Something Big Is Happening With Terafab

Kind: captions Language: en I think we have to assume it's going to be quite challenging. Um, and I'm going to be interested to see how much of this supply chain SpaceX and Tesla decide to vertically integrate just like they just like Tesla's done on the EV side. >> Chips, chips, chips. As Ryan from the Tesla owners of Southwest Washington asked Elon himself, "What are we going to do? We do not have enough chips. They do not exist. What shall we do?" And the answer is, "We're going to make our own." That's crazy. That's impossible. [music] It can't happen. Uh, but it looks like it just might. I want to discover some of the questions that you guys have. I want to get into the answers. I want to figure out, is this even possible? Is this a good idea? Or is it a very expensive bluff andor fool's errand? Well, if I want to have someone tell me what a fool looks like, I bring on Cern Basher, who just points. He's just like, "It's it's this guy. It's this guy over here." And uh, yeah. No, not you. Me. Anyway, I'm Brian. Welcome to Futurazza. [singing] >> So CERN, uh, we've talked at different times about the about the chips and I think we were both there at that shareholder meeting. I went and looked through my selfies. I don't know if I got one. I maybe it's on your phone. I don't know. At any rate, uh, chips when I first heard it, I said, "Well, unfortunately, it's impossible. There is only so many companies that can make chips of this caliber specifically like two or three and only two of them are actually doing it at the time. Intel had stepped back from it and it is that hard. It is that big of a challenge. But then they partnered with Intel and things started to look more interesting. Specifically, I thought that originally it was just a threat to get competition back in gear and to that end it did work. But now it appears that they're actually moving forward with it. What's going on? Tell me all about these chips, man. What are we doing here? >> Well, this is a project like no other. Um, the Terra Fabe is what you're speaking about is very ambitious and like you said, it's born out of necessity and when something is born out of necessity, it sort of forces you to like Elon does with everything is focus on first principles to try to solve the problem. Uh, so there's a number of things we can look at here in terms of the roles that they're hiring already hiring for. uh roles for terra fab based in Austin. Um one of them I think is in Palo Alto but of the 35 I think 34 of them are Austinbased. Um the ambition is absolutely enormous. The fact I think that they will also line up I think some pretty important players. You mentioned Intel. Um I think I personally think that Nvidia would be a fool not to be involved. Uh so we'll see how that develops. Also, I would say that um I watched last night a CNBC special on uh SKH Heinix, South Korean company that builds memory, and they literally had the government move mountains, move a mountain to create enough flat land to build their uh next fabs that they're they're going to invest $400 billion dollars into. >> Wow. >> To create more memory chips. And what was interesting is when the chairman of SKHEX was asked about expanding into the United States, he seemed very skeptical about the ability for them to do that. They're open to it. Of course, that's the right answer to say that they're open. But when pressed a little bit, he seemed to indicate that he didn't feel that the United States had the talent, the people, the knowledge, the knowhow to build memory chips. >> Is he wrong? >> That remains to be seen. Well, historically, I mean, the US has built chips for a long time. Intel has had its fab in the US since forever, but they fell out of first place. They were doing really well for a long time. The NBA's got a little too ambitious in their involvement and uh made some mistakes that wound up costing them more over the long term. The first step to all of this, though, is answer me this question. I think it's a pretty easy one. Did Tesla and SpaceX's partnership uh announcement that they're going to get into cutting edge chip fabrication spur the competition to increase their supply? >> I think it if it hasn't yet, I think it will. >> They both announced Samsung and TSMC announced additional expansions in the US for their fabs. That to me is already a win. >> That's right. And by the way, uh, some work that that Phil BL and others have done have basically shown that one terap is not enough. So even if these companies massively accelerate their expansion plans, what's nice about that is it might buy Elon a bit more time to get Terap up and running, which is probably not a bad thing because the enormity of this task is is crazy. >> It is, >> right? So if you can buy yourself another year or two, then you've done well. And those companies building those advanced chip fabs today does mean they will have a head start if Tesla executes flawlessly everything flawlessly. It does buy them a couple years. Uh and during that time they can absolutely make hay. Uh so in terms of the real cost you said SK is looking to spend 400 billion dollars to flatten land and build the new uh chip fabrication facility. We know uh TSMC and Samsung have spent hundreds of billions or or will have spent hundreds of billions on theirs. Tesla and SpaceX believe they can do it leaner. How realistic is that? >> If it was anybody but Elon, I would say it's totally unrealistic. I think TSMC is projected to spend about $290 billion in Arizona on their FBS. SKHEX is four FBS for 400 billion. Some big numbers. >> Some big numbers, >> right? And so can Elon build terraab for 119 billion is the highest number I've seen so far. Probably not. It'll probably be more than that. And by the way, that terapab is never really finished. Uh all these fabs, you're constantly, you know, just like we've seen with Giga Texas, they're constantly making changes and improving it and adding equipment and so on. So the the capital expense continues >> forever. Forever. And that is the thing with a Tesla factory. I joked with Joe Tagmire in a video recently. I said, "I'm watching your drone footage and I see they're bringing in new equipment. What? I thought it was done." And he had a good laugh because that's not a thing. It's never done. Uh so in terms of the cost, I appreciate hearing that it is likely to have cost overruns because that budget was already surprisingly low. But as long as you can begin production, as long as you can start having products go out the door in terms of now, we talked about job postings are already up, but they're breaking ground in both Giga Texas on the on the compound there in Austin and on uh in Grimes County. Uh what's going on there? >> Well, in Giga Texas, I think we'll probably see first steel pretty soon. >> Uh >> by the time this airs, it may it may have already reached that point. It may have. I certainly been working on the foundations in Grimes County. Uh they've been clearing the land and Joe TechMy has done a great job. I took a drone out there and and got some great footage. They've done quite a bit of work in clearing the land. It's it's heavily forested uh plot of land. At least half of it appears to be heavily forested. Um so they've been working on that. That'll take some time to clear the land and then of course flatten it out and then start working on the foundation. That's going to be fascinating to watch just how big that that foundation is. Um, and Brian, I'm looking forward to you to count the columns again like you did with Giga Texas. >> I don't know, man. I don't know. Those are a whole lot of work. Uh, but the thing on that construction is, and I can't believe they're playing the greatest hits, but the skeptics looked at it and said, "That's not construction. They're pushing dirt around." And I said, "Is it 2019? Are you looking at the drone footage from Shanghai?" because that's exactly what they said. >> I don't know if any of you viewers out there know how construction works, but first you clear the land and push the dirt around and then you put the foundations in and build it. So that is to me at least a pretty logical first step. Push some dirt. I I don't know. >> Steel in a forest that that's gets to be quite difficult. It can be difficult to put steel just just dodge the trees, man. How hard could it be, >> right? >> Uh, ridiculous is what that would be. By the way, we're not done with this conversation, but I did want to take a quick uh intermission here to mention that uh you can, should you choose to subscribe both to me and to Cern Basher's channel. Uh this is uh a very silly uh bit of AI that was generated. I say, "Hey, you know what? uh, can you make it look more like us? And it's the same picture but with smaller eyes. And now I'm apparently have no hair instead of four or five remaining. And I pointed out, no, no, I I do have more hair than that. And it said, okay, you've got a full head of hair, Brian. You're now a werewolf. Uh, so that's us. You can subscribe. uh because CERN does uh great work again both on X where he posts frequent frequently and with tremendous depth of analysis and on his YouTube channel links in the description all that good stuff. So on this uh the logistics of chip manufacturing is challenging. TSMC found that they were unable to get certain elements uh certain compound get the acid of sufficient uh purity to actually make their chip fabs work. The US is going to have to reinvent parts of the supply chain that either never existed or have been deprecated to the point where they need to be reinvented. How much of a challenge does that pose? You suppose >> I think we have to assume it's going to be quite challenging. Um, and I'm going to be interested to see how much of this supply chain SpaceX and Tesla decide to vertically integrate just like they just like Tesla's done on the EV side. >> Um, you know, how much of this will be necessary for them to tackle on their own, >> right? >> It's one thing to say we're going to build, you know, chips and memory and packaging in one factory, which has never been done before, but to your point, what else is there that they're going to find that they also need to to solve, >> right? Will they need to buy a company that makes special seismic equipment mounts perhaps? I have no idea, but that's just the kind of little tiny details that are easy to overlook. Uh that can make or break it. >> The other thing I think that is is nice and instructive is they're building this advanced technology fab and the Austin campus and that's kind of their test facility to rapidly iterate on different ideas. And it sounds like Elon conceptually has some very different ideas about how semiconductors can be made. Whether it's on the >> whether it's on the EUV lithography side and different technology there or just this whole idea of being able to eat a cheeseburger and smoke a cigar in a chip factory, >> right? >> I don't know if I'm allowed to share it, but I have uh it has been explained to me how that would work. And as soon as it was explained to me, I said, "Yeah, that actually would do it." Uh because but I I'm not I don't know what I'm allowed to share on that. Uh so there are a number of things. And this is just like Tesla's uh philosophy on I asked a worker, why is the R&D spend so low? Shouldn't your R&D spend be five, 10 times higher? And he said, oh yeah, yeah, yeah. It's [snorts] because we're all R&D. >> Yeah. We're all doing the R&D every day. And if we're not providing some R&D, what are we doing? They will. So you are encouraged to participate in the refinement. No, no, no. Not encouraged. We are required. If somebody doesn't ever put in ideas, the boss will say, "Quick question. Do you think things are perfect here or could they be better? How could they be better? What could we do to make this safer, more efficient, more affordable? Eliminate a step." And sometimes you'll get really bad ideas and sometimes you'll get amazing ones. And this isn't to say that Elon is some super genius. Oh my gosh, he's a very smart man. He's a very clever engineer. But if you wish to hate him, you can still hate him and recognize that his engineers are really, really good. And you might have someone from Drivetrain who comes over to autonomy and says, "Could we do it this way?" And sometimes those crazy ideas that seem really dumb at first blush, you sit back and go, "Wait a minute." That might actually work. And that's one of the ways Tesla's been able to have breakthroughs. While at other companies, they tell the employee to stay in your lane. This ain't your department, buddy. And >> the the other interesting thing too, Brian, is I've heard a story about SpaceX uh in terms of building these data centers. Uh I think this was according to semi analysis. They said that at the Colossus uh 2 site, Macroh harder and mini hard, there's about 3,000 workers that go to that site every day. And they said at a similarly sized site somewhere else, I'm not sure where it was, maybe a couple of them, that the number of workers going into those sites is more like 10,000 a day. So somehow Elon is able to build these data centers faster with less people. >> That's uh unsurprising to me, I guess, is how I'd say it. when you saw the cable management uh beauty, the absolute feast for the eyes that was the cable management at Cortex 1, uh I knew that somebody there was really paying attention. And then when you found out that, oh yeah, we stood it up in a month at a at a cluster size previously considered unworkable, uh you knew they were on to something. So uh I think that's a good place to wind it up, guys. It's chips. There's a lot to know. So, what did we miss? What do we misunderstand? Leave it in them comments. You got to realize the videos where we try to make you a little smarter. The comments are where you try to make us smarter. Although on some videos I do ask for wrong answers. On those, the objective is to make us dumber, but with a laugh. Uh, not today. We want to be smarter. So guys, what did uh what do we do? What do we got to do? What do we got to fix this? Do we need to address more of your concerns? Let's find out. Everybody else, like, subscribe, do what you do. Stay tuned and juicy. And I cannot wait to hear from you clever robots in this our exciting future. Uh check out CERN's channel.