SpaceXAI Launches HUGELY Profitable AI Business

Hinzugefügt: 2026-05-16

[00:00:00]So, compute power is now the number one bottleneck for the AI industry, and we have just witnessed something absolutely historic. Anthropic's compute rental agreement with SpaceX and xAI gives Claude AI full access to Colossus 1 supercomputer in Memphis. Now, mind you, this is one of the world's largest AI clusters. xAI originally built Colossus 1 for Grok, but now it mostly uses Colossus 2 for its AI training. So, what sort of an edge does this give Anthropic, and what does it mean for competitors like OpenAI? Also, is this the clearest sign of SpaceX's impending dominance in the orbital AI data center domain? I have Brian Wong with me, and he's going to break it all down for you today. Brian is a futurist thought leader, and he runs the science the prominent science blog nextbigfuture.com.

[00:00:54]Welcome, Brian. Great to have you back.

[00:00:57]Great to be here. A lot of exciting news happening.

[00:01:00]Absolutely. So, [snorts] quickly, let me just bring this deck up, and if you could take us through this historic agreement. Um Mhm. Before we begin, when do you I think I think Elon said there's going to be a new logo for a combined SpaceX and xAI entity? Is it going to be SpaceX AI? Any thoughts on the new logo?

[00:01:22]Um no, I don't know what the new logo will be.

[00:01:26]Um but it you know, maybe focused more on the AI data center in space type thing. Um where it's basically getting both to be prominent because that will be the dominant revenue impact aspect.

[00:01:40]So, I think that In the long run, right?

[00:01:43]In the long run. But, you know, like as short as four three four years, it'll be a major aspect of what they're doing.

[00:01:49]Yeah, do you think it'll be bigger than than Starlink?

[00:01:53]Oh, the the AI? Oh, for sure. It'll it'll be way bigger before.

[00:01:57]Well, so so Starlink is 30,000 satellites V3, right? They're going to go from the 10,000, 11,000 now, to 30,000 of the high-speed internet ones, and then 15,000 of the um uh for for direct to cell phone, right? Versus 1 million for the AI. So, you're looking at 20 times as many satellites for for the other thing. So, yes, it will be bigger. And then that will scale continue to scale up.

[00:02:23]Um yes.

[00:02:24]Right.

[00:02:25]Okay, so back to your deck. Okay. Okay, so yeah, so um for people who don't know um Anthropic and uh is renting the closest one as as as you described, and that's for about 300 MW. Um Epic AI says there's about 445 MW there. So, um there could be still some usage 100 MW uh from XAI as needed, but you know, they'll have to share in some way uh when they get fully ramped up.

[00:02:57]Um so, this is immediate revenue for Anthropic.

[00:03:01]So, they have already started um promoting their um increased service.

[00:03:06]So, they're increased usage limits, where they're doubling the um Claude code 5-hour rate limits for each of their tiers of their subscription service.

[00:03:15]And they remove peak hour limits. So, basically like if you want to spend and give them more money, they will let you do it because they have more capacity.

[00:03:24]Um so, they could double their revenue on the API side by increasing these limits, which are already happening. So, I I I think they're already using the XAI uh thing. But then some things maybe doesn't start till end of the month, unclear. So, these are the token increases, where the basically the level one tier one becomes as good but even better than what the what the the tier two was. So, tier two used to be 450,000 tokens, and the tier one was the only 30,000, which was very low limit. And now tier one is gone to 500,000. So basically they bumped up all tiers. So it's like what was 1 2 3 4 is now 2 3 4 5 kind of thing.

[00:04:03]And then increasing the maximum maximum usage per limit. So again, people are paying for token, so then if I increase the usage, they can make more money right away.

[00:04:12]And this is money right away to XAI, right? So the 300 megawatts, so it because it's dedicated data center, then that give the premium. So Amazon charges 2.4 times the price of what it takes to build what they're doing, right? For a dedicated large scale data center.

[00:04:37]Right, so there's a premium for it.

[00:04:38]People think that there's a a discount for the volume. It kind of works in the reverse in the situation that they have described here. So then if it was the full data center, 7 billion dollars to build it, 2.4 times 7 billion would be 16 billion.

[00:04:54]But if it's kind of only part of it, like 60 70%, 300 megawatts, then maybe it's 2 times 5 billion because not all 445 megawatts are like 300 megawatts.

[00:05:06]But I can if the demand is off the hook, which it appears to be for Anthropic, then maybe it end up going to the full 445 megawatts and then they pay them more money. You know, like go up to 16. I'm anticipating that the demand being so large, if which is what Dario's talking about on interviews, one anticipating an 80 times increase in usage versus 10X, right? Then they can use everything that they can be given and then XAI can make more money. It will help with the IPO to say hey, XAI is not losing money, not losing a billion dollars a month. It's actually breaking even or better by renting out only 15% of our capacity.

[00:05:51]Right? If you do, you know, based on H100 equivalents, right? It's more if more like 1/3 based upon GPUs, like with the cost of one of H100s and H200s versus the four times better B200s, B300s, right? So, we There's still some um thing we don't know, which I would expect after the S1 filing to say, "Here Here's the date which we expect maybe next week.

[00:06:20]Here's the date when we're going to do this thing."

[00:06:21]Then there'll be a time when they say, "And here are the financials of XAI."

[00:06:26]And then this will say, "Our financials are great. You thought they weren't going to be good. Our XAI financials are great. Our SpaceX financials are great.

[00:06:33]And we have a partner for the future with space-based AI."

[00:06:36]So, the high end I think is 16 billion, the low end is eight, and the super low end bear case is five. But I'm thinking eight to 10, but then if you use up the full 445 megawatts, then um it could go up to 16 or even more.

[00:06:51]So, this is So So, what's the timeline for this? And And how much of this is a combination of um space-based uh and like terrestrial AI data centers? This is This is right now just talking about Colossus one. This is like not talking about anything else. This is only Colossus one rental revenue. Right? So, I'm not talking about the space based stuff. I'm not talking about distributed AI.

[00:07:17]Mhm. This is just I'm not even talking about if they were to rent half of Colossus two.

[00:07:23]Right? Like they could do that, and then they would make even more money, right?

[00:07:27]Yeah. Um but basically they're they're getting the um So, it's good for Anthropic. They're making more revenue.

[00:07:33]Instead of a run rate of 44 billion, they go to like 60 billion, maybe even this month or next month. Right? So, suddenly it's like they're know, growing fast. 30 billion run rate a few months ago, 44-billion just you know, like in the last month. And then with this deal they go to 60-billion plus. Wow.

[00:07:50]>> Right?

[00:07:51]And that like in May June, right? And then good for XAI and SpaceX IPOs because now instead of worrying about XAI losing money they only have They have a revenue stream. I have a revenue stream that's making me money. And I'm still doing the all the training and all development of the of the code. Yeah.

[00:08:10]Yeah. And then so then we get both IPOs out. XAI SpaceX first, Anthropic next.

[00:08:15]Bad for OpenAI. They're the third IPO for sure. Yeah. Yeah.

[00:08:19]>> Right?

[00:08:19]Yeah. So this is this is interesting because I I how much of this is a 5D chess move by Elon? Mhm.

[00:08:28]Well, it it's um it's a it's a win-win for for him.

[00:08:33]It's a it's a win-win-win for for him.

[00:08:35]So he he wins because XAI gets better more more revenue. On stuff that he wouldn't fully use Liven. Two, it's a win for Anthropic. And then they also have a deal. I'm sure they have a better deal now with Anthropic because they needed to have the Cursor data.

[00:08:49]And Anthropic could try to limit them.

[00:08:53]Because I think even before the the Cursor deal, they were saying, "Hey, XAI, you're using our Anthropic stuff.

[00:09:00]We're cutting you off because we don't want you learning from our models and training against our models, right?" But then Elon did the Cursor deal and Cursor is 1/3 of the revenue for Anthropic. So Anthropic can't cut off his right arm in order to cut you off, right? But then now with this deal it's like they're going from enemies to frenemies. Mhm.

[00:09:21]And then they and they have the data. So they have the data to to catch up. And then they maybe even are semi-partnering for all I would think definitely they'll semi-partner on all the future stuff.

[00:09:33]But then they both hit their enemy total enemy OpenAI.

[00:09:37]>> Where it really hurts. So, it really hurts them. So, it's like and that's like the cherry on top. It wasn't the reason you did it, but it is an extra benefit of like and I get to screw him.

[00:09:47]That's that's nice. So, this is net net very positive for both the SpaceX IPO as well as the Anthropic IPO that's expected maybe later this year or later next year. But before Anthropic, we were all expecting the OpenAI IPO.

[00:10:02]Right.

[00:10:03]>> And together with the with the case that Elon has in court currently against >> Right. OpenAI and Sam Altman. And these are two really big negatives for that IPO, isn't it?

[00:10:14]Right. Right. It's it's it's a huge Yeah, it's it's even if Elon doesn't win that that uh the court case, um I mean, which I think their chances are low.

[00:10:25]No, I think the chances of them winning some part of it is high.

[00:10:30]That they can get like something like, you know, you know, but if getting the full thing of like forcing a restructuring, that's low, right? But saying, "Hey, you wrote in your diary that you lied to me and, you know, you you you you you you said that these things would show that you, you know, the $40 million would would defraud in some way, right?

[00:10:51]So, getting some millions of dollars or $40 million or $100 million, that that I think that the odds of that are pretty good, right? Because they clearly did bad stuff. The other thing is that if you allow them to freely do this um um nonprofit to profit conversion, Mhm.

[00:11:14]right? If I say, "No rules, you can do whatever you want. You can change your everything you you did." Right?

[00:11:22]That opens the door for $5 trillion of hospitals, universities, all these other nonprofits to go. So, then there'll be this massive um um you know, conversion for for the people who troll the non-profits and they say I'm going to you know, make a bunch of for-profit stuff in a hospital and then pocket the money. Cuz they had restrictions, you know, in the '80s and '90s and various other things, right?

[00:11:47]And then they're saying "Oh, I can just freely do that? Great." Cuz that is something where you would definitely make instead of 3% 5% profit because it's a non-profit and then go to "Well, now I'm unlimited. I can have one part that's the the the for-profit, one part is a non-profit. I can make my hospital charge 20% margin or 15% margin and then I'll you know, give some up-cut to the non-profit side, right? So, basically all the hospitals in the United States would go right? If you do not limit it. But, that doesn't mean that this current decision will be that, right? Because um I it could be in appeal or in the Supreme Court where they would then uh make that limitation. But, there's no way they can allow the uh precedent-setting thing of what OpenAI did to stand. They cannot just let that happen.

[00:12:44]>> that opens a whole Pandora's box here.

[00:12:48]Right. But, anyway, so so the I expect Elon to get something. The question is how much? It's unclear.

[00:12:54]Mhm. Okay. So, that's that part. Um so, there was the shift of 0.3 gigawatts between So, we have this table of all of them. So, in the middle are OpenAI and xAI in the range of 2 2.5 gigawatts and then 0.3. So, xAI drops to 1.7 and then OpenAI goes to 2.8, right? So, and but then OpenAI's uh xAI is still building stuff. And so, this is uh a significant shift in the immediate um um building of of data centers. Uh but, we have a long-term game where everyone's building a lot as much as they can uh, going into next year and and beyond. And the big advantage that XAI has is that they can do inference computing at the superchargers, with the cars, and with the powerwalls. So, superchargers are about um, 7 gigawatts, and then you're doubling every 2-3 years, right? Uh, based on the pace that they're trying to increase to.

[00:13:54]And then, you already have the power there. So, if I place a small rack at that location, then, you know, if I have like 10,000 racks, I can tap into those 7 gigawatts and continue to build out. So, it's um, easier to add that. I don't have to build new grid, I don't have to build new stuff that they're already kind of like in place with the solar superchargers.

[00:14:18]But the cars, once they get the digital Optimus to work, there's like 1 gigawatt just from the the hardware for cars lying around. So, that'd be three times bigger than the 300 300 megawatt deal of Open AI.

[00:14:35]Yeah. Just from >> we we should we should do a separate deep dive on digital Optimus because I know you've been doing a lot of tremendous work. Yeah. And I would love to chat with you about that separately because that that is fascinating. Right.

[00:14:48]The other thing is powerwalls.

[00:14:50]I think there's about a million powerwalls out there, each with about 13 kilowatts. So, that's about 13 gigawatts of powerwalls. And a new company, a startup spanning AI, working with um, Nvidia, is putting a like a dozen chips, 16 chips chips, Nvidia chips, into each home with the equivalent of about a powerwall.

[00:15:12]So, that is another way to tap into more AI inference. So, the thing is, XAI has the Amazon capability to toss on a a large amount of compute for AWS. So, for Amazon AWS um for e-commerce Amazon built a lot of cloud, and then and then they used it for themselves, but then they added on AWS where they built like four or five times more and then they sold it to other people.

[00:15:41]This is the model that XAI will be showing, that they can quickly tap into more inference via their energy network of superchargers, cars, and powerwalls.

[00:15:52]And then they can then rent it out to themselves or to OpenAI.

[00:15:58]Oh, sorry. Or to or to Anthropic. Yeah, I had the two mixed up.

[00:16:02]>> Brian Brian, at this stage I want to ask you a question because then just so that our viewers understand. So, this there are two levels where or two stages where you need um AI compute, right? One is for training. Yeah.

[00:16:17]More powerful models, and the other is for inference, right? Now, inference can be distributed.

[00:16:24]Mhm.

[00:16:24]But training when it comes to training, you need your chips all together in the same place, right? Physically all together in the same place. So, if you zoom out and you look at how um XAI is building its infrastructure here terrestrially on Earth and also looking to build in space.

[00:16:48]How much of um an advantage does this give quantitatively and qualitatively to XAI and SpaceX's ability to keep building news new infrastructure um and and moving its training as well as inference to the latest hardware either terrestrially or in space while at the same time giving out uh you know, its previous generation of hardware to anybody like Anthropic who may want to hire it.

[00:17:23]Give us the big picture.

[00:17:24]So, I think that the, you know, 13 gigawatts of power wall increasing, you know, 20-30% every year. The superchargers 7 gigawatts also increasing at 20-30-40% every year.

[00:17:37]And then the cars also increasing, you know, at probably slower rate, probably 10-20% every year. Although with cyber cams, that could also have a surge in growth. So, you add those up, you're looking at like 20-30 gigawatts right there, and then that will keep growing, right? So, it'll take 3 years say that to fully tap that in, but if you have an average of adding 10 gigawatts per year, that would be huge. You would cuz the whole world is adding maybe 20 gigawatts per year. So, suddenly you can get to 50% of the additional um growth of of AI by being able to tap into these less conventional distributed sources where I can do it without, you know, with less grid building or with less whatever. Um and then for AI in space, if I get to 10,000 launches per year with Starship, which I can see how that can happen by 2030, right? And 10 10 launches this year if if we get the the next launches right.

[00:18:40]Um 100 launches the year after because we have um three major uh launch facilities, Texas, Boca Chica, um two in in Cadavra, one already built another and with already with FAA approvals up to 120.

[00:18:58]And then I think that in the in 2028, by shortening, you know, getting your operations down to Falcon 9 efficiency where it's only a half an hour to get my launch off where I shut down the the the air traffic um exclusion zone 20 minutes ahead of time, get it up lifted in 10 minutes, then instead of 3 hours, you know, then I can maybe get um you know, 500 launches, 200 per location. Which is about what I'm doing for for Falcon 9. That gets to 600 launches. And then with the money I've raised this year, I would have it take 6 months, 9 months to build new launch towers, you know, probably another 6 months to get the flight locations and all a bunch of other things. But by 2028, I should have new um launch towers, new sites online, and I can ramp up the number of launches to where a few thousand, right? By 5 like 6 to 10 locations in 2028, 20 locations in 2029.

[00:19:58]Mhm. That's I think what what you need in order to really scale to the 10,000. At 10,000 launches per year with 10 MW, so like 100 um 100 kW satellites. So 10 MW in each one. Mhm. Then that 10,000 launches is 100 GW. Yeah.

[00:20:19]>> Right? So that's the kind of scale that has to happen to do that. But once I'm doing 100 GW, then I'm five times more than the regular 20 GW that everyone else is adding on Earth. And even with 10 10 extra GW, 20 extra GW of distributed power, um you know, from superchargers and other stuff.

[00:20:35]>> So the AI the ultimate AI data center data centers, would they be for training or inference or both? Inference.

[00:20:42]Inference. Purely inference. Purely inference. Purely inference. Okay.

[00:20:46]>> Right. Right. But thing is inference is the you know, 100 times thousand times larger thing because I'm profitably you know, you know, running the models to make money. So it's like I'm I'm building it less, I'm selling it more. So the ratio of selling more is increasing increasing. Which is what, you know, Jensen has said, other people have said it's going to happen. So then for training Grok models, more more powerful Grok models, um do you see like subsequent Colossus data centers being built out all across terrestrial data centers? Right, right.

[00:21:24]You you would you're going towards from the 1 gigawatt to 2 gigawatt on the um drawing board, on the plan is 5 gigawatt and 10 gigawatt, right? So that was, you know, like you know, Meta, Prometheus, Hyperion, they're 5 gigawatt, 10 gigawatt facilities.

[00:21:43]Um you know, Colossus 3 hasn't been announced, but I'm sure with the Reuben chips, they will have they will target a 5 gigawatt facility and probably 10 gigawatt facility. So you by 2030 you will have a 10 gigawatt facility, which is what um uh Leopold Aschenbrenner, you know, who runs his 5 billion, 10 billion dollar fund, says is needed for AGI. So we will need to get to 10 Yeah, go ahead.

[00:22:09]And um so I'm just by extension, when you look at Terafab, we know that Terafab is going to be building chips for different purposes, like for cars, for Optimus, for um the space um AI data centers.

[00:22:22]Um will Terafab, when will Terafab, and will it also build chips for terrestrial AI data centers for training models?

[00:22:33]Yes, it will be building chips for everything. It will be making AI 5 chips, AI 6 chips, which will be you know, on Earth, in space, wherever needed. Dojo 3 chips for training. Yes, so they will be making chips for all purposes. And I have a bit um on the next slide, I think, that will discuss that. Where I discuss So in the Google thing, in the the the column to the right, it talks about the TPUs that they have, and the Tranium chips. So each of the big players, the hyperscalers, are making their own chips, right? As much as possible, because they don't want to pay the 50 to 80% margin to Nvidia where they can avoid it, right? But Nvidia is so good that they need to still have them, right?

[00:23:16]>> Right. So, then um going on to the uh next slide. So, as I mentioned, part of the these deals are the Cursor training data. That is key to xAI with Cursor catching up to Anthropic, right? And Anthropic will block it less because, "Hey, we're we're buddies now. You know, I'm helping you make a ton of money. Get off my back about um um the fact I'm using the data, you know, which I should have right to anyway, um to do this thing." So, they will basically make a truce on that training data thing, where basically um Cursor and xAI will use that data. And there're going to be future partnerships with space AI. And I think for distributed AI, because in the AWS model, I need people to buy it up. And if you have an insatiable demand, then I have I will have insatiable supply that I cannot leverage as fast as I can bring it on. Basically, they can bring on distributed AI, I can bring on space AI as fast as possible.

[00:24:16]When I need to throttle Anthropic down because my Grok models have finally caught up, or I made a new Cursor code model, I say, "Hey, I'm going to make four times as much money with my model versus only twice as much um as my cost for for the other thing." So, it's just in more margin, more money, and then having Can I ask you a question, just going back to one of your earlier slides about the revenue? We're looking at as much as 16 billion a year, right?

[00:24:48]Mhm. Right. Um how much does it offset the cash burn that uh xAI It it could it could it could offset completely cuz though it's about 1 billion 1 billion dollars a month is what they were doing, right? Of losses.

[00:25:03]Although in the revealed uh financials for 2025, they were saying SpaceX makes 5 billion dollars and then we go to -5 billion dollars when we add in um XAI. So, that'll be 10 billion dollars. It'll be not quite 1 billion dollars per month.

[00:25:19]It'd be like 800 million or something like that, right? So, if I did grow some other things, ex-payments, ads, blah blah blah, then you know, the at 5 billion dollars, it would be close to 400 to 500 million dollars per month coming from this. If it's you know, 8 9 billion, it fully offsets. If it's to the higher numbers, 10 12 16, then they're making a profit from XAI. And it's only using 15% or maybe 10% of their actual total compute built, right?

[00:25:51]So, that that means that they can you know, if they were consistently doing 50% for other people, 50% inference for us, and then we have still have all of our own training, right?

[00:26:04]Then they can comfortably be very profitable, right?

[00:26:09]With the 50% of the of the data center stuff. So, that means they can um raise money, build as fast as they can, build you know, big data centers, build um the distributed stuff at the cars, superchargers, and homes. So, the reason I um I'm kind of trying to zoom out and look at um SpaceX post-IPO, right? One of the biggest concerns about um post-IPO SpaceX was the cash burn for XAI, right? So, here is a potential solution to that problem, right?

[00:26:46]>> Right. Mhm. Um immediately. And then of course this becomes this takes on a life of its own, and the more Space XY scales with data centers, the more revenue comes in.

[00:26:59]And it's a flywheel that keeps going.

[00:27:01]Now, what happens to the likes of AWS, Google, um uh Oracle, even, you know, and and all these major infrastructure companies supplying infrastructure for AI compute?

[00:27:18]Because the problems that exist in America in building data centers terrestrially on American soil continue to exist.

[00:27:28]There's growing opposition, and communities increasingly don't want a data center near them or in their communities. So, I'm wondering what how you think of this dynamic playing out.

[00:27:38]And what's the lead that does give does that give Space XY? So, Dylan Patel said, you know, in a in a interview he did recently, and he tracks things closely with semi-analysis, where they look at all the token economics that probably think. So, the Anthropic has shown that they can be profitable with um with their AI inference, right? And then they've cut deals with Google, with Amazon, all for it. So, they he says, "Any AI data center that's built, first tier, second tier, everyone sells out.

[00:28:13]Everyone Whoever you want to sell, you will sell it, you will make your money, you'll be profitable. And then the one using it, Anthropic, whatever, they will be profitable. So, we've crossed over from we're building without profits to a building with profits. So, then, whatever you can build, you turn it on, you overcome the blockages to making it, that new data center will make money. All your old data centers make money. So, that's why, you know, think the re-rating where, you know, all the AI players are making more money is because this is now we're past the point of will it make money to yes it makes money and the question of who makes money is everyone who can build anything.

[00:28:55]Right. So what one more question before we move on and this is about the life cycle of these chips, right? They're anything between three to six years.

[00:29:02]Right.

[00:29:03]So how old is Colossus one now? It's about almost three years?

[00:29:07]No, I think it's like two years. Two.

[00:29:09]Okay. So what happens once that life cycle is over? What happens to Colossus one do you think? Is it is it going to be upgraded or is it going to be hired or the services of Colossus one going to be taken over by or hired by companies that are Yes.

[00:29:28]level two Right. Right. So um the the H100s the which are H100 H200 that are in Colossus one, right? The rental rates increased. You know, so they went you know, started at a certain level they went they're going back up because demand is higher. It's because if that H100 H200 can still serve out tokens the fact that it can serve out you know, a billion tokens mean I can make my million dollars. Right? So that means that it will still have use and utility if I'm constantly having a shortage of energy compute then I'm still paying for the old stuff just less relative to the new stuff but still at a price >> earning revenue. Yeah. It's still earning revenue.

[00:30:16]Right. So then it won't drop per se you know, like if the tokens [snorts] start dropping you know, five 10 times or whatever like that then you know, goes to 10 times then the amount of money they can make from these tokens may start being less and then at some point it becomes I have my shell the building. I have my energy and so then I would look at swapping out chips where I could finally take them out and then pop in the new ones, right? So, um but that probably isn't for for 3 years. But it's it's assuming I have an abundance of chips to do it, right? But I have a lot of chips.

[00:30:57]I can't stick them anywhere else. Well, let me just revamp this one. Which you see that with um the uh Bitcoin miners.

[00:31:07]Like like Coreweave and some of the other companies that were making, you know, $2 billion a year on Bitcoin mining, they say you know, they say, "Okay, you Bitcoin miners, get out of my my data center. I'm putting in AI and I'm going to make 10 times more money."

[00:31:23]Which is why, you know, some of these names are converting over, right? So, that is a conversion factor. So, it will be you can renew, but you choose the right time when it's like, "Okay, lease is over, get out. I'm putting in the new one, blah blah blah."

[00:31:38]You know, I love talking to you. This is why Exactly.

[00:31:42]You know, [clears throat] it's like we're building a mind map Yeah. as we go along. And I think it's so important and that's why I my viewers absolutely love you, Brian, because uh you know, it's these you don't get these mind maps as clearly with anyone else as as I get with you.

[00:31:59]So, thank you and apologies for keep keeping peppering for I keep peppering you with these questions, but they are just Anyway, let's get back to your take.

[00:32:08]Okay. So, they said that they want to talk about being future and it's kind of like if they if xAI, SpaceX, Tesla build all this extra inference on the ground in the distributed locations and in space, then everyone comes to them. All of the AI players come to them because money matters less than actual working chips and energy.

[00:32:31]So, that is the short resource that they have and if it's short for the next 5 years as far as I can see, then SpaceX XAI have the dominant position.

[00:32:41]If they start making five times, 10 times, 100 times more compute in space than anyone else, then they're the ones with the um the monopoly on energy and power and and compute.

[00:32:55]Um so part of the disputed AI, I think that Yeah, I discussed that. I think that if they can get the superchargers, the cars, and the powerwalls loaded up with AI, that means 10 to gigawatts more more AI, which is uh if you have 30 gigawatts of that, that is 100 times more than the the deal we just did.

[00:33:18]Right? So that's like >> so you're saying that it's inevitable that that gets leased out as well in time.

[00:33:26]It'll be It'll be leased out as well or if Grok can use it himself, become dominant, you know, either or. So either making twice as much money or I'm making, you know, four times, five times as much if you know, 10 times as much because I'm actually got the model.

[00:33:43]Right? But the the the economics are not bad for leasing.

[00:33:48]Right? So I will they will be making plenty of money from that. Right? So it's it's like saying you only have an AWS model.

[00:33:57]Amazon doesn't have their own thing, but they're renting out AWS. Everyone loves Amazon, right? And you're saying, I only have an AWS model.

[00:34:04]It's like, so?

[00:34:07]And I have a chance to to make my own model. It's like, yeah.

[00:34:12]>> So that's Okay. So then here's uh a slide from um Goldman Sachs Investment Research where we're at the point where um the agents are taking over, right? In the the middle of 2026, you can see that consumer agents and enterprise agents just start to take over. They're expecting it to go up 24 times, over 24 times in 2030. Which is more than doubling every year. And this could be this could be conservative. Right? But then you see that the demand is going 24 times, so that means if I can make more tokens, if I have the energy compute to do this, right? Then that's where the money is.

[00:34:50]Okay. Can you Brian, can you for for for my audience that maybe wondering what enterprise agents are, can you describe them for for us, please?

[00:34:59]That is the um Claude co-work, that's the Claude code, where they're using it to make enterprise software, or they're using it to do financial analysis that a company would do. It's not just a chat thing of like, here, let me ask you a question, I get a kind of social answer back, I I get a cup 10 pages it's like I have a cycle of a work and I'm checking it you know, every time, and I'm going towards a big company related answer or a product of like a hundred slide spreadsheet, a 200 page deck, a research report that's like McKinsey grade or something like that.

[00:35:40]That's enterprise caliber work, right?

[00:35:43]It's not just >> Right.

[00:35:45]asking, you know, the the typical questions we have at the chat level.

[00:35:49]It's like I'm working towards something that's replacing consulting uh that I'm paying $100 an hour for or something.

[00:35:56]>> Yeah. Right. Because this this also then it has an impact on the the workforce of each of these companies that deploys these Yeah. enterprise agents, right? Because uh there's been so much of discussion about whether it's going to lead to um the workforce being you know, reduced, or whether it's going to augment the existing workflow and put it on steroids because you get so much more productivity. Right.

[00:36:25]Right. Yeah, so um I think that, you know, in the All-In podcast, they discussed that, um, people are, you know, aren't worried about vibe coding. It's like, um, the back in the day when, you know, like, 30 years ago when VisiCalc or Lotus 1-2-3 came out, "Oh, we're going to replace all the accountants. We're going to replace all the finance people because everyone can do their own spreadsheets, right?" It's like, no matter how easy you make it, the professional doing it will give you better product work, right? Um, although, you know, you'll have something becomes trivial, right? Then the the value of the trivial thing becomes nothing because you still pay for the elite stuff, right? So, the I think the the the thinking that will replace the workers will is going away with the really good people will get productivity increased 10 times, 100 times, and then the quality of the work will go up. Yeah, generally across the board, the bar will be higher. Quality of work will go up. Yeah. Yeah.

[00:37:31]And then, yeah, consumer edge. I think the enterprise stuff may end up being bigger in terms of evaluation part of it. Like, in terms of consumer tokens, this is a rough estimate of that. But, the main thing is we're going to be using a lot more.

[00:37:43]We're going to use 25 ti- times more tokens within 4 years, and this could be on the low end, but just means more inference compute is needed. More data centers are needed.

[00:37:55]And it's all profitable.

[00:37:56]And then, the ROI on the hyperscalers is Google has shown that they're making, you know, $20 billion per quarter on their cloud revenue, and they're increasing by 63%.

[00:38:08]And they're having, uh, operating income of 6.6 billion on capex of 35. So, they're getting, you know, like, over 50% ROI from that, you know, based on this approximation, based on cloud revenue.

[00:38:21]Azure is also doing pretty well, strong revenue growth, reasonable ROI. AWS is showing a positive ROI trend. So, again, the the investments are giving money back for all the big players, and then, you know, Anthropic as well.

[00:38:38]Um this is a very uh small thing. It's just breaking down the ROI as well as the chips. So, each of the players is getting the the TPUs, Traniums, and then on the bottom line, you see that Tesla, SpaceX, XAI, they're going to make their AI 5, A6 Dojo chips, which is instead of paying $50,000 for a Nvidia chip, they'll pay like $6,000 for their own uh chips for inference or chips for training, which is similar to what uh Google's doing with TPUs, which are cheaper than, you know, cuz they're making them themselves, they don't pay the Nvidia thing. So, as long as you have a competitive chip, you will use your own to improve the economics of what you're doing. Mhm. So, I've I've I've put this um on the big three mode so that any of you watching can pause, and you can zoom in because it's a lot of dense text. Um so, use this opportunity to pause, zoom in, and study this. Yeah.

[00:39:36]And then the um money raised is like $42 billion for XAI, and he also got $10 billion in debt. So, a lot of money being spent, but if it's all profitable, then it's good. And then, if I'm reducing the the cost of my chips, I'm, you know, instead of spending $50 million to make a gigawatt, I may only spend $10 billion to do a gigawatt. So then, if I was already making money on $50 million for a gigawatt, I'll be making way more money on $10 billion for a gigawatt. So, the economics are improving faster and faster for this thing.

[00:40:13]So, again, if um Um, I can stand up another gigawatt of compute for only 10 billion dollars, or maybe even only 5 billion dollars because I already have the energy at um uh superchargers and cars, I only need to put in, you know, some other layer, then the like the car gigawatt in the car is nearly free because it's just chips sitting there that I'm just looping in.

[00:40:39]Yeah.

[00:40:40]>> So, so that means that the that Tesla, XAI, SpaceX have the lowest cost of adding in gigawatt of compute.

[00:40:48]Especially when they get the AI 5 chips into the cars and the bots. It'll be the cheapest way cuz they've they've already been paid for. The power is basically paid for, the the chips been paid for.

[00:40:58]It's just um, you know, the incremental cost is practically nothing, and then they can make you know, like um you know, uh 10, 20 billion dollars um per per year from that from that uh compute.

[00:41:15]So, the economic >> because like if you It's like one of those machines that's uh What do you call those machines that kind of um keep generating energy on their own? It's just like Oh, yeah, but but but a flywheel that doesn't need external >> per- perpetual motion machine.

[00:41:34]>> Perpetual motion machines, yeah. This is like a perpetual motion Yeah.

[00:41:37]>> uh mini economy on its own. It's It's crazy.

[00:41:43]Anyway, so so um that goes to, you know, how this will be able to scale to to many tens of gigawatts. It's a matter of like how fast can they bring it online, and then they know they have a customer right away who will take it and they'll make more money.

[00:41:57]So, that goes to how 2026 is already good for Open AI for Anthropic and and XAI and SpaceX and Tesla, and then 2027, 2028 will be even better. Once they're tapping into the cars, the bots, the superchargers, all that extra compute and energy.

[00:42:15]It's um >> [snorts] >> it's an insane amount of money.

[00:42:19]Yeah.

[00:42:19]>> And then so this is the macro hard with the name there. So that's basically it. So I think I've done my deck. Yeah.

[00:42:27]All right. Well, this is like you know, we should build a collection of these lessons, these deep dives with you, Brian, and then >> Yeah.

[00:42:37]collate them and put them together and then you you know, weave them into a book of yours and publish it cuz it's brilliant.

[00:42:44]Brilliant.

[00:42:45]>> Okay.

[00:42:46]Great.

[00:42:47]>> So before we say class dismissed, I have just one question for you and I'd like to know your thoughts on this big um shift back towards CPUs from GPUs. Mhm. And I want your thoughts on this tie up uh between Intel um and XAI for TeraFab.

[00:43:08]Do you see um Space XAI now moving towards a balance of combination of CPUs and GPUs?

[00:43:18]So the um Blackwell chip um like when you have the rack that um you know, of 72 chips or whatever that um that uh Nvidia sells you.

[00:43:30]They have like you know, I think 72 um GPUs and 36 CPUs. So So two to one. If you go back to H100, I think it might have been like eight to one.

[00:43:41]Right?

[00:43:41]>> Mhm. So now Reuben, which they announced, which they start selling um any month now, they um use one GPU and four CPUs.

[00:43:52]Mhm. So the ratio CPUs to GPUs is increasing to CPUs because of the workloads uh agentic workloads that when I do agents, more complex stuff, the routing around is more difficult and I need more CPUs to do it.

[00:44:10]So thus there is this and also I need a ton ton more memory. So basically based on the versions and what I have to do Yeah. I need >> we should have a we should have a a a separate deep dive explana- explain about this brand because there's so much of um kind of confusion over why because we switched from CPUs to GPUs because GPUs allow parallel computing.

[00:44:38]Mhm. Right.

[00:44:39]>> Right. That was the raison d'être raison d'être of the whole reason why we moved to AI and parallel computing. But then if you have sequential computing again in CPUs it's a bit of a I'd love to wrap my head around it and if you can help you know explain this because this is going to this is a fascinating I think this whole shift back towards CPUs is really really fascinating it and we should really do a deep dive in this as well. So yeah.

[00:45:06]>> yes it's basically to to to quickly summarize which we'll go to a deep dive in the complexity of what you're doing with agents I need to do that with CPU that that type of work I need more CPUs to do and I also need more memory. So that's the I still need the GPU that are still core but other stuff around it you know the best way I can do >> So is this a training versus inference thing? Like training on on GPUs but more of inference on CPUs with AI?

[00:45:35]>> it it's it's the inference is getting more complicated.

[00:45:39]Okay.

[00:45:40]So I So it's still inference in both cases but the inference got more complicated. So then I need to to have more CPUs to have the complications. It wasn't you know the the a simple routing thing it became um you know a lot more stuff around the parallel part. Yeah.

[00:46:00]All right. I'm going to look forward to this because I I need to understand this and I need your help. Okay, we'll do.

[00:46:05]All right, we'll leave it at that for now. Thank you so much Brian, it's been so wonderful going down this rabbit hole with you on you know, this whole anthropic space X space X A I deal and you you helped us lay the foundation of our understanding for this. So thank you so much. You're welcome.