Elon Musk claims he can accelerate gas turbine production by manufacturing the components himself.
On Saturday, Musk confirmed that a secret foundry SpaceX is building in Bastrop, Texas, is intended for this purpose. The announcement followed a report from The Information that detailed job listings for a “blades and vanes foundry” and noted SpaceX purchased roughly 830 acres near its Starlink factory between March and June. Corey Trinetti, a specialist who reviews AI infrastructure sites, had previously identified the land acquisition.
“SpaceX and Tesla are each building 100GW/year of solar production capacity as fast as possible,” Musk wrote on X. “but natural gas will still be needed to supplement and bootstrap solar for several years. The limiting factor for nat gas turbine production is casting the blades & vanes. By doing in-house casting at SpaceX, we can accelerate nat gas turbines coming online by up to 18 months, which is a profound game-changer.”
This strategy addresses a specific constraint facing the AI industry. While GPU shortages persist, with Nvidia’s Blackwell chips facing lead times of several months, the physical power grid has become a second bottleneck. The International Energy Agency projects global data center electricity use will roughly double by 2030. GE Vernova, a gas turbine maker, states it has sold out production capacity through 2030 due largely to AI infrastructure demand.
That shortage explains why hyperscalers including Amazon, Google, Meta, OpenAI, and Microsoft are building private gas-fired plants next to data centers. After years of prioritising wind and solar, these companies are now betting on natural gas to get data centres online faster.
The casting bottleneck is severe. According to The Information, the blades inside a gas turbine’s hottest section run at temperatures around 3,000 to 3,600 degrees Fahrenheit. This is roughly 800 degrees hotter than the melting point of the metal alloy used. The blades survive only because of internal cooling channels, thermal-barrier coatings, and a specific casting method. Just four companies worldwide have mastered this process at industrial scale, and all are currently tapped out.
Each blade must be cast as a single, unbroken crystal. The metal is grown slowly inside a vacuum furnace to avoid microscopic seams that cause cracks under stress. This is a difficult process even for smaller jet engine blades. The blades in power-plant turbines are considerably larger, making production at that scale without defects even harder.
If SpaceX succeeds, a Musk-controlled entity would hold a manufacturing capability that every other AI infrastructure builder currently depends on for a tiny oligopoly. This gives SpaceXAI an edge that is difficult for any well-funded but non-manufacturing competitor to copy quickly.
However, faster turbines mean more pollution. Turbines in the ground are already drawing federal lawsuits and peer-reviewed health research over the emissions they produce.
In Memphis, where SpaceXAI has run gas turbines to power its Colossus data centres since 2024, the NAACP has repeatedly accused the company of operating turbines without the permits or pollution controls required by federal law. The organisation states these turbines emit smog-forming compounds and hazardous chemicals like formaldehyde. These pollutants are linked to asthma, respiratory disease, and certain cancers. The site sits near neighbourhoods that already face heavy industrial pollution. University of Memphis researchers said that in their analysis, air pollution grew “slightly worse” because of the data centre.
Memphis is the most visible case, but similar fights are playing out where gas turbines have become the default fix for data centre power shortages. In Virginia’s “Data Center Alley,” a study commissioned by the Piedmont Environmental Council used the EPA’s own COBRA health-impact model. It found that emissions from a single facility’s eight full-time gas turbines could reach more than 2.5 million people across multiple counties. The heaviest impact lands on already-marginalised communities. The study estimates 3.4 to 6.5 additional premature deaths a year, translating to $53 million to $99 million in annual health-related damages.
The list of complaints continues.
What it means
For the people making things, the shift is clear. Data centre operators no longer wait for public grid connections or rely on a few suppliers for turbine parts. They build their own power generation on-site using gas. This allows companies to deploy infrastructure faster but shifts the burden of local pollution directly to the communities surrounding those sites.
