Tesla has plans for two new giant factories in Texas. Let’s have a look. Tesla’s “Project Crystal Sun” Solar Factory Tesla is reportedly planning a giant solar cell factory in Texas. The factory, named “Project Crystal Sun” within Tesla at the moment, is expected to require a $10.1 billion investment — ~$1.5 billion for “real property” and ~$8.6 billion in “personal property.” See some of the documents below prepared for Tesla by Kroll Tax Services. The project would be located about 40 minutes from Houston, near Richmond, Texas. Tesla has long had a penchant for vertical integration, including integrating as much as possible at one site. This is not any kind of revolutionary vertical integration, but in the face of continual tariffs under Trump and previously under President Joe Biden, the company has apparently decided it’s time to bring more of its solar cell production to the US. Instead of just putting solar cells together, the project would include “everything under the Sun” (as they say) when it comes to solar cell production — wafer manufacturing, ingot manufacturing, coating, metallization and printing, cell testing, automation and material handling systems, cleanroom infrastructure, chemical storage, etc. Tesla is forecasting that the “Project Crystal Sun” project will create 9,712 permanent jobs at full operation, while also creating 1,147 peak construction jobs. Construction is expected to go from 2026 through 2028. Commercial operations are expected to begin in Q1 2029. All of that said, the company is still considering another location outside of Texas. Or it just says that it is in order to try to negotiate a better deal with the state. Tesla already produces Megapacks in Brookshire, Texas. Tesla & SpaceX TERAFAB Project Then there’s the ginormous TERAFAB project Tesla and SpaceX are planning in Texas as well. As we’ve reported previously, there’s been an “insane” shortage of computer equipment due to the tremendous AI infrastructure boom, even leading to rising Apple computer costs and all kinds of challenges for companies that rely on buying a lot of computers. Tesla and SpaceX have thus decided they need to build their own computing hardware. As big as Project Crystal Sun is, this is about two to two and a half times bigger — in terms of investment at least. But to see how huge it actually would be, here’s a graphic my friend David Havasi sent me: Elon Musk calls this “the most epic chip building exercise in history by far.” It is a collaboration between Tesla and SpaceX. “The initiative aims to produce over one terawatt of AI compute annually, dwarfing the global industry’s current output of roughly 20 gigawatts per year. Musk framed the effort as ‘the next step towards becoming a galactic civilization,’ positioning it as essential for scaling humanity into a multi-planetary species,” Teslarati summarizes. “We’re very grateful to our existing supply chain … but there’s a maximum rate at which they’re comfortable expanding. We either build the TERAFAB or we don’t have the chips, and we need the chips, so we build the TERAFAB,” Elon Musk added. A few issues come to mind for me here. First of all, there’s the potential AI bubble many of us have been writing about. See here, here, here, here, and here. Do we really need this enormous expansion in chips/computer hardware? Recall that Grok (part of xAI, now under SpaceX) ended up having so much less demand and the company overbuilt data center capacity so much that Musk had to resort to renting capacity to Anthropic, a sworn enemy at the time. It’s very easy to see AI/LLM data center expansion increasing to a point — already — that is financially unsustainable and illogical. Again, see the articles linked above. Secondly, a lot of this is apparently for Tesla’s humanoid robots that haven’t hit the market yet. “Musk outlined annual output targets, which are between 100 and 200 gigawatts of terrestrial compute for robotics, supporting Musk’s vision of producing 1-10 billion Optimus units per year, and the majority (80%) of chips dedicated to orbital AI data centers. Overall, TERAFAB aims to produce 100-200 billion custom AI and memory chips each year.” Hmm…. Yeah…. Let’s see if there’s actually going to be demand for all of these robots. Musk has been talking about an age of abundance lately, that these robots and AI will do all the work and humans can reap all the benefits, to the point there will no longer be any need for money. Is Tesla going to, then, just hand out the robots for free? Are they really going to be able to do all manner of human work when. Tesla is still having humans virtually control them for snazzy demos because of how little they can actually do? Is this really the time to pump billions of dollars into production capacity for a product that isn’t ready and that doesn’t have clear, known customer demand? Another thing that comes to mind is how much the AI data center and investment needs for “Full Self Driving” has grown. There’s been much criticism over the years of how much Waymo’s lidar sensors cost, but do we see Waymo having to build a massive chip factory and ginormous data centers to implement the actual self-driving robotaxis it has deployed around the United States? How much investment does Tesla have to keep making to try to get FSD to a level where it can really live up to Elon Musk’s dreams and hype? With the TERAFAB project, maybe it all works out and is akin to Tesla’s original “gigafactory” in some ways, but I can’t help thinking of the Titanic at the moment, and Icarus. Since our in-house sage Steve Hanley first brought the idea to my mind, I have thought that Icarus is the perfect comparison for Elon Musk. Would it be ironic, or simply destiny, if one of the things that caused a massive crashout for Musk and his empire was a giant TERAFAB project meant to power data centers in outer space? I mean, it almost feels like a story written in Hollywood. Musk describes TERAFAB as “the final missing piece of the puzzle.” That does sound like something destined to fail, at least going by many movies and real-life stories I’ve seen play out numerous times. We shall see.
+20 others
