Self-driving cars, cyborgs, and catching rockets in midair: Elon Musk can’t resist the lure of the impossible.
The world’s richest man has made a habit of taking on the world’s most difficult engineering challenges at Tesla and SpaceX — and has often proved the doubters wrong. His latest target is a tall order even by his standards.
For several months, Musk has been talking about building a “Terafab,” a mammoth factory that would churn out semiconductors critical for Tesla’s ambitious rollout of robotaxis and humanoid robots.
On Saturday, he teased that an announcement was imminent. “Terafab Project launches in 7 days,” Musk wrote in an X post, without providing further details.
In a January earnings call, the billionaire cited chip production as the major long-term headwind to the company’s growth, suggesting that output from suppliers Samsung, TSMC, and Micron would be nowhere near enough to meet Tesla’s targets as the EV giant scales its robotaxi and humanoid robot programs.
“This is definitely going to be sort of a controversial thing, but I think Tesla needs to build a Terafab,” Musk told investors, adding that such a facility would also protect Tesla against geopolitical upheaval.
The Tesla CEO suggested that the company would pursue the hardest possible version of that vision, a “very big fab” that would produce and package logic and memory chips entirely in the US.
Speaking at Tesla’s annual shareholder meeting last November, Musk estimated the Terafab would aim to initially produce 100,000 silicon wafers a month and could eventually grow to 1 million.
SpaceX/Getty Images
Ahead of the Terafab announcement, Tesla has begun laying the groundwork for Musk’s grand plan. The tech giant is hiring a semiconductor infrastructure manager to oversee factory design and construction, per a recent job posting. The role is based in Austin, suggesting the Terafab could be built near Tesla’s gigafactory on the outskirts of the city.
However, analysts told Business Insider that Tesla would face enormous challenges — and a huge bill — as it tries to master one of the most complex technologies on the planet.
“It’s Musk, so I would never count it out. But I suspect this is actually harder than sending rockets to Mars,” Stacy Rasgon, managing director and senior semiconductor analyst at Bernstein, told Business Insider.
Semi-impossible?
The global supply of semiconductors is almost entirely produced by a small handful of companies, many of them based in East Asia.
Manufacturing them is an expensive, complicated, and time-consuming process. Deep within hermetically sealed factories, chip designs are etched onto thin silicon wafers at the molecular level by specialist lithography machines, which are almost entirely made by one company in the Netherlands and can have a waitlist of over a year.
Rasgon said that procuring these in-demand ASML-built machines was a critical roadblock for any would-be chipmaker.
“If you’re a brand new customer, you’re probably waiting a couple of years before getting your hand on one of those,” he said.
Rasgon added that chipmakers usually split up production of logic and memory chips and semiconductor packaging across different factories.
Musk’s suggestion that Tesla could integrate them all into one facility would make scaling the Terafab even more complicated, Rasgon said, as each product has wildly different processes and economics.
: Jim West/UCG/Universal Images Group via Getty Images
Musk is not alone in fearing geopolitical disruption. The threat of a Chinese invasion of Taiwan, which would plunge global chip supply into chaos, has prompted companies, including TSMC, to build new chip fabs in the US.
But the road to US-produced semiconductors has been far from smooth. TSMC’s Arizona expansion has faced years of delays and a total price tag of around $165 billion across multiple facilities.
The industry runs on technical knowledge that is deeply embedded within leading companies. TSMC flew employees out from Taiwan to help the Arizona facility ramp production and brought US workers to its home country to train them.
The need for specialized knowledge will make recruitment critical for Tesla’s Terafab hopes, Rasgon said, adding that the semiconductor industry is already facing a worker shortage.
“These guys don’t grow on trees,” he said.
A ‘Herculean’ challenge
Analysts warned that overcoming these challenges would add to the severe cash burn Tesla is set to face in the coming years.
The company said in January it would spend $20 billion on building out its robotaxi and Optimus production lines this year, a figure which does not include the Terafab project.
Ben Kallo, a senior research analyst at Baird, told Business Insider that investors would have questions about how Tesla plans to fund such an ambitious project — especially considering Musk has also said Tesla will build around 100 gigawatts of solar panel manufacturing.
“Where’s the money coming from? I think that’s going to be a question,” said Kallo, who added that he wouldn’t rule out Tesla raising outside capital for the first time since 2020 to fulfill Musk’s ambitious targets.
Musk hasn’t given a specific timeline for building the Terafab and producing chips, but he said in the January earnings call that he was building it to “remove a probable constraint in three or four years.”
In a Tuesday note, Morgan Stanley analysts led by Andrew Percoco pointed to Micron’s factory in Boise — which began construction in late 2022 but isn’t expected to begin shipping chips until mid-2027 — as evidence of how long it can take to build semiconductor infrastructure in the US.
They estimated that building a factory capable of producing 100,000 wafers for cutting-edge logic chips a month could cost as much as $45 billion. A note from UBS analysts in January estimated that just getting to Musk’s initial production target of 100,000 silicon wafers a month would cost $30 billion.
“Even understanding Elon Musk’s history of doing difficult things, this seems like a Herculean task,” the Morgan Stanley analysts wrote.
