SpaceX Wants to Build Its Own Power Plant Parts to Speed Up AI Data Centers

Elon Musk confirmed on August 30, 2026 that SpaceX is building a factory in Bastrop, Texas, to make key parts for gas turbines, the machines that burn natural gas to generate electricity. Musk said on X that the slowest step in making turbines is casting the blades and vanes, the internal parts that handle the hottest temperatures, and that doing this work at SpaceX instead of buying from outside suppliers could speed up turbine production by up to 18 months. He also said SpaceX and Tesla are each trying to build enough solar panel production capacity to generate 100 gigawatts of electricity per year, but natural gas will be needed to fill the gap for several years while solar ramps up (TechCrunch).

The Information first reported on August 29 that SpaceX was laying groundwork for a turbine-blade factory to address the data center power crunch, noting coverage involving Musk and GE Vernova gas turbines (The Information). SpaceX's own careers page has since listed multiple roles tied to the effort: a Materials Engineer for a "blades and vanes foundry" in Bastrop, an Automation Engineer at the same location, a Senior Gas & Chemical specialist also in Bastrop, and a Mechanical Engineer for gas and steam turbines at Starbase, Texas. Due diligence specialist Corey Trinetti reported that SpaceX acquired roughly 830 acres near its existing Starlink factory in Bastrop between March and June 2026 (TechCrunch).

The technical challenge SpaceX is taking on is extremely difficult. Gas turbine blades in the hottest part of the machine operate at temperatures of roughly 3,000 to 3,600 degrees Fahrenheit, about 800 degrees hotter than the melting point of the metal alloy they are made from. Each blade must be cast as a single, unbroken crystal, grown slowly inside a vacuum furnace using a process that carefully controls how the molten metal cools so it hardens into one continuous piece with no internal boundaries or seams. Only four companies worldwide have mastered this process well enough to produce blades at industrial scale. The barriers are about the science of metals, not just manufacturing: getting the crystal structure right requires precise control over temperature, cooling speed, and furnace atmosphere over long production cycles, and the rate of defects when scaling up is high.

The demand pressure driving this move is real and immediate. The International Energy Agency projects global data center electricity consumption will roughly double by 2030. GE Vernova, one of the dominant gas turbine manufacturers, says it is essentially sold out of production capacity through 2030, largely because of AI infrastructure demand. Nvidia's Blackwell GPUs still carry lead times of several months. Major tech companies including Amazon, Google, Meta, OpenAI, and Microsoft are building private gas-fired power plants next to their data centers rather than waiting years for permission to connect to the regular electrical grid. The bottleneck has shifted from computing power to electrical power, and within electrical power, from generating capacity to the physical supply of the turbine machines themselves.

Musk's own operations illustrate the urgency, and the friction. Reuters reported in August 2024 that xAI was operating gas turbines without permits at its Memphis data center. The NAACP sued xAI and a subsidiary in April 2026, alleging illegal operation of more than two dozen gas turbines. A Reuters analysis in July 2026 found that 30 gas turbines at the unpermitted xAI power project could emit nearly 2,500 short tons of nitrogen oxide, a pollutant that harms lungs, and that the pollution hit Black communities hardest (Reuters). SpaceX agreed in August 2026 to begin removing the temporary gas turbines powering part of its Colossus AI data center complex by 2027 (Gizmodo).

That pattern, acquiring turbines under regulatory pressure and then committing to phase them out, frames the foundry decision in a particular light. If SpaceX can cast its own blades and vanes, it gains its own supply chain for the components that limit how fast turbines can be built, reducing dependence on the four established manufacturers whose backlogs now stretch years. Whether SpaceX can produce single-crystal blades at industrial scale, a process the incumbents spent decades refining, is an open engineering question that the job listings and land purchases suggest the company is committing serious resources to answer.

The broader context here is a collision between three timelines: the AI industry's need for power measured in months, the wait to connect to the electrical grid measured in years, and the turbine supply chain measured in years on top of that. SpaceX entering the foundry business is an attempt to compress the third timeline. Musk has framed gas as a bridge to the 100 GW/year solar buildout he says SpaceX and Tesla are pursuing, which would itself be an unprecedented manufacturing scale-up if pursued to completion. The foundry does not resolve the environmental and permitting concerns that have followed xAI's Memphis operations. It does signal that Musk's companies are willing to move into the most technically difficult components of power hardware rather than wait for existing suppliers to catch up.

For anyone tracking the AI infrastructure buildout, the foundry is worth watching as an indicator of how far companies will go to secure power. The single-crystal casting process is not something that can be improved through the kind of rapid trial-and-error common in software; it is furnace physics, alloy chemistry, and patient process engineering. If SpaceX succeeds, it could meaningfully shorten the timeline for new turbine capacity at a moment when every other path to powering data centers is congested. If it does not, the bottleneck stays where it is today: at the foundry door.