Tesla and SpaceX Now Depend on One Impossible Factory

Kind: captions Language: en Tesla and SpaceX had been quietly building land in Texas and on August 6th, 2026, they showed everyone why. One factory, 100 million square feet. That's about 15 Pentagons and on August 13th, the machines arrived because they're already building it. Every robot, every self-driving car, every satellite Elon Musk has promised you runs on a chip that nobody makes yet and this one factory is where those chips come from and they're committing tens of billions to a factory nobody has built before, nothing this size has ever been attempted and nobody has ever made a chip start to finish in one place. But Elon is going for both at once, aiming for the impossible. So, let me show you what goes inside it and why two companies that build cars and rockets need a factory this size. They call it Terafab, but the size is the least interesting thing about it. The buying started this June when the chief financial officer of SpaceX signed an agreement with a rural county in Texas and committed the company to a payment of $10 million it doesn't get back. Three weeks after the agreement, Tesla hired the man who had been running Intel's most advanced chip factory. Meanwhile, SpaceX bought the land, 2,700 acres of Texas farmland and then it bought the water, the pumps, the intake and the rights to draw from the reservoir on the edge of it. That land is in Grimes County, Texas and what's going up on it is phase one of 10. There's also a second building and that one's going up outside Austin. It's much smaller, ground is broken, the equipment is already ordered and it has a very specific job that I'll come back to because it's the strangest part of this whole plan. Both buildings are part of the same giant project called Terafab. To put it simply, it's a chip factory but not a factory that simply puts chips into things, a factory that makes the chips themselves out of sand and light in rooms so clean that one speck of dust can ruin the entire process. There are only a handful of places on Earth that can do this at the level we're talking about and building a new one is one of the hardest things a company can attempt and this isn't Tesla doing it on its own. It's Tesla and SpaceX together with Elon Musk artificial intelligence company. And in April 2026, Intel joined them publicly. The chief executive of Intel said there's quote no better partner than Elon Musk. So, two of Elon Musk's companies plus one of the oldest chip makers in America all pointed at a single building in a Texas field, which leaves the obvious question, why does it have to be so big? The biggest factory Tesla has ever built is the one outside Austin. By floor area, it's the one of the largest buildings in the United States. The kind of place where you drive along the outside of it for a minute and you're still driving past the same building. Terafab is 10 of them. Not 10% bigger, 10 times. About 100 million square feet across thousands of acres on a site that needs more than 10 gigawatts of power, which isn't factory power. That's the electricity supply of a small country for one building. And this is where Musk says the quiet part out loud. Terafab Texas will be the largest and the most valuable building on Earth by far. But no factory this size has ever been built. Not by Tesla, not by Intel, not by anybody. The largest chip factories in the world took decades and tens of billions of dollars to reach a fraction of this. That's not a footnote. That's the whole risk of the thing. So, what could possibly need a building this size? If you're liking this video, a genuine subscribe would mean the world to me. Thank you so much and now let's get back to it. The reason it's that big has almost nothing to do with construction. It's about the one small part and everything Elon Musk has ever promised you happens to need it. Think about the three things this man is most famous for promising. A car that drives itself, a robot that works in a factory or a house, and satellites that do the thinking for artificial intelligence in orbit about the weather. Three completely different products, almost three different companies, and all of them run on the same thing. Not a chip you can buy off a shelf, a specific kind in quantities nobody has ever made. At the announcement, Elon put the numbers on a screen. The world's total output of this kind of computing power is running at roughly 20 gigawatts a year. And here's what he says he needs against that. >> So, if you add up all the fabs on Earth combined, they're only about 2% of what we need for the for the terawatt project. >> Every chip factory on the planet added together, 2%. He's not being rude about the companies that makes those chips. He said the opposite, that he's grateful to them, that he'd like them to expand as fast as they possibly can, and that he'd buy every chip that they can make. The problem isn't that they're slow, it's that there's a speed beyond which they won't go, and it's nowhere near what he's asking for. >> We either build the terafab or we don't have the chips. And uh we need the chips, so we're going to build a terafab. >> Which leaves only one option. And on Tesla's own earnings call, an analyst asked him whether terafab was really about squeezing his chip suppliers. >> Just literally we we don't see a path to having enough uh especially quantity of AI chips down the road as we scale production to high levels. >> Which brings me back to that second building, the small one outside Austin. A chip is made by printing patterns onto silicon using a mask. Think of it as a stencil. Normally, the company that makes the stencil, the company that prints the chip, and the company that packages it are three different businesses in three different countries. And a change takes months to go around that loop. In Austin, they're putting all of it inside one building. So, design a chip, cut the mask, print it, test it, and start the next version in under a week. So, how many places on Earth already work like that? >> Uh to the best of my knowledge, this doesn't exist anywhere in the world. >> And then there's the light. To print those patterns, you need a very extreme kind of light, and generating it is one of the hardest problem in manufacturing. It's done by firing a laser at a droplet of molten tin tens of thousands of time a second. And here's the craziest part. Somebody looked at the terafab site plans and noticed a structure that looked less like a factory and more like a particle accelerator. They asked Elon whether that was what it was. He replied with three letters, F E L, a free electron laser. So, they're not planning to buy a better version of the machine everyone else uses. They're going to build a small particle accelerator inside the factory and use it to light the whole site. Not one factory on Earth makes chips that way. It's been an idea in physics paper and never production line. So, that's three things nobody has done. A chip factory this size, a whole chip line in one room, and light out of a particle accelerator. They're doing all three at once and on the first attempt. So, all of these, the land, the water, the 10 gigawatts, the accelerator, looks like it's for the robots and the cars. Well, mostly it isn't. Elon Musk was asked directly how the output of this factory would be split. His answer, roughly 1/4 of it for the robot and about 3/4 for spacecraft. 3/4 on the biggest building ever attempted is being built for machines that won't be on this planet. And the size finally makes sense because the limit was never really the silicon. It was the electricity. The target for this project is a terawatt of computing power a year, a trillion watts. The entire United States, every home, every factory, every data center, every city runs on about half of that. So, the chips can be made here. They just can't all be switched on here. There isn't enough power on the ground and no plausible way to build enough of it in time. But in orbit, there's sunlight constantly with no night, no weather, and nobody to object to it. So, build the chips on Earth and run most of them off the planet. The first of those satellites aren't waiting for any of these. They're already being built on existing production lines with other companies' chips. Terafab isn't what starts the plan. It's what makes the plan big. Elon Musk announces a lot of things. This one is already being paid for. The first phase alone is put at about $16.8 billion from SpaceX and Tesla together. At least 3,000 jobs, most of them hired locally. A chip factory now sits inside Tesla's own capital spending guidance. And Tesla's finance chief has said that spending grows for the next 2 to 3 years. That's why this is a bet rather than a purchase. Every dollar going into this building is a dollar not going into the robotaxi fleet or into robot production capacity. Both companies' boards have to sign off on the arrangement between them. And the expertise is being bought, not grown. Job listings across Taiwan and Korea and a process veteran hired away from Intel to run it. That's the fastest way to get good people. It's not the same as having done it before. So, if you own either of these companies, and none of this is financial advice, remember what Elon said. The thing holding back the robot and the self-driving car isn't and it isn't demand. It's supply of one component. Owning the factory turns waiting in somebody else's queue into your own production rate. If it works, the ceiling comes off. If it slips, they've spent years and tens of billions building the ceiling themselves. So, the size was never the point. He's not trying to build a better chip company. He's trying to make sure that when the cars and the robots and the satellites all need a brain at the same time, he's the only one who has any. And if you want the story of what those satellites actually are, I made a whole video on it and it's right here. I'll see you there.