Musk’s Faster Gas Turbine Push Comes With a Pollution Problem

Elon Musk is trying to attack one of the AI industry's newest bottlenecks the same way his companies have approached rockets, cars and chips: manufacture more of the critical hardware in-house.

SpaceX is laying the groundwork for a foundry in Bastrop, Texas, that could manufacture highly specialized blades and vanes used inside industrial gas turbines, according to The Information. Musk says bringing the casting process under SpaceX's control could cut as much as 18 months from the time required to bring new natural-gas turbine capacity online.

That could give Musk's expanding AI infrastructure access to electricity faster at a time when data centers are running into grid delays and shortages of power equipment.

But it comes with a second problem. The gas turbines already being used to power Musk's AI data centers have become the subject of lawsuits, regulatory disputes and growing concern over nitrogen oxides, particulate matter and other air pollutants.

SpaceX Wants to Make One of the Hardest Turbine Parts Itself

The apparent plan centers on turbine blades and vanes, components that operate under extraordinary heat and mechanical stress.

Job listings uncovered by The Information refer to a new "blades and vanes foundry" in Bastrop and positions involving high-temperature superalloys, castings, automation and industrial gas-turbine manufacturing. SpaceX reportedly purchased roughly 830 acres near its existing Starlink operations in the area between March and June.

The reason for bringing production inside SpaceX is straightforward: turbine manufacturers cannot currently supply equipment fast enough to match the pace of AI data-center construction.

Musk said that producing the components internally could accelerate the deployment of natural-gas turbines by up to 18 months. He has argued that gas will remain necessary for several years to provide dependable electricity while solar production and other energy infrastructure scale.

The challenge is that turbine blades are not ordinary cast metal parts.

According to reporting cited by TechCrunch, turbine components can operate in environments reaching roughly 3,000 to 3,600 degrees Fahrenheit, temperatures around 800 degrees above the melting point of the alloys used to make them. Cooling passages, coatings and highly specialized manufacturing processes allow the components to survive those conditions.

The most demanding blades are formed as single-crystal structures to avoid microscopic boundaries that could become failure points.

Successfully manufacturing them at industrial scale would therefore give SpaceX control over a component that has traditionally been supplied by a small group of established turbine manufacturers.

AI’s Power Problem Is Getting Bigger

The push comes as electricity is becoming almost as important as GPUs in determining how quickly companies can deploy new AI infrastructure.

The International Energy Agency estimates global data-center electricity consumption reached around 485 terawatt-hours in 2025 and could rise to approximately 950 TWh by 2030.

Electricity use from AI-focused data centers is expected to triple over that period. Data-center power consumption already jumped 17% in 2025, far faster than the roughly 3% increase in overall global electricity demand.

In the United States, the pressure is even stronger. The IEA expects data centers to account for nearly half of the country's electricity-demand growth through 2030.

Building the computing hardware is therefore only part of the race. Companies also need substations, transmission lines, transformers, generation capacity and grid connections capable of supplying enormous amounts of continuous electricity.

Data centers can sometimes be constructed in two or three years, while large additions to electricity infrastructure can take considerably longer.

That mismatch is encouraging AI companies to build their own power systems instead of waiting for utility grids.

Gas Turbine Manufacturers Are Struggling to Keep Up

The shortage Musk is trying to bypass is visible in the order books of the world's biggest energy-equipment companies.

GE Vernova reported a $176 billion backlog in the second quarter of 2026 and said it expects to have at least 125 GW of gas equipment under contract by the end of the year.

The company is working to increase annual gas-turbine output from about 20 GW in 2026 to 24 GW in 2028 and 30 GW by 2030. Its data-center orders exceeded $5 billion during the first half of 2026, more than double its total for all of 2025.

The numbers illustrate how AI is reshaping the energy-equipment market.

Gas turbines offer something data-center developers value enormously: large amounts of dispatchable electricity that can operate regardless of whether the sun is shining or the wind is blowing.

For developers facing multi-year utility delays, putting power generation beside the data center can significantly shorten deployment schedules.

Musk Has Already Used Gas Turbines to Move Faster in Memphis

Musk's AI operations have already demonstrated the approach around Memphis, Tennessee.

Temporary mobile natural-gas turbines have been used to supply electricity to the massive Colossus computing infrastructure while permanent power generation and grid capacity are developed.

Reuters reported in July that xAI had installed 59 turbines, including 57 in Southaven, Mississippi, near the Tennessee border. The report estimated that the equipment could potentially produce as much as 2,500 tons of nitrogen oxide emissions annually, alongside other pollutants, if operated at reported levels.

The company disputes claims that its operations violate environmental rules and has said temporary turbines were permitted to operate under applicable exemptions and state regulatory decisions.

The situation has also continued changing as the site expands.

On July 30, SpaceXAI said there were 69 temporary mobile turbines powering its Southaven facility and announced an agreement with the Mississippi Department of Environmental Quality setting a timetable for their removal. The company said removal would begin in August 2026 and all temporary units would be gone by July 2027.

They are ultimately expected to be replaced by a permanent 1.2 GW power plant containing 41 turbines, which SpaceXAI says received a Clean Air Act permit in March.

The Turbines Have Triggered a Clean Air Act Fight

Environmental and civil-rights organizations argue that the speed advantage provided by temporary turbines has come at the expense of normal pollution controls and permitting.

The NAACP filed a Clean Air Act lawsuit in April against xAI and its subsidiary MZX Tech, alleging that dozens of turbines had been operated without required air permits. It later sought a preliminary injunction aimed at stopping the allegedly unpermitted pollution.

The legal dispute is contested.

The U.S. Department of Justice moved to intervene in June and sought dismissal of the lawsuit, arguing that Mississippi regulators had determined a permit was not required for the equipment at issue.

The fight illustrates a broader question AI infrastructure developers are beginning to face: whether equipment described as temporary should be regulated differently when dozens of units together effectively operate as a large power plant.

That question could become increasingly important if more developers adopt similar strategies.

The Pollution Issue Extends Beyond Musk

Musk's projects are the most visible example, but they are far from the only data centers exploring on-site fossil-fuel generation.

In Virginia, where some of the world's densest clusters of data centers have been built, a study commissioned by the Piedmont Environmental Council examined the potential impact of eight full-time gas turbines at one Vantage data-center facility.

Using the U.S. Environmental Protection Agency's COBRA health-impact model, researchers estimated permitted emissions could contribute to 3.4 to 6.5 additional premature deaths annually across the affected region.

The analysis put potential health-related damages at roughly $53 million to $99 million per year and estimated pollution could affect more than 2.5 million people across Northern Virginia and parts of the Washington metropolitan area.

Those figures are modelled estimates rather than documented health outcomes, but they show why local resistance to turbine-powered data centers is growing.

America Is Building More Gas Capacity as AI Expands

The wider energy shift is becoming visible at the national level.

Recent Global Energy Monitor analysis found the United States had around 378 GW of proposed gas-fired generating capacity in its development pipeline, up sharply from earlier levels. Around half of planned capacity was linked directly or indirectly to the surge in electricity demand from AI and data centers, according to reporting on the analysis.

The IEA expects natural gas generation serving data centers globally to increase by approximately 175 TWh through 2035, although renewable energy is projected to provide the largest portion of new supply.

That means gas increasingly looks like a bridge between the speed at which technology companies want to build AI and the slower speed at which grids and new low-carbon generation can expand.

Musk Could Solve One Bottleneck and Create Another

For Musk, producing turbine blades at SpaceX could deliver a significant competitive advantage.

AI companies can buy GPUs. They can raise billions to construct data centers. But if they cannot connect those facilities to enough electricity, expensive chips can remain unused.

Controlling turbine-component manufacturing would push Musk's strategy deeper into the physical infrastructure underlying AI, reducing dependence on another constrained supply chain.

Yet removing the manufacturing bottleneck would also make it easier to deploy more fossil-fuel generation quickly.

That puts two competing priorities on a collision course. AI companies want gigawatts of dependable electricity as quickly as possible, while communities and regulators want industrial-scale power generation subjected to pollution limits, monitoring and public review.

If SpaceX succeeds in cutting turbine deployment times by as much as 18 months, Musk may have found a faster route around one of AI's biggest infrastructure constraints.

The next question is whether environmental regulation can move just as quickly.