SpaceX is building a foundry in Bastrop, Texas, to independently manufacture gas turbine blades and vanes to address power shortages for AI data centers.Article author and source: Wall Street Journal
Facing the industry-wide pain point of power shortages in AI data centers, Musk is applying his signature vertical integration logic to forcefully break through bottlenecks by targeting the weakest link in the supply chain.
On August 30, according to the technology media The Information, SpaceX is building a foundry in Bastrop, Texas, dedicated to producing blades and vanes for large gas turbines—among the most technologically advanced and critically scarce components in the global turbine supply chain.
After the report was published, Musk personally confirmed the plan on the social platform X, stating that SpaceX's in-house manufacturing of turbine blades and vanes could bring forward the deployment timeline of gas turbines by up to 18 months, calling it "a profound game-changer."
Musk also stated that SpaceX and Tesla are each building solar capacity at the fastest possible pace to reach 100 GW per year, but "natural gas will remain a necessary complement and backup for solar over the next several years."
Meanwhile, research by Morgan Stanley analyst Adam Jonas corroborates the above report, noting that the foundry will also meet the casting demands for SpaceX’s Raptor rocket engine turbopumps, enabling cost-sharing between aerospace and AI operations.
Bottleneck: The Hidden Constraint Stalling Global AI Computing Expansion
Reports state that gas turbine blades and vanes must operate at extreme temperatures of 3,000 to 3,600 degrees Fahrenheit (approximately 1,650 to 1,980 degrees Celsius), about 800 degrees Fahrenheit above the material's melting point, placing extremely demanding requirements on the casting process.
Currently, only four minting companies worldwide possess this technology, creating a highly monopolistic landscape. This oligopoly has directly resulted in orders for major turbine manufacturers like GE Vernova being booked through 2030, becoming one of the key bottlenecks accelerating AI infrastructure development.
In February this year, Musk explicitly stated on a podcast that blades and vanes are the "bottlenecks" preventing faster delivery of turbines, and warned that, at current production rates, "I'm not sure there will be enough available electricity to power all the AI chips being produced."
Bastrop: From Satellite Factory to Turbo Casting Hub
Reports indicate that SpaceX has established several industrial facilities in Bastrop, approximately 30 miles from Austin. The company has previously operated a facility in the area related to its Starlink satellite business.
According to Corey Trinetti, a due diligence expert at Measured AI, an AI infrastructure research newsletter, SpaceX purchased approximately 830 acres of surrounding land between March and June this year. Satellite imagery shows that two new buildings appeared next to the Starlink factory in July, whereas no structures were present at that location a year ago.
At the hiring level, SpaceX has recently posted several positions, including Senior Operations Engineer, Senior Materials Engineer, Automation Engineer, and Senior Tooling Engineer responsible for the construction, expansion, and operational ramp-up of a new foundry, as well as a programmer tasked with programming high-temperature superalloy milling for a new industrial gas turbine (IGT) blade and vane manufacturing line in Bastrop.
However, Trinetti noted that no permits or construction filings for the foundry have been found in Texas’s environmental or building records, and it remains unclear whether SpaceX has initiated the approval process.
Dual-use: Cost-sharing logic between aerospace and AI
In his report this month, Adam Jonas of Morgan Stanley noted that SpaceX is establishing a turbine component foundry in Texas to serve both data center demands and the production of castings for Raptor rocket engine turbine pumps.
"Raptor turbine pump castings and data center turbine airfoil components use the same alloys, furnaces, and skilled workers, allowing a single foundry to spread fixed costs across both aerospace and AI businesses," Jonas said.
This dual-purpose logic significantly enhances the commercial viability of the foundry, while also prompting cautious observation from outsiders regarding whether SpaceX can truly replicate the decades-long casting expertise accumulated by industry leaders such as Precision Castparts.
Transition period: Acquire bridging power through acquisitions and partnerships
Before its own power generation capacity comes online, Musk is seeking gigawatt-scale transitional power supplies through multiple channels. According to The Information, Musk has recently collaborated quietly with Wall Street hedge fund Fortress Investment Group, which has invested in at least two turbine suppliers.
According to a filing with the U.S. Federal Trade Commission in May, Fortress has sold APR Energy, a supplier of mobile industrial turbines, to Musk himself. In January, APR Energy stated that its fleet consisted of approximately 1 GW of turbines, meaning these assets are now accessible to Musk.
In addition, SpaceX may also be purchasing turbines from FTAI Aviation—a company specializing in modifying and overhauling commercial aircraft engines.
FTAI was spun off from another Fortress-affiliated company and, in July this year, announced through its Chinese joint venture partner a $1.465 billion turbine sales contract with a "leading international cloud service provider." Adam Jonas has listed FTAI as a potential supplier to SpaceX.
Morgan Stanley estimates that Musk could secure approximately 3 to 4 GW of power by the end of 2027 through a series of procurement agreements.
Carbon Emission Controversies and Industry Divisions
However, the report also noted that Musk's further expansion of fossil fuel power generation capacity has raised concerns among advocates of low-carbon energy.
BloombergNEF estimates that if all planned off-grid natural gas projects for data centers are fully implemented, they will increase U.S. carbon dioxide emissions by 20%.
This figure does not yet account for Musk’s latest plans but includes the 2 GW of off-grid natural gas capacity planned by SpaceX’s xAI division for its data center in Memphis, Tennessee, to be operational by year-end.
There is significant disagreement in the industry over whether Musk can truly achieve this ambitious plan. Dylan Patel, founder of semiconductor research firm SemiAnalysis, believes Musk can accomplish it; however, clean energy entrepreneur Jigar Shah, who previously led the Biden administration’s Office of Energy Loan Programs, directly countered: "That’s because you have no idea how to build infrastructure."
Musk previously shocked the industry with the rapid deployment of the Memphis Colossus data center—although its facilities are not top-tier, SpaceX has already generated substantial revenue by leasing the facility to Anthropic. Analysts are closely watching whether this bold bet on power infrastructure can again replicate such speed.