Tech • AI • Robotics
New Elon Musk’s Terafab $119B Largest Building on Earth Shocked Inside!
9.4/10TL;DR
Tesla is advancing two unusually large manufacturing bets in Texas: a proposed Terafab chip complex meant to secure AI processor supply for Optimus, robotaxis and Starlink, and a next-generation vehicle production system built around modular assembly, giant castings and deeper automation.
KEY POINTS
Terafab’s scale
Elon Musk has described Terafab Texas as a 100 million-square-foot semiconductor campus, about 10 times the floor area of Gigafactory Texas. The project has been framed as a facility large enough to outscale any existing chip plant and produce integrated logic, memory and advanced packaging in one location rather than across multiple countries.
Texas tax deal and local footprint
Permit filings have been reported in Grimes County, Texas, and SpaceX has already paid the county $10 million ahead of schedule under a tax-abatement agreement. The arrangement also requires $20 million annually over 35 years, while the project is expected to create more than 3,000 jobs.
Why Tesla wants its own chips
The immediate rationale is supply. Tesla’s self-driving systems, robotaxis and Optimus humanoid robots all rely on heavy AI compute, and current suppliers cannot deliver enough chips to support the company’s ambitions at full scale. Musk’s broader claim is that AI chips, not vehicle engineering, are becoming the limiting factor.
Output targets are extraordinary
Musk has tied Terafab to a goal of 1 terawatt of AI computing capacity per year and annual production of 100 billion to 200 billion custom AI chips. Even allowing for aggressive assumptions, those figures imply an industrial expansion far beyond a conventional automotive supply project and place the proposal in direct comparison with global semiconductor leaders such as TSMC, Intel and Samsung.
Intel’s role and manufacturing expertise
Intel joined the effort in April, a notable step because leading-edge chipmaking depends on process discipline and yield management that carmakers do not possess. Gary Zhang, a veteran of Intel manufacturing operations and its 18A program, was named as Terafab’s first director, signaling that the venture is trying to import semiconductor know-how rather than improvise it.
Two chip families, two missions
One chip roadmap centers on Tesla’s AI 5 platform, with an initial ramp targeted for 2026 and broader production in 2027, followed by AI 6 and AI 7. A second family, identified as D3, is intended for Starlink and would need to tolerate radiation, vacuum and large temperature swings, making it a very different design problem from automotive AI silicon.
A high-risk industrial gamble
Building a state-of-the-art fab is far harder than scaling a battery line. Advanced nodes require ultra-clean rooms, expensive EUV lithography tools, tight chemical supply chains and yields where microscopic defects can destroy a wafer batch. That makes Terafab one of the riskiest manufacturing projects tied to Musk’s companies, even if it follows the same playbook Tesla used when it absorbed battery expertise and pushed into 4680 cells.
Parallel push in vehicle manufacturing
At the same time, Tesla is overhauling auto production around its so-called unboxed process. Instead of moving one nearly complete car through a long serial assembly line, major modules such as front and rear structures, body sections, interior and battery are built in parallel and merged later, with the aim of cutting factory footprint by 50% and reducing capital spending by billions.
Cybercab at the center of the new line
The new system is tailored to a simplified robotaxi often referred to as Cybercab, a two-seat vehicle without steering wheel or pedals. Fewer parts, less wiring and a simpler cabin make it better suited to modular assembly, but the challenge is synchronizing thousands of robots and material flows so each subassembly arrives at the final integration point within extremely tight tolerances.
Gigapress and paint-shop disruption
Musk has also floated a future 50,000-ton gigapress, far beyond today’s 6,000-ton and 9,000-ton machines, to produce very large castings at high speed. In parallel, Tesla is exploring reaction injection molding for exterior panels with color embedded in the material, which could reduce reliance on the energy-intensive paint shop, traditionally one of the most expensive and failure-prone parts of a car factory.
CONCLUSION
The common thread is vertical integration at an extreme scale: control the chips that power autonomy and robots, and redesign the factory that builds the vehicles using them. If both bets work, Texas could become a central manufacturing hub for Tesla’s push beyond cars into AI, robotics and high-volume autonomous transport.
Explain this