinct drawbacks. According to the New York Times cited by TechCrunch, converted jet engines are less efficient, pollute more than conventional gas plant turbines, and require more units to match standard power output. Experts suggest these environmental and efficiency trade-offs may become major issues in a year or two.

Why it matters for small operations teams

If you are running workloads on shared AI infrastructure or planning your own localized agent clusters, infrastructure bottlenecks at the massive data center level directly impact your scaling timelines and cloud costs. When major players like OpenAI and xAI face a seven-year wait for traditional utility connections and resort to temporary jet-engine power plants, it signals that the physical constraints of AI are very real. For small teams building human-AI hybrid companies, this means computing costs and availability will remain volatile. You cannot assume cheap, infinite cloud compute will always be instantly accessible as providers scramble to secure basic power.

What to do this week

1. Audit your current AI workload dependencies to identify whether your operations rely on single-region cloud providers vulnerable to regional grid and power constraints.
2. Build redundancy into your agent deployment strategy by testing multi-region or multi-provider fallback options for critical automated workflows.
3. Review your compute forecasting for the next 12 to 24 months to account for potential price fluctuations driven by broader infrastructure and energy shortages.

Sources

• Data centers are turning to old jet engines as the wait for power grows — TechCrunch - AI