Inside Elon Musk's $119 Billion Terafab — Nothing Like This Has Ever Been Built

Kind: captions Language: en If you look closely at the plans for what is being built in the heart of Texas, your brain might just stop working. We are not talking about a new car factory or a rocket launchpad. We are talking about Terafab, a project so massive, so audacious, and so incredibly expensive that it defies logic. Elon Musk is planning to pour $119 billion into a single facility, a 100 million square foot mega structure designed to be the largest factory in human history. When it is done, it could produce up to 1 million wafers every single month. That is enough to shock the global semiconductor industry to its core. But, the most amazing part is not the size or the cost. It is a huge circle sitting right in the middle of the whole thing. Musk has hinted that this circle could hold technology that changes how advanced chips are made. It would move away from the way every other chip maker on Earth does it. To understand why Musk is pouring $119 billion into this giant place, you first need to understand the huge problem he is trying to fix. His companies are about to hit a wall. Without Terafab, they might not be able to keep growing. The story starts with four companies that seem totally different, but all share one giant headache. Elon Musk has several huge projects going on at the same time. They all point to a future that needs tons of computing power. Tesla is building the Cybercab, a car that drives itself with no steering wheel or pedals. Tesla is also building Optimus, a human-like robot that walks, lifts things, and helps people at home and at work. SpaceX is growing Starlink to bring internet to the whole planet. They also want to put smarter computers in space. Then, there is XAI, a company building huge AI models that need tons of computing power to learn and grow. Cars, robots, satellites, and AI seem like totally different businesses, but they all run into the exact same problem. They all need advanced [music] chips, and they need them in numbers that are hard to even picture. For a cybercab to drive safely on its own, it has to look at video from cameras right away. It has to spot people walking, read traffic lights, find lane lines, and react to a world that changes every single second. Think about how hard it is to drive in a busy city. You have to watch for kids running into the street, cars stopping suddenly, and weather changing the road. Now, imagine doing all of that with no human brain, just a computer making choices in a fraction of a second. Optimus, the robot, needs even more processing power. It has to balance on two legs, control dozens of motors, see objects, and talk to people without messing up. A human brain does all of this without thinking about it, but a robot has to calculate every single movement, every step, and every grip. It is like playing a video game at the highest difficulty level, except in real life with real consequences. If Tesla was only building a few thousand of these machines, buying chips from other companies might work. But Musk does not want just a few thousand. [music] He talks about building millions or even tens of millions of smart machines. When you add the computers needed for XAI and SpaceX, the need for chips jumps to a level the world has never seen. Imagine if every home in America suddenly had an Optimus robot. That alone would require more chips than all the phones and laptops combined. Musk has said that the whole world right now might only make about 2% of the AI chips his companies will need in the future. That means the problem cannot be fixed by just ordering more from companies like TSMC or Samsung. If a factory can only make 100 chips a month, you cannot buy 1,000 from them, no matter how much money you have. The machines to make them simply do not exist. It is like trying to buy 1,000 cars from a dealer who only gets 20 shipped to his lot each month. No amount of money changes that. As Musk put it, they either build their own giant chip factory or they do not get the chips they need. So, they decided to build Terafab. Terafab is not just a place to make simple chips. It is built to do everything under one roof. It will make special computing chips, fast memory systems, and advanced packaging. It will test the parts and get them ready for use. This turns the project from a normal factory into a whole city built for AI. The Cybercab and Optimus are the main reasons this factory has to exist. They're the products that will need more chips than anyone has ever made. Without Terafab, these products would wait years just to get the parts they need. But, knowing the demand only explains the why. To understand the how, you have to look at the size of this building. It is so big that every other factory on Earth looks like a tiny toy next to it. 100 million square feet of pure factory. When you hear the number 100 million square feet, it is hard to picture what that means. Think about it this way. Terafab is planned to be almost 10 times bigger than Gigafactory Texas. That factory is already one of the largest buildings in the world. It makes thousands of electric vehicles every week, and it takes hours just to walk through it. Now, imagine something 10 times that size. Once finished, [music] Terafab would be about 50 times bigger than the Pentagon. The Pentagon is one of the biggest office buildings on Earth and it has its own zip code. Terafab would make it look like a garden shed. If you put that into numbers people can understand, 100 million square feet is about 9.3 million square meters. [music] That is the same as 1,300 soccer fields. From the sky, this place would not look like a factory. It would look like a giant city built just computer chips. Picture 1,300 green soccer fields placed side by side in a giant grid. Now replace the grass with clean rooms, machines, and testing labs. That is what Terafab looks like from above. If a person stood at one end of Terafab and wanted to walk to the other side, it would not be a quick trip. The building is so long that getting across it might need a special transport system inside. It would need its own vehicles, roads inside the building, and spaces that feel more like a small town than a factory floor. You might need to ride a cart or a small electric vehicle just to get from the memory production area to the testing area. Musk has even said it will make some of America's biggest landmarks look small. The goal is not just to build another chip factory. The goal is to make something bigger than the world has ever seen. This giant size also explains why Terafab is expected to create more than 3,000 good jobs. A place this massive needs more than just a few engineers. It needs building crews, operations teams, shipping workers, repair staff, security guards, and many other workers to keep it running. It is like building a small town inside one building. The people working there might need their own cafeterias, rest areas, and maybe even housing nearby. But, the most important thing about this space is that it is not just for one production line. Terafab is designed as a system with many areas for making, testing, and finishing parts all inside the same walls. This turns the 100 million square feet into space that can grow in stages. Tesla does not have to build a new factory every time they need more chips. They can just open the next part of Terafab. The first plan might start smaller, but the big goal is to reach 1 million wafers per month. To make that many, the building needs space for huge clean rooms. A clean room is a special space where the air is filtered to remove every tiny speck of dust. Even one piece of dirt can ruin a tiny chip. It is like trying to paint a tiny portrait inside a dust storm. Workers wear special suits that cover their whole body, so hair or skin does not get in the air. The factory also needs room for machines, storage, packing areas, testing spots, and inside roads that run almost non-stop. The machines used to print chips are as big as school buses. Each one costs hundreds of millions of dollars. Terafab will need hundreds of these machines lined up in perfect rows. When you see the 100 million square-foot size, the 119 billion-dollar cost starts to make more sense. This is not just a big building meant to impress people. It is a project meant to keep growing for many years, maybe even decades. It is a living machine that gets bigger as the need for AI grows. But, the size of the building is only half the story. The real secret of Terafab is not how big it is. It is what is hiding in the very center of it. If you look at the plans for Terafab, one thing stands out more than anything else. In the middle of the huge complex, there is a giant circle. This could be one of the most costly and hardest parts to build in the whole project, yet it has gotten very little attention. When someone online asked Musk if there might be a particle accelerator under this area, he gave a very short answer. He wrote f e l f t, which means free electron laser for the win. If that hint is true, the circle could be the key to fixing one of the biggest problems in making advanced chips today. To make computer chips smaller and stronger, chip makers need to print super tiny patterns onto silicon wafers. They do this using light to print the patterns, kind of like how an old camera projects an image onto film. But instead of making a picture, they are making circuits so small that thousands of them could fit on the tip of a finger. To print the smallest possible parts, they need light with a very short wavelength. Today, the best chips are made using a system called e u v, which stands for extreme ultraviolet lithography. Modern EUV systems use light with a wavelength of about 13.5 nanometers. To make this light, a strong laser fires at tiny drops of melted tin 50,000 times every second. The laser hits the tin and turns it into super hot plasma. That plasma gives off the EUV light. The light then bounces through a set of mirrors, so perfect they are among the flattest objects ever made by humans. If you expanded one of these mirrors to the size of Earth, the biggest bump on it would be shorter than a person. This technology is amazing, but it has big problems, too. It uses a huge amount of electricity. The systems are very complex, and there is always a chance that debris from the tin could hit and damage the mirrors. Those mirrors cost hundreds of millions of dollars. A free electron laser does it in a totally different way. Instead of shooting lasers at tin, it speeds up electrons to almost the speed of light. These fast electrons go through a set of magnets called an undulator. As the electrons wiggle through the magnets, they give off a beam of light. This light can be controlled very precisely. It can be tuned to the exact wavelength needed. In theory, a strong free electron laser could make a clean, powerful EUV beam without the mess of exploding tin drops. It would skip the debris problem that threatens the delicate mirrors in current machines. What makes the Terafab idea truly special is not just using a new light source. It is turning that technology into shared stuff for the whole factory. In a normal chip factory, every single EUV machine needs its own light source built inside it. That is [music] like every house on your street having its own power plant in the basement. It works, but it is costly and takes up space. But if Terafab plans to make 1 million wafers a month, it would need a ton of these machines. Building a separate light source for every machine would be a huge problem. A central free electron laser could make a huge amount of EUV light and send it through pipes to many machines at the same time. It is like having one giant power plant that serves the whole neighborhood, instead of a tiny generator in every home. This would change the whole design of a chip factory. [music] Instead of buying 50 machines that each have their own light, Terafab could have one giant light source in the middle powering everything around it. This is why the circle is so important. It is not just a design choice. It is the heart of the whole factory. But, building a machine that powerful is also very risky. Building an advanced chip factory is nothing like building a car factory. It needs a level of care that is almost impossible to reach. That difficulty is exactly why Musk needs help from a very surprising place. Even if Terafab solves the problems of chip design and extreme ultraviolet light, a bigger challenge is waiting. A 100 million square foot place running huge machines 24 hours a day will need power that never stops. A chip factory can handle many kinds of delays, [music] but a power loss in the middle of making chips is a disaster. If the power drops for even a few seconds, it can ruin the wafers inside the machines. It can break equipment and cause huge losses of money. Imagine baking a cake and the oven turns off right in the middle. The cake is ruined and you have to start over. Now, imagine that cake costs $50,000 and takes 3 months to make. That is what a power outage feels like in a chip factory. Chip making needs super clean spaces, chemicals that are almost perfectly pure, and control at a level close to the size of atoms. In a chip factory, even a tiny mistake can destroy a whole batch of wafers. The process of printing tiny lines on silicon is so touchy that a shake from a truck driving by outside can ruin a chip. Some chip factories are built on giant shock absorbers just to stop vibrations from the ground. The air inside is filtered so thoroughly that a single particle of dust is considered a serious problem. The chemicals used have to be so pure that one tiny impurity can change how a chip works. The yield is is percentage of chips that actually work after they are made. It is the most important number in a factory. When you start a new factory, your yield might be very low. Most of the chips you make will be broken and you have to throw them away. It can take years of testing and fixing to get the yield high enough to make money. If your yield is only 50%, you are throwing away half of your costly materials and time. To make a profit, a factory needs a yield of 90% or higher. Getting there is one of the hardest things in the world. [music] Some of the biggest chip companies spent decades learning how to do it. Money cannot instantly buy decades of hands-on factory experience. That is why Terafab needs to work with people who have done this before. Reports say a former Intel leader with 17 years of experience is set to become the boss of the Terafab project. This person used to run Intel's 18A work, which is some of the most advanced chip making in the world. Intel has been making chips for over 50 years. They have seen every problem you can imagine from yield crashes to equipment failures that took months to fix. That kind of knowledge lives in the heads of the people who were there. Tesla brings large-scale building skills. SpaceX brings fast engineering and the ability to solve problems quickly. xAI brings a huge need for computing power, which means the factory will always have a buyer. But, the former Intel people bring the one thing none of those three companies have ever had. They bring real experience in taking an advanced chip process and making it work at a huge scale. Getting the yield up is a dark art. It means understanding how every chemical, every temperature, and every machine setting works with the silicon. It is not something you can learn from a book. You have to live in the factory, watch the machines, and fix the problems one by one. A tiny change in temperature can mean the difference between a chip that works and one that fails. The former Intel leader has spent 17 years learning these secrets. By bringing in someone who has already done this, Terafab has a chance to skip the years of painful trial and error that most new factories go through. This is the last piece of the puzzle. The need for chips is there. The 100 million square foot building gives the space. The free electron laser gives the light. The natural gas and battery plants [music] give the power. And the Intel leader gives the knowledge to actually make it work. Terafab is an attempt to build a totally new kind of industrial world. It controls everything from the power plant to the AI model running on the final chip. If Terafab works, it will not just change Tesla or SpaceX. It will change the whole global chip industry. It will prove that a giant central place can beat the old way of making chips. It is a 119 billion dollar bet on a future where AI, robots, and self-driving cars need more computing power than the world can make right now. The size of this dream is almost impossible to grasp. But the pieces are slowly coming together to make it real. Now, why on earth would a chip factory ever need to be that big? The answer is that Terafab is not really one factory. It is every single stage of making a chip crammed into the same complex. Normally, building an advanced processor is a relay race spread across the whole planet. One company designs it, another prints the circuit board, a third makes the memory, a fourth does the packaging, and a fifth handles the testing. These are often done in different countries, and chips are shipped back and forth at every step. Musk's plan is to take that entire scattered supply chain and pull it under one roof. Nothing leaves until it is finished. That means more speed, faster iterations, more control, and no waiting on a factory on the other side of the world. Now, the reason why comes down to one word that Musk has been repeating over and over again. Terafab is being built to produce over 1 terawatt of AI compute every year. That's double the entire annual chip consumption of the United States today. Now, what could possibly need that many chips? The answer is basically Musk's entire vision of the future. Tesla wants to build literally billions of Optimus robots, and every one of them needs a brain. It wants fleets upon fleets of cyber cabs driving themselves around the cities, and self-driving demands an enormous amount of onboard computing. SpaceX wants to turn its satellites into data centers in orbit running AI in space. Every one of those ideas is starving for chips, and there simply are not enough being made in the world to feed all of them at once. So, instead of standing in line behind everyone else, Musk decided to build the line itself. And the final preparations are already underway. Intel has joined the project as a partner. Filings show that the plant will pull its water from a nearby reservoir instead of competing with local communities for ground water, and it is expected to create at least 3,000 jobs in the surrounding Texas counties. This is the America that Innovation Core is watching closely. We are witnessing the greatest industrial mobilization in modern history, a $119 billion bet on American soil, a bet that we can out-build, out-engineer, and out-manufacture the rest of the world. The era of relying on foreign supply chains for our most critical technology is ending. From the natural gas turbines powering the clean rooms to the Intel engineers teaching the robots how to etch atoms, this is a symphony of American grit and genius. We are not just building a factory, we are building the brain of the 21st century. And as we watch these 6,000 acres of Texas dirt transform into a city of the future, one thing is clear. The future belongs to those who build it. Thank you for joining us here on Innovation Core. If you believe in American ingenuity, hit that like button, subscribe, and turn on notifications. Because the next industrial revolution is happening right now, and we are just getting started. The next industrial revolution is already underway.