Elon Musk’s Terafab: What does this mean for the semi industry?

Investing.com -- SpaceX and Tesla Inc (NASDAQ:TSLA) plan to build two advanced chip plants at a large Austin, Texas site, CEO Elon Musk said Sunday, with one facility focused on powering vehicles and humanoid robots and the other targeting AI data centers in space.

The remarks came a day after Musk unveiled plans for “Terafab,” an advanced AI chip manufacturing complex in Austin.

“Terafab” marks one of the most ambitious attempts yet to reshape the semiconductor landscape, with plans to scale compute production to 1 terawatt annually—around 50 times current global output.

“While he intends to continue purchasing chips from existing suppliers, he notes that 1 TW is ~50x the current global compute supply (~20 GW), and (correctly) observes that those suppliers are, shall we say, hesitant to attempt a capacity addition of this magnitude,” Bernstein analyst Stacy Rasgon said in a note.

“Hence the requirement for Musk himself to attempt the “Terafab”,” he added.

The project is expected to begin with an advanced facility in Austin capable of producing logic, memory, packaging, and masks, enabling rapid design iteration. Tesla intends to focus on edge inference chips for vehicles and humanoid robots, alongside more advanced compute tailored for space applications.

Barclays analyst Dan Levy said that chips are now “truly the pillar for Tesla’s growth over the next decade+,” forming the backbone of its broader AI ambitions.

The scale of the effort is unprecedented. Bernstein estimates that delivering 1 terawatt of annual compute would require between 7 million and 18 million 300mm wafer starts per month, driven largely by demand for high-bandwidth memory (HBM).

This would equate to roughly 140 to 360 new leading-edge fabs and imply capital spending of $5 trillion to $13 trillion—levels comparable to, or exceeding, the current global semiconductor manufacturing base, Rasgon noted.

Musk said that roughly 80% of compute output would be for space and only 20% for terrestrial use. He also envisions producing around 1 billion chips annually, supported by large-scale infrastructure including solar power, launch capacity, and robotics.

However, both analysts raised questions about the feasibility of the project. Rasgon said “a true Terafab feels like a stretch,” particularly under current compute paradigms.

“Perhaps we might see partnerships at some point with other current manufacturers if Elon decides he cannot make it happen on his own? Or perhaps Elon has something more off-the-wall in mind to improve things,” he wrote.

Meanwhile, Barclays’ Levy described the project as a “show-me story, and expect to see much smaller-scale aspirations, at least in the near/mid-term,” pointing to execution risks such as limited manufacturing experience, technological complexity, and long equipment lead times.

“Our key question is how "real" Terafab plans are. The announcement in many ways reminds us of the grand targets Tesla provided at battery day 2020 (i.e. 3 TWh of battery capacity by 2030) which Tesla has not come close to achieving,” he said.

For the semiconductor industry, near-term implications appear limited. Rasgon said the announcement is unlikely to matter “beyond the hype” for now.

Still, if pursued, the buildout would represent a major demand driver for semiconductor equipment, with the analyst noting investors would want to “buy semicap (and buy it and buy it and buy it).”

“Perhaps Elon making his own chips might read negative for the current incumbents though we think in a world where compute is this strong any player is going to see far more upside than they could ever handle,” Rasgon continued.