Elon Musk’s “Terafab” Could Change AI Forever

Kind: captions Language: en Welcome to the Elon Effect, your daily check-in with everything orbiting the worlds of SpaceX, Tesla, X, Neurolink, and the Boring Company. I'm your host, Miles Vega. Let's dive in. Elon Musk may have just unveiled the most ambitious project across all of his companies. And it's not a rocket, a car, or even a robot. It's a chip factory. Over the weekend in Austin, Musk introduced what he calls the Terrafab, a massive new initiative that would bring together Tesla, SpaceX, and XAI to build advanced AI chips entirely in-house. And when Musk says massive, he means it. The Terraab facility is planned for the north campus of Giga, Texas, and is expected to be even larger than the existing factory, already one of the biggest buildings on Earth. The goal to control every step of chip production under one roof from design to fabrication to memory and packaging. Musk says the reason is simple. The world isn't making enough chips fast enough. Right now, Tesla relies on manufacturers like TSMC and Samsung to produce its AI processors. But according to Musk, even the best case output from those suppliers won't meet the future demand from Tesla, SpaceX, and XAI. So instead of waiting, he's building his own. Terraab is expected to cost between 20 and 25 billion initially, with some estimates pushing total investment closer to 40 billion or more over time. And notably, that spending isn't even included in Tesla's current capital expenditure plans. So, what are all these chips for? Two main categories. First, edge AI chips designed for Tesla's full self-driving systems, robo taxi fleet, and most importantly, its Optimus humanoid robots. Second, high-performance chips designed for space, powering SpaceX satellites, orbital data centers, and XAI's growing AI infrastructure. And this is where the scale becomes almost hard to comprehend. According to analysts, Tesla's Optimus program alone could require tens of millions of chips per year. If Tesla ever reaches its long-term goal of producing 100 million robots annually, that number could jump to over 200 million chips per year. That's more than 50 times Tesla's current chip demand. To meet that kind of scale, Terraab is targeting cuttingedge 2nometer chip technology, the most advanced manufacturing process currently being developed. Musk's ultimate goal is even bigger, producing over a terowatt of AI computing power per year, with much of it eventually deployed in space. But make no mistake, this is an incredibly difficult undertaking. Building advanced semiconductor fabs is one of the most complex industrial challenges in the world, typically requiring decades of expertise and massive capital investment. Even optimistic projections suggest Terraab wouldn't produce meaningful chip output until at least 2028. And Musk himself has no direct background in semiconductor manufacturing. Still, if history tells us anything, it's that Musk tends to pursue projects that others initially think are impossible. Electric vehicles weren't supposed to scale. Reusable rockets weren't supposed to work. And now Musk is betting that vertically integrating AI chip production could unlock the next phase of growth across Tesla, SpaceX, and XAI. If he's right, Terraab could become one of the most important pieces of infrastructure in the future of artificial intelligence. If not, it could go down as one of the most expensive bets in tech history. Either way, the race for AI compute just entered a whole new phase, and Elon Musk wants to build it himself. That's your tech news for today. If you enjoyed this episode, please consider subscribing and leaving a quick review. The Elon effect is a Cerebra media production. I'm Miles Vega and I'll see you back here tomorrow. Take care.