Terafab: Elon's Most Insane Project Yet 🏭🔥

Kind: captions Language: en Elon Musk just unveiled a $25 billion chip factory joint venture between Tesla, SpaceX, and XAI, targeting 1 terowatt of AI compute per year, more than every chip company on Earth combined, and 80% of it will be sent to space. This isn't a press release. Intel just joined this week. It's real and it might change everything. Terrafab was officially unveiled on March 21st at the defunct Seaholome power plant in Austin, Texas. Musk called it the most epic chip building exercise in history by far. And for once, the hyperbole might actually be justified. The project is a joint venture between three of Musk's companies, Tesla, SpaceX, and XAI, which SpaceX recently acquired in an all stock deal. Together, they're building a vertically integrated semiconductor facility that combines every stage of chip production under one roof. Design, lithography, fabrication, memory, advanced packaging, testing, all-in-one building on the north campus of Giga, Texas. The target is 2 nanmter process technology. The most advanced node currently entering commercial production. Initial output 100,000 wafer starts per month. Full-scale target 1 million wafer starts per month producing between 100 and200 billion custom AI and memory chips per year. Estimated cost 20 to$25 billion. And it's not even in Tesla's record 20 billion capex budget for 2026. This is on top of everything else. Musk's rationale is simple. The global chip industry can't scale fast enough to meet Tesla and SpaceX's future demand. On the Q4 earnings call in January, Musk told investors that Tesla faces a supply constraint within 3 to four years. Samsung, TSMC, and Micron, his current suppliers, can only expand so fast. Musk claimed that every chip fab currently operating on Earth produces only about 2% of what his companies will eventually need. Optimus is the biggest driver. Morgan Stanley estimates Giga Texas alone will need capacity for 10 million humanoid robots per year, which requires 20 million chips. That's six times Tesla's entire current auto chip demand. If Tesla hits its 100 million robot target, it needs over 200 million chips annually. Then there's Dojo for full self-driving training, Grock for XAI, Cyber Cab for autonomous taxis, and the piece that breaks most people's brains, space-based AI data centers. Here's the part that sounds insane until you understand the math. Musk said 80% of Terrafab's compute output will be directed towards space-based orbital AI satellites, not data centers on Earth, data centers in orbit. The argument solar radiance in space is roughly five times greater than at Earth's surface. Heat rejection in vacuum scales differently, and once Starship is flying regularly, launching a satellite data center becomes cheaper than building one on the ground with grid power. SpaceX and XAI are betting that within 2 to 3 years, running AI workloads in orbit will be cheaper than running them terrestrially. Sam Alman called the idea ridiculous. Experts are divided, but Musk is building the chip factory to fuel it. That tells you how serious he is. And here's the development that changed everything. Just this week, Intel officially joined the Terapab project. Intel CEO Lip Bhutan hosted Musk at Intel facilities and confirmed the partnership publicly. Intel's own language on X was telling. They said their ability to design, fabricate, and package chips will help accelerate Terrafab's 1 terowatt goal. That's not a supplier relationship. That's Intel running the fab. Tesla and SpaceX finance and own a facility. Intel builds and operates it. It's essentially a co-anchored Intel foundry expansion rebranded as Terrafab. And for Intel, this is huge. The company has been hunting for Ankor customers to support its foundry business. And it just landed two of the most valuable and fast growing companies on Earth as clients. Intel stock jumped 3% on the news. The foundry thesis just got a massive credibility boost. So, what's the chaos capital take? Terrafab is Musk's most ambitious infrastructure bet since Gigafactory 1. The timeline is aggressive. The cost is staggering and the execution risk is real. This is the same playbook as battery day 2020, which promised a revolution that took 5 years longer than expected. But the strategic logic is undeniable. If Tesla produces 100 million robots, if SpaceX deploys space-based data centers, if XAI needs frontier compute at scale, external suppliers will never keep up. Owning the silicon stack is the only path for exposure. Tesla benefits from vertical integration if it works. SpaceX benefits even more through the satellite angle. Intel just became a legitimate AI infrastructure play again. And the entire semiconductor supply chain from ASML to lamb research to applied materials benefits from another massive fab buildout. Elon Musk doesn't do small bets. Terrafab is the most audacious physical infrastructure project announced this decade. Whether it ends in another battery day disappointment or the foundation of an orbital AI empire, you'll want to be paying attention. The capital markets are chaos. Embrace it and I'll see you in the next