According to CNBC, Tesla CEO Elon Musk stated the company may need a “gigantic” semiconductor fabrication plant for its artificial intelligence and robotics initiatives.
Musk expressed this need at Tesla’s annual shareholders meeting on Thursday, noting the challenge in securing sufficient chips. He stated, “One of the things I’m trying to figure out is — how do we make enough chips?”
Tesla currently contracts Taiwan Semiconductor Manufacturing Company (TSMC) and Samsung Electronics for chip production. Musk also indicated consideration of a partnership with U.S. chip company Intel. He added, “But even when we extrapolate the best-case scenario for chip production from our suppliers, it’s still not enough.”
Musk anticipates a “gigantic” chip fab, which he termed a “Tesla terra fab,” stating, “I can’t see any other way to get to the volume of chips that we’re looking for.” Demand for microchips, powering various modern technologies from consumer electronics to data centers, has increased significantly due to the AI boom. Tech organizations, including Tesla, seek more supply from chipmakers like TSMC, which is the world’s largest and most advanced. Musk indicated Tesla’s prospective fab would initially target 100,000 wafer starts per month, with an eventual scale-up to 1 million. Wafer starts per month measure new chip production. In comparison, TSMC reported an annual wafer production capacity of 17 million in 2024, approximately 1.42 million wafer starts per month.
Tesla designs custom autonomous driving chips but does not yet manufacture them. The company currently outsources production of its “AI5” chip. Musk detailed this chip will be cheaper, power-efficient, and optimized for Tesla’s AI software.
Musk also announced that Tesla will commence production of its Cybercab, an autonomous electric vehicle without pedals or a steering wheel, in April. Musk characterized AI and robotics as fundamental to the global economy’s future. He noted, “With AI and robotics, you can actually increase the global economy by a factor of 10, or maybe 100. There’s not, like, an obvious limit.”
