Terafab: Can Musk Build the Biggest Chip Factory Ever?

Kind: captions Language: en But I don't I don't know how to build a fab yet. I'll figure it out. We either build the Terafab or we don't have the chips. And uh we need the chips, so we're going to build Terafab. >> This is Terafab. Tesla calls it the largest chip manufacturing facility ever. For now, it's a plan. So, can they build it? First, how big is it? >> Uh to give you a sense of scale here, uh we expect that the Terafab is going to be around 100 million square feet, uh which is 10 times the size of the uh the Tesla Gigafactory Texas. >> Why does it need to be 10 times Giga Texas? At the launch in Austin in March, he showed this chart. >> To give you a sense of um what we're talking about, uh the current output of AI compute uh is roughly 20 gigawatts per year. This chart explains why we need to build the Terafab, because all of the rest of the output from Earth is about 2% of what we need. So, if you add up all the fabs on Earth combined, they're only about 2% of what we need for the for the terawatt project or Terafab project. >> A podcast host had the obvious question. >> Why don't you do the Jensen thing and just prepay TSMC to build more fabs for you? >> Uh I I've already told him They're building fabs as fast as No, uh They're building They're building fabs as fast as they can. Um and so is Samsung. They're like that like that They're They're pedal to the metal. I mean, they're going, you know, balls to the wall. You know, they're as fast as they can. So, still not fast enough. >> So, they'll build their own. The The is scale. Chips for cars, robots, and AI satellites. In June, Musk did the math for a terawatt a year. >> But fundamentally, it's about scale. So, even if there were no uh fundamental technology breakthroughs, uh it's and and uh you simply you you could simply scale uh the existing chip-making technology uh with a lot of difficulty uh to a terawatt of chip output per year. Um that's if you look at it just from the logic die standpoint, that's like having a billion chips per year with a a kilowatt per radical. So, it's a a billion full radical equivalent chips uh each doing a kilowatt. And then you're going to need a lot of memory to go with that. >> Our own rough math? That many chips would need about all of TSMC's capacity in 2025 just for logic. Three things have to work. First, the machines. >> You know, the the the fabs today all basically use um machines from like five companies, you know, so you have got ASML, Tokyo Electron, KLA-Tencor, you know, um etc. So, um So, so I I I I first I think you'd have to get equipment from them and then uh modify it or work with them to increase volume. Um but I think you'd have to build perhaps in a different way. Um so, that I think that the logical thing to do is to uh to use conventional equipment in an unconventional way to get to scale. Uh and then uh and then what and then start modifying the equipment uh to increase the throughput rate. >> Second, memory. Right beside the chip holding the data it works on. >> And I'd say my biggest concern actually is is memory. Um so, the I think there's there's a the the the path to creating logic chips is more obvious than the path to um having sufficient memory to support logic chips. That's why you see DDR prices going ballistic in these memes about like you know you're marooned on a desert island you write help me on the sand nobody comes you write DDRM with best chips come swarming in. >> Later in the same conversation. >> That's the plan with Terra fabs? Millions of wafers a month of advanced process nodes? >> it's currently a number north of a million I think. >> You got to do the memory too. >> I think the Terra fab's got to do memory it's got to do logic memory and packaging. >> Third. Making good chips at huge volume. Musk on how long that takes the big chip makers. >> And then and then if I ask TSMC or Samsung okay what what's the time frame to get to volume production the point is that is it's not you've got to you've got to build the fab and you've got to You've got to you've got to start production then you've got to climb the yield curve and reach volume production at high yield. That that from start to finish is a five-year period. >> Nvidia's Jensen Huang whose chips TSMC makes on what TSMC gets right. >> And somehow they're running a factory with high throughput high yields really great costs excellent customer service they they take their work seriously they take their promises seriously then when your wafer because they know that you're helping they're helping you run your company when the wafers when the wafers were promised to show up the wafers show up you know so that you could run your company appropriately and so their system their manufacturing system is completely miraculous. >> Musk's approach from February. >> Well um we make a little fab uh and see what happens. Uh make our mistakes at a small scale and then make a big one. >> Is a little fab done or is or is it >> No it's not done which I mean who would I'm not going to keep that cat alive. That cat's going to come out of the back room. They'll be like drones hovering over the bloody thing, you know. You know, you'll be able to see its construction progress on X, right? You know, in real time. Um So, you know, we we we we I mean, like I don't know. We could just flounder and fail, you know, to be fair. It's like not success is not guaranteed. >> That little fab is going up now. A research fab at Tesla's Giga Texas in Austin. The plan has a trick. In March, Musk explained. >> So, in the advanced technology fab uh we will we will have all of the equipment necessary to make a chip of any kind, logical memory, and we will also have all of the uh equipment necessary to make the uh lithography masks. So, in a single building, we can create a lithography mask, uh make the chip, uh test the chip, make another mask, and and and have a an incredibly fast recursive loop for improving the chip design. >> On Tesla's earnings call in October, Musk gave an update. >> So, we have all the equipment on order for that. And we probably will be able to I think we'll probably be able to make something useful by the end of next year. Not not at scale, but as Gwynne said, to crawl walk run. We we got to try to you know, at least uh figure out how these machines work, you know? Like we don't know how they work. They We never built a chip line before. >> Pause. I saw that you >> The giant Terafab is planned in Grimes County, Texas, with the first phase of about $16.8 billion. Uh Intel has joined and TSMC is in talks. So, can Musk build the biggest chip factory ever? So far, it's a plan. But the first fab is going up in Austin, and if the loop works, a new chip goes from mask to test under one roof again and again. And all those chips need one more thing. >> Limiting factor once you can get to space is chips, but the limiting factor before you get to space will be power. >> That's our next story.