Terafab EXPLODES from $20B to $119B: Musk's MASTER PLAN Revealed!

Kind: captions Language: en In March 2026, Musk announced Terafab with a stated figure of $20 billion. Less than 2 months later, official filings submitted to Grimes County, Texas, disclosed the actual number, $55 billion for the first phase alone, $119 billion in total. Why would a single project nearly sextuple in size within just a few weeks? Tesla is currently using chips from TSMC and Samsung. SpaceX depends on outside suppliers. So, what drove Musk to decide to build his own chip factory instead of continuing to buy like before? Let's dive right in. >> [clears throat] [music] >> Every chip factory on this planet, all of TSMC, Samsung, Intel, everything combined, is currently producing approximately 20 gigawatts of AI computing capacity per year. That number may sound large, but Musk said it plainly, it's only 2% of what Tesla and SpaceX need to fully realize their plans. So, what does Tesla need chips for on a scale that enormous? Most of us still think of Tesla as a car company, but in reality, Tesla is simultaneously running three extraordinarily chip-hungry operations at the same time. Full self-driving autonomous vehicles, the Optimus humanoid robot, and the Dojo supercomputer for AI training. Every Tesla vehicle that rolls off the line needs image processing chips, AI inference chips, and control chips. Every Optimus robot does, too. And Musk isn't talking about a few tens of thousands of robots, he's talking about 1 billion Optimus robots in the future. 1 billion, not 1 million. Now, that 2% figure starts to make sense. But the problem doesn't stop at sheer numbers. Even for the portion of chips Musk is currently able to buy, he doesn't truly control that supply. Tesla is dependent on TSMC in Taiwan and Samsung in South Korea. These are two locations sitting squarely in the most geopolitically sensitive region in the world right now. If anything were to happen in the Taiwan Strait, even just diplomatic tension, let alone armed conflict, Tesla's entire chip supply chain could be severed within a matter of weeks. Apple recognized that lesson and spent nearly a decade breaking free from Intel, designing its own M1 chip in-house. The result? The M1 MacBook Air, released in 2020, became one of the most highly acclaimed products in Apple's history. Dramatically superior performance while consuming significantly less power. Musk saw that. And he wants to do the same thing, but not over 10 years, and not just designing chips. He wants to manufacture them himself at a scale never before seen in history. That is exactly why Terafab was born. Not personal ambition, not ego, but the survival of an entire ecosystem that Musk is building. This is the part that most people skip over when reading the news. And that's why they don't understand why the number jumped from $20 billion to $119 billion dollars just a few weeks because Terafab is not one factory. Terafab is three entirely distinct phases with different scales, different objectives, and different budgets. And people lumped them all together into a single number and started making comparisons. No wonder everyone is confused. The first phase is a pilot facility at Giga Texas in Austin. And this is actually the most modest step in the entire plan. Tesla will build an experimental chip manufacturing facility right on the existing Gigafactory Texas campus at a cost of approximately 3 billion dollars and a capacity of only a few thousand wafers per month. That sounds small relative to the 119 billion dollar vision, but this is precisely the key differentiator Musk is going for. In conventional chip manufacturing, when you design a new chip and want to test it, you have to send the design to a factory in Taiwan, wait for them to produce it, wait for it to ship back, run tests, find bugs, send it back for revisions, then wait again. One cycle like that can take three to six months. And in a world where AI is changing week by week, three months is an enormous amount of time. Tesla's pilot facility will do everything inside the same building, chip design, wafer printing, packaging, testing, and then revisions right on the spot. The development cycle shrinks from months down to weeks. The first product designed for production here is the AI 5 chip, Tesla's fifth generation AI chip, projected for small-scale production in late 2026 and mass production in 2027. The phase that truly stunned the global tech world is the large-scale facility in Grimes County, Texas. On May 6th, 2026, official documentation was filed with the Grimes County Commissioner's Court. The phase one figure, 55 billion dollars. Total across all phases, 119 billion dollars. For comparison, the entirety of TSMC's project currently under construction in Arizona, previously considered the largest semiconductor project in US history, costs approximately 65 billion dollars. Terafab, if completed, would nearly double that figure. The division of roles here is quite clear, yet few people have paid attention to it. Tesla handles the research and development phase, while SpaceX takes full responsibility for this entire large-scale production phase. Musk stated it plainly during Tesla's Q1 earnings call. SpaceX will take on the initial phase of the large-scale Terafab. Tesla is the lab, SpaceX is the factory. That's the simplest way to understand this structure. The production target at full capacity is 1 million wafers per month, equivalent to 100 to 200 billion chips per year. And here's something that very few outlets have mentioned. On May 8th, 2026, just days before this video was made, aviation observer Joe Tegmeyer flew over the area and documented actual construction activity taking place on the eastern portion of the site. Materials being moved, trucks, equipment, River Road is being widened. Terafab is no longer just a promise on a presentation slide. The farthest reaching and most ambitious phase, the one that gives even long-time Musk watchers pause, involves chips specifically designed to operate in orbital space, called the D3 chip. These chips will be launched aboard Starship, housed in orbital data centers, and powered by solar energy 24/7. Why would we need data centers in space? Because Earth's power grid is becoming the bottleneck for AI. The world's largest data centers currently consume electricity equivalent to that of a mid-sized city. And in many places, there simply isn't enough power to expand further. Space is different. No grid limitations, no electricity bills, no national regulations. Just continuous sunlight and D3 chips running above it all. This is also why SpaceX acquired XAI. And XAI will now be renamed SpaceX AI. The full AI capability of the Grok model will run on this infrastructure. April 7th, 2026. Intel, the company that once dominated the chip industry for decades before gradually losing market share to Nvidia and AMD, posted a brief announcement on X. Intel is proud to join the Terrafab project alongside SpaceX, XAI, and Tesla. Intel's stock rose more than 3% that same day. And April 2026 became the best month in Intel's history. Its share price nearly doubled in a single month. Why did such a short announcement carry that much power? Because Intel is one of only three companies in the entire world capable of manufacturing chips below 5 nanometers at commercial scale. Those three companies are TSMC, Samsung, and Intel. Terafab wants to break free from dependence on TSMC and Samsung, but still needs advanced manufacturing technology. Intel became the most natural answer, and also the most strategically sound answer given the current geopolitical landscape. Because Intel is an American company manufacturing on American soil. More specifically, Terafab will use Intel's 18A process, the most advanced manufacturing node Intel has ever developed. Intel also brings something Terafab could not build on its own in a short time frame. EMIB chip packaging technology, which allows multiple smaller chips to be interconnected into a high-performance cluster, rather than fabricating one single large chip that is increasingly expensive and difficult to produce. TSMC has a comparable technology called CoWoS, but those production lines are currently running at full capacity because Nvidia has placed its entire order volume there. And no one else can get in. Analyst Adrian Sanchez from Yole Group offered a more grounded reading. Terafab is essentially an expansion of Intel's fab with Tesla, SpaceX, and xAI as anchor customers, rather than a fully independent manufacturing joint venture. That framing sounds less exciting, but it's actually good news for the project's viability. Because if Terafab truly had to build a fab from scratch, no equipment, no technology, no workforce, the story would be entirely different and the risks far greater. By building on Intel's existing infrastructure, Terafab shaves years off its deployment timeline. For Intel, this is also a lifeline arriving at exactly the right moment. CEO Lip Bu Tan made no effort to hide his excitement. Elon Musk has a proven track record of redefining entire industries. This is exactly what chip manufacturing needs. And for Intel, Terafab is what they've been searching for. An anchor customer large enough to prove that Intel's foundry division is not a failed venture. $119 billion already sounds enormous. But analysts estimate that to reach the full 1 terawatt computing capacity that Musk has set as the target, the actual cost could fall somewhere between $5 trillion and $13 trillion. Not billion, trillion. A figure larger than the entire GDP of Germany and larger than the entire US defense budget across 20 years combined. Chip analyst Ben Bajarin from Creative Strategies calls Terafab a 15-year strategy. Not 5 years, not 10 years as the headlines have been implying. And Musk is well known for missed deadlines. Starship took far longer than originally projected. The Cybertruck was delayed nearly 3 years. Full self-driving has been promised as complete next year, every year from 2016 to the present, and still isn't legally complete in any meaningful sense. Terafab will likely be no exception. But there is a technical problem even more serious than timelines, and almost no one is talking about it. To manufacture advanced chips, you need EUV, extreme ultraviolet lithography machines. And there is exactly one company in the world that makes them, ASML, a Dutch company. Each EUV machine costs approximately $200 million. ASML produced only 48 machines in all of 2025, and their entire order book has already been allocated to TSMC, Samsung, and Intel through at least 2027. Terafab, as of this point, has no publicly confirmed ASML orders on record. No EUV machines, no advanced chip manufacturing as planned. It's that simple. This is also precisely why leveraging Intel's existing infrastructure, which already has ASML equipment in place, becomes the most practical short-term solution. Beyond that, Tesla and SpaceX are two entirely independent companies, despite Elon Musk being CEO of both. Every coordinated decision must pass through two separate boards of directors and clear conflict of interest review processes. In the chip manufacturing industry, where speed of development is a core competitive advantage, organizational delays can cost more than any technical challenge. And the most noteworthy detail, in my view, is this. Within the actual Grimes County filing itself, a legal document, not a PR statement, SpaceX explicitly wrote that the company has no guarantee of achieving Terafab's targets within the projected time frame, and does not rule out the possibility of complete failure. The filing also acknowledges having no long-term contracts with many direct chip suppliers, and concedes that the company will remain significantly dependent on third parties for an extended period. This is not performative humility. This is mandatory legal language in filings for tax incentive applications and IPO preparation. When a company commits language like this to a legal document, they genuinely believe it could happen. There is one more detail that few people have connected. SpaceX is preparing for an IPO as soon as June 2026. The public S-1 filing is expected to be submitted between approximately May 18th and 22nd. The estimated valuation stands at roughly $1.75 trillion, targeting pro- ceeds of approximately $75 billion, which, if successful, would be the largest IPO in history, breaking Saudi Aramco's record. The timing of the Terafab filing in Grimes County, just days before the IPO window, is not a coincidence. Terafab is part of the story SpaceX needs to tell investors. "We don't just launch rockets. We are building chip infrastructure for the entire AI era." This means Terafab is simultaneously a real project and a brand positioning tool in service of the IPO. Those two goals are not mutually exclusive, but viewers need to understand that not every announcement reflects 100% of the technical reality behind it. So, after everything we've just walked through, what is the true picture of Terafab? For optimists, Terafab is proof that the private sector is doing what the government cannot do fast enough. The US is spending tens of billions of dollars through the CHIPS Act to bring chip manufacturing back home, but progress has been slow and weighed down by bureaucracy. Meanwhile, SpaceX, Tesla, and Intel sat down together and started building, even without enough equipment in place, even without enough contracts secured. That is a build first, figure it out later mindset, and it is precisely that mindset that has produced Silicon Valley's biggest breakthroughs over the past 20 years. For those who are more pragmatic, Terafab still raises more questions than it answers. $119 billion is an investment figure, not revenue, not profit. Chip fabs take years to build and more years to reach capacity, and the AI market is moving too fast for anyone to be certain that chips designed today will still be relevant 5 years from now. Nvidia is releasing new GPUs every year. AI architectures are changing constantly. Building a chip fab takes 5 to 7 years, but the market waits for no one. For skeptics, this is one of those projects that history may remember in one of two ways, either as the moment humanity truly began building chip infrastructure for the AI era, or as a textbook example of ambition outpacing execution capacity in an industry that demands the highest levels of patience and precision. But there is one thing I want to say plainly, and this is my personal perspective. Regardless of which side you're on, Terafab has accomplished something that very few projects manage. It has forced the entire world to seriously ask, "Who will control AI chip infrastructure over the next 20 years?" TSMC is in Taiwan. Samsung is in South Korea. Nvidia designs chips, but does not manufacture them. And until Terafab was announced, not a single private project had dared to set a goal of changing that structure. Musk once said something that I think is more fitting right now than at any other moment. We either build Terafab or we have no chips. And we need chips. That doesn't sound like a marketing statement. It sounds like someone who sees a wall and decides to go straight through it rather than around it. Whether Terafab is right or wrong, whether it succeeds or fails, the question it poses is the right question. Can the global chip industry expand fast enough to keep pace with the AI era? Or does humanity need to rebuild that entire foundation from the ground up? That is the question Terafab is attempting to answer with 119 billion dollars and the years ahead. On June 3rd, 2026, Grimes County will convene to decide whether to grant SpaceX a tax exemption. The first legal milestone for the large-scale Terafab site. Between May 18th and 22nd, SpaceX will file its public S-1 IPO prospectus. The first time the world will have access to the company's actual financials. And by the end of 2026, we will know whether the prototype fab at Giga Texas has actually succeeded in producing AI 5 chips. Those three milestones will tell us whether Terafab is history being written or just another page being turned. Tech Revolution exists because I believe everyone deserves to understand where the world of technology is heading. Whether you're an engineer or have never heard the word nanometer in your life. If this video helped you understand even one more thing then I did my job right. Subscribe so you don't miss the next updates on Terafab. Leave a comment if you think I missed an angle. This community grows because of you.