So we're going to talk about China, US Middle East, and then US infrastructure, and then a bit of a Q&A if we have time. I don't know. I yap a lot, so it's going to be hard. Oh, I got 20. I got 20. That's plenty. Time's ticking. Time's ticking.

Should I just stall so I don't have Q&A? Crowd's tough. So I wanted to talk about Huawei's chips, right? Because we never get to talk about them. Huawei is really, really cracked. They absolutely destroyed everyone in 5G and telecom by just engineering better. And now they've done something that's super interesting.

So they have the Ascend 910B and C, which you may have heard of, which are chips that they make. And then they've turned it into this really cool system architecture, right? You know, NVIDIA talks a lot about Blackwell and VL72. It's one rack with 72 GPUs connected together in NVLink.

Huawei has done something similar. It's called the Cloud Matrix 384. 384 of their chips, right, connected together in not one rack, but like 12. And there's a ton of optics and power connecting them all. What's interesting is that their architecture is actually one that NVIDIA tried to deploy and they failed at.

And that was called DGX-H100 Ranger, right? Which was 256 NVIDIA GPUs connected together in one NVLink network with optics. They tried and it did not work. They could not bring it to production because it was expensive, power hungry, and unreliable. And that sort of was the impetus for why Blackwell, they instead stopped going, you know, hey, 10, 20 kilowatts per server to 120 kilowatts in one server.

That is effectively one whole rack because they wanted to keep it all in copper, right? Whereas DGX-H100 Ranger and also the Huawei system uses optics to connect all the GPUs at super high bandwidth in the high bandwidth, you know, NVLink or I can't remember what the Ascend Cloud Matrix pod is called, their scale-up network.

But it's quite interesting that Huawei was able to engineer something that NVIDIA effectively failed at, right? Now, obviously, this is way more, you know, power hungry, of course, potentially unreliable. There's no data. And it would be expensive except for the fact that, like, you know, China's really good at making things cheap, right?

So that's something interesting about the Huawei system. What's interesting about the geopolitics of this is that despite the fact that Huawei is a sanctioned entity, China is a sanctioned country, they're still able to actually access TSMC to manufacture these chips, right? The chip is manufactured at TSMC through Sofco, which is a Bitcoin cryptocurrency mining company that pretended to not be attached to Huawei and just bought these chips.

And then there's, you know, HBM, which is high bandwidth memory from Samsung and Hynix in Korea. And then all the equipment to package it together is from the U.S., Netherlands, and Japan. So it's funny that sanctions are completely useless because they're still able to access this stuff. And what's interesting is, you know, what I'm showing on the right is that they've made, you know, they have roughly 3 million chips worth.

They have 2.9 million chips worth from TSMC. That's allegedly stopped now, right? The U.S. gave TSMC like a billion dollar fine for $500 million of revenue that they made, which seems like a slap on the wrist, honestly. But, you know, SMIC is also going to, which is China's TSMC, is going to start making it as well.

The other thing I would say is what's funny is that HBM was also banned to China entirely. But instead of just like, hey, HBM's banned, there's cool ways to circumvent this, right? Which is Samsung sells to this company called Coasia in Taiwan, and then Coasia sells to this company called Faraday, which packages it into a chip.

Like, there's a fake chip, basically, right? A chip that does literally nothing, and it has HBM packaged on it, and they ship it to China, and then they take the HBM off of that chip, and then put it on the Ascend. And this actually circumvents all the rules and regulations, so this is completely legal, which I think is, like, very funny.

And so Huawei has stackpiled roughly 13 million HBM stacks, you know, already. And they're continuing to receive more shipments, which is fun. Very, very interesting that, you know, that's kind of possible. The other thing is, you know, domestically, they have not been able to manufacture in the past, but now they're able to, right?

China's SMIC, right there, TSMC, has enough tools for 50,000 wafers a month. Today, the only 7-nanometer chip that anyone's found from SMIC is their smartphone chip, right? Smartphone chips are smaller, easier to make. They yield better than large AI chips. So when you look at, like, hey, 5-nanometer, the first 5-nanometer chip was an iPhone chip in 2020, but NVIDIA didn't release a 5-nanometer GPU until 2022, 2023 with the H100, right?

And so likewise, right, SMIC is making 7-nanometer for phones. They're going to get to the point where they can start making 7-nanometer for AI chips, likely in very high volumes this year. And based on the yields, they can actually get, like, you know, millions and millions, right? So it's, you know, the thought that China will not have equivalent compute is, like, kind of wrong.

They will have a lot of compute, right? Which will be interesting, right? Because, you know, there's already, like, big announcements from DeepSeq that they're going to work and use Huawei chips to try and train their next generation models. And so that'll be really interesting. The other thing that's interesting is, you know, recently, NVIDIA got banned from selling their H20.

This is, like, a cut-down H100 or H200 to China. They wrote down $5 billion worth of inventory. And on the earnings call, Colette, the CFO, said, if we didn't have export restrictions, we would have sold $50 billion worth of GPUs to China this year. Which I thought was, like, a very interesting comment.

But the ban stopped about a million GPUs. And if you, yeah, it's quite interesting that, like, you know, that amount of compute was also blocked. I think the other sort of geopolitical thing that everyone's talking about, besides China, is the Middle East, right? So recently there was a deal in the Middle East.

You know, Trump went to the Middle East. He didn't go to Israel. He only went to Saudi and UAE, which I thought was quite interesting. You know, not to get political, but, you know, that's interesting. No, no more, no more, no more. So the details of the deal are cool, right?

So on the right is a satellite photo of a data center complex in the UAE that G42 is making. And the deal is basically that G42 can buy 500,000 GPUs a year, and they get to keep 20% of them to do whatever they want. The other 80% have to go to U.S.

hyperscalers and cloud companies and AI companies, right? And so G42 is building a 5-gigawatt data center campus. The one pictured is the first parts of a gigawatt data center campus, right? That is absolutely, ridiculously big, right? XAI's data center is like 200 megawatts, right? You know, OpenAI and all these other guys are 200 or less megawatts for the models they've released.

So these are absolutely ginormous, right? Stargate, the first six parts of it in total is 1.2 gigawatts, right? So, like, this is a massive data center that they're building. G42 has a large customer and investor, Microsoft, and so Microsoft is going to be a big one. But if everyone remembers back, like, in 2023, Sam was always like, hey, $7 trillion, right?

You know, he's throwing this crazy-ass number out, or everyone kept reporting about it. The interesting thing is that part of this was that OpenAI wanted to build GPU clusters in the Middle East, right? And so not, like, publicly stated, but it's like OpenAI is going to have a cluster in the Middle East, right?

And so that's a big part of this deal as well. And in concession, right, the U.S. gets a couple benefits, right? UAE is providing matching investments to U.S., right? For any dollars they spend in UAE. The UAE spends in UAE on AI infrastructure, they're also going to spend in the U.S.

And so that's already started, right? G42 has sites in, like, Kentucky and New York that they're spending in. Nothing that's five gigawatts, but I'm sure they'll get something. We'll see if this, like, all follows through. The other thing is that most of the compute goes to U.S. companies, right?

80% of it. Is this intended for inference or training? Like, what's the goal of this data center? I think, you know, you could spin it any way you want, right? But the lines are also blurring, right? People today now, in their inference clusters at night in the U.S., are just running reinforcement learning with verifiable rewards, generating, you know, trajectories and then, you know, keeping the good tokens when there's low utilization, right?

So, like, inference clusters are now also training clusters, right? So I think that there's, like, the lines are blurring between what is an inference and training cluster. But also, this is a five gigawatt data center. Like, that's big. You could do training, for sure. Can we keep questions? Yeah, sure, sure, sure.

Yeah, we can actually do questions in this round, but come up to them afterwards. Sorry. No worries, no worries. The other one is, I think if we want to, we can yell questions, you know. The other one is the Kingdom of Saudi Arabia. DataVault is a company there. If you've heard of the line in Saudi Arabia, it's an absolutely ridiculous project.

It's actually really cool. There's building a city that's a straight-up line, and on both sides, they have, like, huge sets of, like, mirrors, so they don't have a ton of sun, like, making the city way too hot and it's clean and whatever, right? Like, it's a really interesting thing.

There's a lot of YouTube slop videos that you could watch that show how cool it is. But anyways, part of that line project is also DataVault, who is making a data center. And this is real. This is definitely happening. They've already broke ground on, like, a two gigawatt data center.

And again, like, for context, XAI's entire training infrastructure today is 200 megawatts, right? And they've spent, like, $10-plus billion on it, right? Like, you've got to rationalize these numbers. It's, like, 10x more, right? So DataVault is going to invest $20 billion in U.S. data centers in addition to building a bunch in the Middle East.

And then a bunch of American companies are going to invest as well. So the total investment number is, like, $80 billion. DataVault is the data center company for Humane, which is sort of the neocloud for the data center company. And so Humane, they've signed deals to, like, get custom CPUs from Qualcomm and buy a bunch of AMD GPUs as well as NVIDIA GPUs.

Humane is actually the Aramco folks, the Aramco, Aramco Digital folks, which is the vast majority of Grok's revenue as well. So there's likely Grok stuff in there. I think, like, 90-plus percent of Grok's revenue or they've got money from them is coming from the Aramco Digital folks. And then there's also an AWS thing, right?

So all these American companies are also, like, coming together with the Middle East. So the question is, like, you know, why the heck are we sending all these GPUs to the Middle East is what, like, the EAs would say. And then the, like, capitalists will be like, fuck, yeah, GPUs and money, right?

So, like, it depends on what side of the fence you're on. And I like to think, you know, I'm on one side of the fence, so I'll tell you which side. You could guess. You could guess. So, you know, there are real criticisms, right? Like, what if these GPUs get smuggled, right, to China?

Because it's not like Saudi Arabia and the UAE are allies. They are – they conveniently play both sides, right, and always have. So there's always that risk. There's all these security requirements like, okay, what if – okay, they're not – maybe the GPUs don't get smuggled there. Maybe they get rented to China, right?

You know, there's the argument that these GPUs would have been used in the U.S. anyways. And so – and we know that export restrictions and things like that, the enforcement of them sucks. So it's not like the U.S. government apparatus is going to actually be able to effectively enforce this, right?

So there's – there are real risks here. And also, like, giving power to authoritarian countries and – that have, like, you know, kingdoms, right? It's the kingdom of Saudi Arabia. It's princes, right? You know, it's a little bit odd. So that's sort of the arguments against it. And then the supporters are obviously, right, like, yes, more GPUs.

Most of these GPUs go to American companies, right? Like, an easy example is OpenAI, right? They want a lot, lot more GPUs. Microsoft OpenAI deal sort of split up partially because OpenAI wanted so many GPUs, and Microsoft was like, no, you're crazy. Like, that's way too much. Like, you don't have revenue.

And OpenAI is like, but we can convince someone to build it for us, right? Because OpenAI is really nice, right? Like, they get to rent the GPUs. And so, like – but the thing is, their provider has to build the data center and buy all the GPUs, and then they don't get payback for two or three years, right, of renting GPUs, even though the contract is for, like, five years.

So, yeah, there's all this theoretical profit, but you don't actually get the money back on the cluster until year two or three. So, if Microsoft now builds a $100 billion data center for OpenAI, and then OpenAI doesn't raise $150 billion to pay for it, then, you know, Microsoft is just left holding the bag.

And who wants that big of a data center, right? So, unless you believe, like, AI's demand is, like, unlimited, then, you know, you kind of – there's a big risk here. And the nice thing is, like, you know, a fool is easily parted with their money, and SoftBank and the Middle East are the most stupid investors in the world, potentially.

And so, you know, actually, I think they're fine. I think they're fine. I think these are good deals. But, like, you know, OpenAI gets to have someone pay for the cluster, buy all these GPUs, build all the data centers, and then, like, with the promise, they're going to rent them, right?

And the same applies to all the other AI labs across the West, right? And so, the argument that, like, I would make is that, you know, OpenAI, because of this, will have more compute in 2027 or 2028 than they would have had without it, right? It wouldn't have been built in the U.S.

Also, there's big problems with America in terms of being able to build GPUs, right? And that is that there is a massive deficit of power, right? So, there's no access on the chart on the left for a reason, all right, for people taking a picture. But on the right, you know, Middle East is building, you know, by 2030, like, four gigawatts, right?

I said they promised five, but, like, from what we see, they'll have, like, four by 2030. The U.S. has a humongous, humongous deficit on power. I will skip forward a bit. Power, China's good at making power. U.S. is not, right? This is an interesting one, right? So, I presented this to the Secretary Wright and Bergman in February in D.C.

And it's a really interesting chart, right? It's, like, so on the right is, like, sort of our data center data, right? It's saying there is a 63 gigawatt shortfall of power in the U.S., right, based on the data centers that we see under construction. And we're tracking every single one site by site, et cetera, building by building, with satellite photos, permits, regular providing, all this stuff, right?

And on the left is the U.S. power grid, right? So, you can have a lot of, like, strong assumptions about renewables, biobatteries, and then all of the single cycle and dual cycle gas reactors that are being installed. And then you say, hey, actually, almost none of the coal is going to turn off that is planned to turn off.

And you still have this massive, massive, you have 44 gigawatts being added, but you have on the power side, and that's assuming, you know, with the fluctuations and all that, right? Whereas you're adding 100 gigawatts of data center capacity. So, the U.S. simply just doesn't have enough power unless we do something.

And so, the argument sort of is, like, well, geopolitically, the U.S. can't build enough data centers, can't build enough power, whether it's for the lack of skilled labor, whether it's for the lack of, you know, regulatory regulators holding stuff up, federal not as much anymore, but local, state, and also in America, we have this beautiful thing called utility companies that are regulated monopolies that get to do whatever they want, right?

If anyone has a power bill in California, they understand that the utilities fucking suck. Anyways, there's all these issues with building power in the U.S., right? And if we go back, China has no problem, right? They added an entire U.S. grid in seven years, right? And so, and we're talking about adding, you know, you see four terawatt hours on the right graph.

I'm talking about adding 100 gigawatts as a problem, right? Because the U.S. just doesn't know how to. Or rather, hasn't done it in, like, four decades. And so, that's the big sort of challenge. So, for every Stargate that's out there, right, you know, 220 megawatts for these four building, like, things, and there's two of them already almost completed, and then there's another six going up.

That's 1.2 gigawatts total for Stargate. For every one of these that's happening, there's so many projects that are failing, right? Like, I've literally pitched a coal stock to my clients in Ohio, and the stock's 3x because there's power issues. Like, it's crazy how power-constrained America is, right? And so, the geopolitics here is, like, do you build in the Middle East or not?

China's got no power problems. Even if their chips are less efficient, it doesn't really matter, right? Like, I think the AI race is a very geopolitical and interesting one. And so, I'll sort of leave it there. I know there's a couple more slides, but, yeah. A couple questions. Two minutes.

Is SMIC dependent on ASML or what do you work on at all? SMIC, China's TSMC, is dependent entirely on Western tools today, but there are many Chinese tool companies. What's interesting is they'll buy billions of dollars of American, Japanese, and Dutch equipment, including ASML, but then they'll put it next to their domestic tool, and then they'll run wafers through both, and then they'll just, like, learn how to improve it.

They'll also tear them down and reverse engineer them. Yeah, so it's cool. We'll see you next time. We'll see you next time. We'll see you next time. We'll see you next time.