Elon Musk has announced plans to build a semiconductor manufacturing facility that would serve both Tesla and SpaceX, according to TechCrunch AI. Musk revealed the project at an event in downtown Austin, Texas on Saturday night, calling the planned facility “Terafab” and indicating it would be built near Tesla’s existing Austin headquarters and gigafactory.
The reasoning is straightforward: Musk says current chipmakers can’t keep up with his companies’ demand for AI and robotics chips. “We either build the Terafab or we don’t have the chips, and we need the chips, so we build the Terafab,” he said.
What Musk is targeting
The numbers he threw out are staggering:
- 100 to 200 gigawatts of computing power per year manufactured on Earth
- A terawatt of computing capacity in space
- Chips designed specifically for AI and robotics workloads across Tesla and SpaceX
No timeline was provided for any of this.
Why this matters
This announcement sits at the intersection of two major trends: the global semiconductor shortage that’s been squeezing AI companies, and the growing push by major tech players to vertically integrate their chip supply chains. Apple designs its own silicon. Google has its TPUs. Amazon builds Graviton and Trainium. Microsoft is working on Maia.
Musk entering this space isn’t surprising given the trajectory. Tesla already designs its own inference chips for Full Self-Driving, and xAI (Musk’s AI company) has been building out massive GPU clusters for training. The demand is real. But there’s a massive gap between designing chips and actually manufacturing them.
The reality check
Building a chip fab is one of the most capital-intensive, technically demanding endeavors in modern industry. TSMC spent over $65 billion on its Arizona facilities alone. Intel has poured tens of billions into new plants. These companies have decades of manufacturing expertise.
As Bloomberg noted in its reporting, Musk doesn’t have a background in semiconductor manufacturing. He also has a well-documented history of overpromising on goals and timelines. Remember the Robotaxi fleet that was supposed to arrive years ago? The million-mile battery? Full autonomy by 2020?
That said, people also doubted SpaceX could land reusable rockets. The track record is mixed, which makes predictions tricky.
What to watch
The key questions going forward:
- Funding: Who pays for this? A modern chip fab costs $20-40 billion minimum. Will this be a Tesla/SpaceX joint venture, or will outside capital be involved?
- Technology partner: Will Musk license existing manufacturing processes or attempt to develop proprietary ones? Building from scratch would add years.
- Timeline: The absence of any target date is telling. Even with unlimited money, a cutting-edge fab takes 3-5 years to build and ramp.
- Government support: The CHIPS Act has been funding domestic semiconductor production. Would Terafab qualify?
The ambition is classic Musk: enormous scale, vertical integration, and a stated urgency that borders on existential. Whether it materializes as described is another question entirely. For more details, check out the full report on TechCrunch AI.