SpaceX and Tesla Pick Texas for AI Chip Manufacturing Plant Set to Be World's Largest Building
Key keywords: SpaceX, Tesla, Texas AI chip manufacturing plant, world's largest building, Dojo supercomputer, AI hardware supply chain, advanced semiconductor production, xAI training accelerators
In a joint official announcement released this week, Elon Musk-led aerospace firm SpaceX and electric vehicle maker Tesla confirmed that they have selected a 2,100-acre site in the Austin metropolitan area, Texas, to build their co-owned AI chip manufacturing facility, which will also be the world’s largest single building by total floor area once completed. The planned structure will span over 18.6 million square feet, surpassing the size of Boeing’s Everett assembly plant, the current record holder for the world’s largest building, and 30% larger than Tesla’s existing Texas Gigafactory that produces Model Y and Cybertruck vehicles.
The decision to locate the plant in Texas came after 18 months of site evaluations across 12 U.S. states, with factors including low regulatory burden, competitive tax incentives, abundant industrial land, low energy costs, and proximity to existing core operations of both firms weighing heavily in the state’s favor. The site is located 25 miles from Tesla’s Texas Gigafactory and 180 miles from SpaceX’s Starbase facility in Brownsville, enabling seamless cross-team collaboration between manufacturing, R&D, and testing teams.
The $15 billion facility will focus on the mass production of custom-designed AI accelerators, including the D1 chips powering Tesla’s Dojo supercomputer used for full self-driving training, high-performance chips for SpaceX’s Starlink satellite constellation ground processing and Starship navigation systems, as well as custom training chips for Musk’s artificial intelligence startup xAI to support the iteration of its Grok large language model. The plant will adopt 3nm and 5nm advanced process technologies, with an initial annual production capacity equivalent to 1.2 million 12-inch wafers once the first phase comes online in 2026, enough to power over 300,000 high-performance AI training servers per year.
The project is expected to create more than 22,000 direct full-time jobs, ranging from semiconductor process engineers and manufacturing technicians to supply chain managers and AI hardware testing specialists, with average annual salaries exceeding $120,000. The Texas state government has approved $320 million in tax abatements and workforce development grants for the project, and has partnered with the University of Texas at Austin to launch a dedicated semiconductor talent training program to supply qualified workers for the plant. The facility will also run on 100% renewable energy, powered by Texas’ extensive wind and solar farms paired with Tesla Megapack energy storage systems, cutting the carbon footprint of its operations by 75% compared to traditional semiconductor fabs running on fossil fuel energy. Industry insiders noted that the project also aligns with the U.S. federal CHIPS and Science Act requirements, and the two firms are expected to apply for up to $2 billion in federal manufacturing subsidies in the coming quarter.
Featured Comments
As a semiconductor industry analyst covering North American supply chains, this move is a total game-changer. Elon Musk’s firms are cutting their reliance on third-party GPU suppliers at exactly the time when global AI chip demand is outstripping supply by 40% this year. Texas’ combination of low energy costs, friendly regulatory policies and proximity to both firms’ existing operations makes this site choice a no-brainer, and I expect we’ll see other tech firms follow suit to set up advanced manufacturing in the state over the next 5 years.
I own a commercial construction firm based in Austin, and we’ve already started getting inquiries for subcontracting work for this project. The 22,000+ jobs this plant will bring are going to lift the entire regional economy, from housing to food services to professional services. I’m just glad Texas’ pro-business environment is attracting projects of this scale that benefit regular working people, not just coastal tech executives.
As an AI research lead at a mid-sized tech firm, I’m really curious to see how these custom Tesla/SpaceX AI chips perform once they hit production. If they really deliver 30% better performance per watt than Nvidia’s H100 as rumored, they could upend the entire AI hardware market. It’s also smart that the plant is powered mostly by renewable energy from Texas’ wind and solar farms — AI training is incredibly energy intensive, so cutting the carbon footprint of chip manufacturing and operation is a huge win for sustainability too.