Is Terafab Genius or Madness? Elon Musk's $25 Billion Plan

Kind: captions Language: en 10 9 ignition sequence start. 6 5 4 3 2 1 zero. All engine running. Liftoff. >> On July 16th, 1969, millions of Americans gathered around their televisions to watch something unprecedented. A rocket carrying three men and a dream lifted off and aimed for the moon. Many said it was impossible. They claimed the math didn't add up and the technology didn't exist. They were wrong. Elon Musk often does the impossible. He created electric cars when the auto industry said nobody wanted them. He landed rockets vertically when aerospace engineers called it a fantasy. He did it again and again. But on the night of March 21st, 2026, Musk walked onto a stage and announced something that made even his biggest supporters stop in their tracks. He called it Terraab. At its core, Terraab is a semiconductor factory or Chip Fab. But calling it that is like calling the Apollo rocket a metal tube. The scale of what Musk describes is almost mind-boggling. It's a joint venture between Tesla, Spac X, and XAI. It will be built right next to Giga Texas in a building that's already planned to be one of the largest structures on Earth. The initial budget is between 20 and 25 billion. The goal is to produce more than a terowatt of AI computing power each year. For context, that's more than all chip manufacturers in the world combined can produce today or even by 2030. There will be two types of chips. One will be for Earth, powering Tesla's self-driving cars and Optimus robots. The other will be for space, fueling AI data centers in low Earth orbit, relying on pure solar energy. Musk put it simply. We either build the Terra Fab or we don't have the chips. And we need the chips, so we build the Terra Fab. But why does Musk need this many chips? The answer is robots. Musk estimates that Optimus, Tesla's humanoid robot, could eventually be produced at 10 to 100 times the volume of Teslas. Think about that. Tesla makes over a million cars a year. Musk is talking about millions of robots, each needing a chip to function. Then there's the AI side. Every Tesla on the road, every robo taxi, every query on X runs on computing power. Right now, the world simply doesn't make enough of it. Intel, one of the largest tech companies, has already signed on, calling Terraab's goal of a terowatt of computing power each year, a way to speed up advances in AI and robotics. The chips will use two nanometer process technology, the most advanced manufacturing method available. Even industry giants like TSMC are just starting to master it. However, not everyone is convinced. Some of the brightest minds in semiconductors think Musk is aiming too high. It's estimated that producing a terowatt of AI silicon each year would require over 100 state-of-the-art fabrication facilities, each costing about $30 billion. That totals around 3 to5 trillion. That's more than 70% of the entire US government's annual budget. Building a single chip fab typically takes around 5 to 10 years and requires billions in capital. Sourcing the equipment is nearly impossible. ASML, the company that produces the lithography systems required for cuttingedge chip manufacturing, ships only a limited number of these highly specialized machines each year. Terrafab would need many of them. Musk also lacks experience in semiconductor manufacturing, and his history of ambitious timelines makes the industry skeptical. But here's the thing about Elon Musk and those who say something can't be done. The end goal for Terrafab isn't only cheaper AI. It's a galactic civilization. Musk envisions Terraab chips powering solardriven AI satellites in low Earth orbit. These satellites would send intelligence to a planet where robots manage all labor, scarcity vanishes, and every need can be met. And from there, it would expand to the moon, Mars, and beyond. It sounds like science fiction, and Musk knows it. He mentioned Isaac Azimoff and Yain Banks by name. We want to make that real, he said. Whether you believe him or not, it's clear that the age of AI is a chip war. On July 16th, 1969, they launched a rocket that many thought would never fly. 57 years later, the man who built the next generation of rockets is betting that the key to humanity's future isn't a spacecraft. It's a chip the width of two atoms built on a scale the world has never seen. Maybe he's right. Maybe he's wrong, but if history has taught us anything about Elon Musk, it's that he's worth paying attention to.