SpaceX Builds Texas Foundry to Accelerate AI Gas Turbines
In-house casting of turbine blades aims to bypass global supply bottlenecks and speed off-grid power deployment.
SpaceX is constructing a specialized turbine blade and vane foundry on 830 acres in Bastrop, Texas, to address power infrastructure bottlenecks facing AI data center expansion. Elon Musk confirmed the facility aims to produce hard-to-manufacture gas turbine components in-house to accelerate the deployment of natural gas power plants. By manufacturing critical casting components directly, the company expects to cut gas turbine deployment timelines for AI compute infrastructure by up to 18 months.
Key details
The primary constraint in scaling natural gas generation for off-grid AI data centers is the precision casting of single-crystal turbine blades and vanes. Operating in temperatures reaching 3,000 to 3,600 degrees Fahrenheit, these components require complex internal cooling channels and thermal-barrier coatings. Global production is currently limited to a tiny oligopoly of four specialized suppliers, all of which are sold out through 2030 due to surging data center energy requests.
To overcome this supply chain bottleneck, SpaceX acquired 830 acres adjacent to its Starlink production facility in Bastrop between March and June. The company plans to deploy in-house vacuum furnace casting technology to manufacture the large-format turbine blades required for utility-scale power generation. Additionally, Musk stated that SpaceX and Tesla are targeting 100 GW per year of solar manufacturing capacity, though supplemental gas generation will remain necessary for several years to bridge the grid capacity gap.
Why this matters
As electric power grids struggle with multi-year interconnection backlogs, hyperscalers and AI infrastructure developers are increasingly turning to behind-the-meter natural gas plants to energize gigawatt-scale data center campuses. Producing specialized turbine components in-house gives SpaceX and xAI a direct manufacturing advantage, bypassing multi-year lead times for power generation hardware. However, accelerating natural gas deployment raises major environmental and regulatory concerns regarding localized air pollution.
Context
Global data center electricity consumption is projected by the International Energy Agency to double by 2030, driven largely by high-density AI training and inference workloads. Major turbine manufacturers like GE Vernova report sold-out order books through the end of the decade, forcing tech firms to seek non-traditional supply chain solutions. Off-grid gas power installations have faced mounting public scrutiny, including federal Clean Air Act lawsuits and health impact assessments highlighting emissions of nitrogen oxides, formaldehyde, and particulate matter in surrounding communities.
Risks and open questions
Single-crystal turbine casting is technically difficult, and manufacturing large-scale power plant blades without structural defects presents substantial engineering hurdles. Questions remain over whether SpaceX can successfully achieve industrial-scale yields within its 18-month target window. Furthermore, increased reliance on unpermitted or fast-tracked gas turbines risks intensifying legal challenges from environmental groups and local communities over air quality and public health impacts.
Source: TechCrunch Published on AI Usage Global, author: AUG Bot
