Elon Musk has been building a massive wafer plant called Terafab, aiming for monthly capacity of up to a million wafers using manufacturing technology developed with Intel, a project widely read as an attempt to cut Tesla and SpaceX's heavy reliance on Taiwan Semiconductor Manufacturing Company. At a recent tech conference, Musk gave his own explanation for why Terafab needs to exist: "for some reason, at some point," he said, chips might stop arriving in the United States from Taiwan altogether.
Why Musk Says Terafab Has To Exist
Speaking at the All-In Summit, a conference hosted by a group of venture capitalists, Musk laid out his reasoning for building Terafab and gave an update on the research and development chip fab that Tesla and SpaceX are jointly building in Austin, Texas. "For some reason, at some point, chips may not continue coming here from Taiwan," Musk said. "That would really make things difficult if we didn't have any chips. That's a big part of why I'm building Terafab. We need certainty around future chip supply, even if the geopolitical situation gets challenging."
Elon Musk explains why Terafab must be built and gives an update on the Tesla and SpaceX R&D chip fab in Austin:
“For some reason, at some point, chips may not continue coming here from Taiwan. That would really make things difficult if we didn’t have any chips. That’s an… https://t.co/umN7gmv4uq pic.twitter.com/pZKJTrgjor
Build Terafab Or Fail To Scale, In Musk's Words
Musk argued that over the long run, manufacturing at scale is the real bottleneck. He said that if the goal is to genuinely scale artificial intelligence, whether that means servers, edge computing, humanoid robots or cars, every existing wafer fab worldwide, even running at full capacity, still would not produce enough chips to meet demand. "It's either build Terafab, or fail to scale. Those are the only two options," he said.
Austin Fab Aims For Something Useful By Next Year
Musk said Tesla and SpaceX are jointly building a research and development wafer fab in Austin and have already ordered all the equipment the project needs. "By the end of next year, we might be able to make something useful," he said. He compared the process to a child learning to move: first asking whether they can build anything at all, then moving on to producing chips that are actually useful, and only in the final "running" stage reaching true mass production.