# AI data centers turn to jet engines for power as grid delays bite

Published 2026-11-03T22:24:17+00:00 · Updated 2026-11-03T22:24:18+00:00 · By superintelligencenews@gmail.com
Canonical: https://superintelligencenews.com/companies/ai-startups/ai-data-centers-jet-engines-power/

> AI data centers are increasingly using converted jet engines to produce their own electricity while waiting for delayed grid hookups. The approach is faster, but it raises efficiency and emissions concerns.

## Key facts

- TechCrunch reported that xAI has installed eight converted jet-engine turbines in Mississippi for its Colossus 2 data center in Memphis.
- TechCrunch said connecting 100 megawatts to the grid in Northern Virginia can take about seven years.
- Mint said FTAI Aviation plans to sell a modified Boeing 737 engine for data centers this year.
- Mint reported that Crusoe walked away from a $1.25 billion deal in September to buy 29 turbines from Boom Supersonic.
- Mint said FTAI expects annual delivery capacity of 100 turbines, equal to about 2.5 gigawatts.
- TechCrunch reported that Rystad Energy analyst Audun Martinsen said jet-powered units can get a data center running in 24 months.

AI developers are increasingly installing converted jet engines at data centers to generate electricity on site while they wait for grid connections and new power plants, according to TechCrunch and Mint. The appeal is speed: the units can be deployed far faster than traditional grid buildouts, even though they are less efficient and more polluting than conventional gas-plant turbines.

## What is driving the shift?

The immediate problem is power. AI campuses need large, reliable electricity supplies, but grid hookups and new turbine projects can take years in many markets. That has pushed operators toward interim generation options that can be installed much more quickly, especially where the artificial intelligence buildout is outrunning utility infrastructure.

Rystad Energy analyst Audun Martinsen framed the trend bluntly in comments to TechCrunch: “It’s all about getting power as fast as possible.” He added that with jet-powered units, “they can have a data center running in 24 months.”

That timing is attractive when compared with utility timelines that can stretch far longer. TechCrunch cited the International Energy Agency saying that connecting 100 megawatts to the grid in Northern Virginia can take about seven years. Mitsubishi Power’s chief executive also said most new power capacity is not expected online until 2031, underscoring why some operators are looking for stopgap solutions now rather than waiting for the normal utility cycle to catch up.

## How do converted jet engines work?

The units being deployed are retired aircraft engines rebuilt to burn natural gas. In power-industry terms, they are a form of aeroderivative turbine: machines derived from aviation engines and adapted for electricity generation. The core attraction is not efficiency. It is the ability to get meaningful on-site power up and running much faster than a conventional gas plant or a lengthy grid interconnection process.

That trade-off is central to the emerging market. The engines can be installed relatively quickly, but they are generally less efficient and more polluting than larger conventional turbines. Industry executives and analysts see them as a bridge rather than a destination unless those shortcomings improve.

## Where are these turbines already being used?

The best-known example so far is xAI. TechCrunch reported that the company has already installed eight converted jet-engine turbines in Mississippi for its Colossus 2 data center in Memphis. The units come from ProEnergy, which rebuilds a GE engine commonly used on Boeing jets.

[OpenAI](https://openai.com/) has also leaned on gas turbines. TechCrunch said the company used gas turbines at its first Stargate site in Abilene, Texas. Crusoe, which is building that campus, ordered 13 ProEnergy converted units in April, according to the report.

Crusoe later walked away from a separate $1.25 billion deal in September to buy 29 turbines from Boom Supersonic, another sign that the market is still fluid and that operators are comparing suppliers, timelines and technical packages rather than committing to one standard approach.

| Project / company | Power approach | Reported timing or scale | Source |
| --- | --- | --- | --- |
| xAI, Colossus 2 | Converted jet-engine turbines from ProEnergy | 8 units installed in Mississippi | TechCrunch |
| OpenAI, Stargate site | Gas turbines | First site in Abilene, Texas | TechCrunch |
| Crusoe | ProEnergy converted units | 13 units ordered in April | TechCrunch |
| Crusoe / Boom Supersonic | Natural-gas turbines based on modified engines | $1.25 billion deal for 29 turbines dropped in September; delivery expected in 2027 | TechCrunch, Mint |
| Microsoft / Nscale | Utility-backed data center project | Anchor customer exited after power was pushed to the early-to-mid 2030s | TechCrunch |

## Why it matters

This is not just an equipment story. It is a sign that power scarcity has become one of the main bottlenecks in the AI buildout. The scale of capital chasing new compute is colliding with a slower-moving electricity system, and the result is a growing market for temporary or semi-temporary generation that can be installed near the load.

For readers, the implication is that where and how AI infrastructure gets built may increasingly depend on access to local fuel, turbine suppliers and land for on-site generation, not just fiber and real estate. For the market, it creates a new demand stream for turbine makers and engine converters. For the region, it raises familiar concerns about emissions, local air quality and the politics of allowing large industrial users to generate their own power.

The policy backdrop matters too. TechCrunch reported that the Trump administration is encouraging data centers to generate their own electricity, which could make on-site generation more attractive in the near term. That does not solve the underlying capacity problem, but it may ease the approval path for operators trying to move faster than utilities can build.

## Background: how the industry got here

The current rush reflects two facts that have become hard to ignore. First, AI infrastructure consumes immense amounts of power. Second, many power systems are not expanding fast enough to meet the pace of demand. TechCrunch’s reporting suggests the mismatch is especially acute in places such as Northern Virginia, a major data center hub where even a 100-megawatt connection can take about seven years.

That kind of delay is incompatible with the cadence of the AI race, where companies are spending aggressively to secure compute capacity now. Mint said planned AI technology capital expenditure this year totals $700 billion, a figure that helps explain why fast power solutions are suddenly receiving so much attention.

Converted jet engines are attractive because they turn existing industrial hardware into a faster route to electricity. They also sit in a broader family of aeroderivative turbines, a category long used in power generation. The difference here is that the equipment is now being marketed specifically to AI data centers, where time-to-power can matter more than long-term efficiency, at least at the start of a project.

### How do they compare with conventional turbines?

Heavy-duty gas turbines are generally more efficient than smaller aeroderivative units, but they often come with longer lead times. That is the fundamental trade-off. Smaller converted units can arrive faster, but they are not the best answer if a project is designed to run for decades and must minimize fuel use and emissions.

Scott Strazik, chief executive of GE Vernova, pushed back on the idea that the smaller units are direct competitors to the industry’s mainstream offerings. “We don’t really view those smaller units to be in competition,” he told Mint. But he also made clear that efficiency cannot be ignored, saying “when you’re underwriting 20-year business cases, efficiency matters a lot.”

That split captures the market tension. For a rushed AI campus, speed may trump all else. For a utility planning a 20-year asset, the lower fuel burn of a larger turbine remains hard to beat.

## What Mint reported about the business opportunity

Mint’s reporting widened the lens from one-off projects to the possible size of the market. It said FTAI Aviation plans to sell a modified Boeing 737 engine for data centers this year, and that its shares rose about 42% after the power-turbine announcement. Jefferies estimated the business could add $750 million in annual EBITDA, a figure Mint said would amount to roughly 52% of FTAI’s expected EBITDA.

Mint also said ProEnergy sells turbines adapted from the Boeing 747 engine, while Boom Supersonic said in December it would sell a modified engine as a natural-gas turbine. Boom’s first customer is Crusoe, which Mint said should receive the turbines in 2027.

The business case appears to rest on a simple proposition: if data center operators need power quickly, then converted aircraft engines become a product category rather than an engineering curiosity. Mint said FTAI expects annual delivery capacity of 100 turbines, equal to about 2.5 gigawatts. Boom, by contrast, wants to reach more than 4 gigawatts of annual manufacturing capacity by 2030.

### By the numbers

| Metric | Figure | What it refers to |
| --- | --- | --- |
| Time to run a data center with jet-powered units | 24 months | Rystad Energy estimate cited by TechCrunch |
| Grid hookup in Northern Virginia | 7 years | IEA estimate cited by TechCrunch |
| xAI turbines installed | 8 | Converted jet-engine turbines in Mississippi |
| Crusoe ProEnergy order | 13 | Converted units ordered in April |
| Crusoe-Boom deal size | $1.25 billion | Deal dropped in September |
| FTAI conversion time | 30 to 45 days | Time to convert one engine, per David Moreno |
| FTAI redesign timeline | 18 months | Time spent redesigning the turbine |
| FTAI annual delivery capacity | 100 turbines / 2.5 GW | Mint estimate |
| Boom target by 2030 | 4 GW+ | Annual manufacturing capacity goal |

## Why it matters for the market

The turbine makers are not starting from zero. Mint said about 1,600 commercial aircraft engines are retired each year, and if even a fraction are converted into power units, the available supply could be meaningful. It cited a theoretical estimate of 13 gigawatts if one-third of retired commercial engines were converted at FTAI-like scale, and 40 gigawatts from retired military aircraft engines based on an EIA estimate.

That said, supply is only one part of the story. Mint noted that the gas turbine market is still dominated by GE Vernova, Mitsubishi Heavy Industries and Siemens Energy, which together hold about 80% market share. Those incumbents remain the benchmark for efficiency, reliability and large-scale industrial deployment.

The scale numbers help explain the attraction and the limits of the trend. Morgan Stanley estimated global annual gas turbine manufacturing capacity at 47 gigawatts across 2023 to 2025, while the planned AI infrastructure spending boom continues to pull in more demand. In that context, even a niche supply of fast-to-install turbines can become strategically important.

## Where the reporting differs

TechCrunch and Mint are telling the same broad story from different angles. TechCrunch focused on the operational pressure points: grid delays, named data center projects and the practical reason operators are choosing converted jet engines. Mint focused more on the manufacturing market, the stock reaction and the companies trying to commercialize the idea at scale.

They also differ in emphasis on the technology lineage. TechCrunch described units built from retired aircraft engines and mentioned ProEnergy’s rebuilds of a GE engine commonly used on Boeing jets. Mint described ProEnergy’s turbines as adapted from the Boeing 747 engine and highlighted FTAI’s plan to sell a modified Boeing 737 engine. The common thread is that the industry is mining aviation hardware for a power-generation shortcut.

## What happens next?

Several questions will determine whether converted jet engines become a durable category or a temporary fix.

- Will more AI operators choose on-site generation rather than wait for grid access?
- Can FTAI, Boom and ProEnergy scale manufacturing to the levels they are describing?
- Will efficiency and emissions concerns limit wider adoption?
- Will more retired aircraft engines be redirected from aviation parts supply into power generation?
- Which other cloud or AI companies will sign up for similar systems?

The near-term watch item is delivery. Mint said Crusoe should receive Boom turbines in 2027, while FTAI says it wants to start selling modified Boeing 737-based turbines this year. At the same time, TechCrunch’s reporting suggests the pressure to secure electricity is only getting worse, not better.

That leaves the market in a transitional phase. AI developers need power now. Utilities and conventional turbine makers are moving on much slower timelines. Converted jet engines sit in the gap between those realities: imperfect, faster and, for now, increasingly in demand.

Whether that gap closes into a real industry category will depend on more than engineering. It will hinge on policy, fuel costs, manufacturing scale and how much emissions trade-off AI customers are willing to accept in exchange for speed.

## One clear takeaway

AI data centers are turning to converted jet engines because the grid cannot always keep up with their power needs, and the shortage is forcing operators to choose between fast, dirtier on-site generation and much slower utility connections. The trend is still early, but the companies building these systems are already racing to scale.

## Frequently asked questions

### Why are AI data centers using jet engines for power?

AI data centers are using converted jet engines because they can be deployed faster than waiting for grid hookups or new large-scale power plants. The trade-off is that these units are generally less efficient and more polluting than conventional gas turbines.

### Which companies are already using converted jet-engine turbines?

xAI is the clearest example cited in the reporting, with eight converted jet-engine turbines installed in Mississippi for its Colossus 2 data center in Memphis. TechCrunch also said OpenAI used gas turbines at its first Stargate site in Abilene, Texas.

### How long can grid connections take for data centers?

Grid connections can take years in some markets. TechCrunch cited the IEA as saying that connecting 100 megawatts to the grid in Northern Virginia can take about seven years, which is one reason operators are looking for faster alternatives.

### Are jet-engine turbines a permanent solution?

No. The reporting suggests they are mainly a bridge solution. Industry observers say the units are useful because they are fast to deploy, but efficiency and emissions concerns could limit how widely they are used over the long term.

### What companies are trying to sell this technology?

FTAI Aviation, ProEnergy and Boom Supersonic are all trying to commercialize adapted aircraft-engine turbines for data centers. Mint said FTAI plans to sell a modified Boeing 737 engine this year, while Boom and ProEnergy are also active in the market.

## Sources

- [TechCrunch](https://techcrunch.com/2026/11/03/data-centers-are-turning-to-old-jet-engines-as-the-wait-for-power-grows/)
- [mint](https://www.livemint.com/technology/how-jet-engines-are-powering-data-centers/amp-11771325969187.html)