The NVIDIA Bank: Jensen's $500B Wall Street Deal and GPUs as an Asset Class

We unpack the massive $500 billion financing commitment tied to NVIDIA GPUs. Does this represent a new financing structure or a circular trade?

We also cover the risks around AI demand, GPU pricing, and whether this approach could reshape how AI infrastructure is funded.

------
🔒 Check Out Our Sponsor: LEDGER AGENT STACK 🔒
https://developers.ledger.com/docs/ai-tools/overview/?utm_source=Audio&utm_medium=Podcasts&utm_campaign=Limitless

------
🌌 LIMITLESS HQ ⬇️

NEWSLETTER:    https://limitlessft.substack.com/
FOLLOW ON X:   https://x.com/LimitlessFT
SPOTIFY:             https://open.spotify.com/show/5oV29YUL8AzzwXkxEXlRMQ
APPLE:                 https://podcasts.apple.com/us/podcast/limitless-podcast/id1813210890
RSS FEED:           https://limitlessft.substack.com/

------
TIMESTAMPS

0:00 AI Bubble or New Asset Class
2:09 Jensen Orchestrates the $500B Deal
3:17 How the GPU Financing Works
9:38 Where the Money Comes From
11:32 GPUs vs Mortgage-Backed Securities
14:04 Why Supply Still Looks Tight
19:28 Agents Drive Near-Term Demand
20:36 The Bear Case Risks
22:50 Tracking the Real Warning Signs
25:01 Why the Bull Case Still Holds
28:52 NVIDIA and the GPU Future

------
RESOURCES

Josh: https://x.com/JoshKale

Ejaaz: https://x.com/cryptopunk7213

------
Not financial or tax advice. See our investment disclosures here:
https://www.bankless.com/disclosures

Josh works with Anthropic as a contractor. All views expressed are his own and do not represent Anthropic, its leadership, or its affiliates. Nothing in this episode is investment advice.

Episode Transcript

Ejaaz: Earlier this week, the most powerful people in finance stood around a table (00:00):
undefined

Ejaaz: next to Jensen Huang and announced they'd raised $500 billion to buy NVIDIA GPUs. (00:03):
undefined

Ejaaz: Now, if you're listening to this and you're thinking this is just an AI bubble (00:09):
undefined

Ejaaz: circular economy type thing, you might not actually be wrong. (00:13):
undefined

Ejaaz: Larry Fink, the head of BlackRock, actually likened this deal to mortgage-backed (00:16):
undefined

Ejaaz: debt securities of the early 2000s. (00:21):
undefined

Ejaaz: And if that sounds familiar, to which he created. Yes, to which he created. (00:23):
undefined

Ejaaz: And if you're likening that to a kind of like a PTSD flashback, (00:27):
undefined

Ejaaz: that's because that's exactly what happened in the 2008 financial crisis itself. (00:31):
undefined

Ejaaz: But if you look at the news in general, if you look at the way that this deal (00:37):
undefined

Ejaaz: is structured, it might actually hint at something completely different. In fact, the opposite. (00:40):
undefined

Ejaaz: GPU prices for renting has gone sky high. It's up 40% on the year. (00:46):
undefined

Ejaaz: And there's not enough GPUs to back a lot of the deals that Microsoft, (00:50):
undefined

Ejaaz: Google, Anthropic and OpenAI are signing with NVIDIA. So the question that we're (00:54):
undefined

Ejaaz: going to unpack on the show is, is this very much a bubble back deal? (00:58):
undefined

Ejaaz: Or is this something completely different that we're missing? (01:02):
undefined

Josh: A new paradigm of investing, a new paradigm of financial manufacturing and construction. (01:04):
undefined

Josh: This is a new investable asset class. Yeah, this is an entirely new thing. (01:11):
undefined

Josh: GPUs. Who would have thought? (01:14):
undefined

Josh: Michael Burry, the guy who's responsible for the big short, he was like, (01:15):
undefined

Josh: no, the price of these things are going down only. It turns out he could not have been more wrong. (01:18):
undefined

Josh: And now Jensen has assembled Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR. (01:22):
undefined

Ejaaz: It's the Avengers of finance. It's the Avengers of finance. (01:28):
undefined

Josh: It's like Apollo alone has a trillion dollars. (01:32):
undefined

Josh: Of assets managed. Blackstone has over 1.3 trillion. (01:35):
undefined

Josh: Brookfield is over a trillion. And combined, that's 3.4 trillion. (01:38):
undefined

Josh: BlackRock is bigger than all three of those combined. (01:42):
undefined

Josh: And they're all doing this together. And together, they've signed this thing (01:45):
undefined

Josh: called a memorandum of understanding. Now, I had to actually look up what this (01:48):
undefined

Josh: means because I had no idea. (01:51):
undefined

Josh: A memorandum of understanding, or an MOU, is a formal, usually non-binding document (01:53):
undefined

Josh: signed by two or more groups. (01:57):
undefined

Josh: It shows that the groups share a common goal and plan to work together. (01:59):
undefined

Josh: So this is not a contractual obligation. We have to start with that. (02:01):
undefined

Josh: This is not a guarantee that $500 billion is going to flow into this new economy. (02:04):
undefined

Josh: But it is an intention that all of these people are going to be aligned to work (02:09):
undefined

Josh: together towards funding this next build out of AI. And what I found most interesting (02:12):
undefined

Josh: is that this was actually orchestrated entirely by Jensen. (02:17):
undefined

Josh: Jensen reached out to all of these banks himself personally. (02:20):
undefined

Josh: And he said, hey, I'd like to work together on this thing. And not a single (02:24):
undefined

Josh: bank that he reached out to said no. (02:27):
undefined

Josh: So here we are now with a moment on CNBC in which they're all sitting around (02:29):
undefined

Josh: a table talking about how they are committing $500 billion to this new asset class. (02:32):
undefined

Josh: And this is unbelievable. This feels like a, for better or worse, (02:37):
undefined

Josh: a brand new paradigm for the AI era in which now the collective force of the (02:41):
undefined

Josh: United States banking system is like starting to get behind this. (02:47):
undefined

Josh: And I should say this is not for AI as a whole. This is purely for NVIDIA as a company. (02:52):
undefined

Ejaaz: Yeah, and I want to take a moment to actually explain what's happening here, (02:57):
undefined

Ejaaz: because I think there's a lot of confusion. (03:01):
undefined

Ejaaz: There's a lot of headlines that people are getting worried over. (03:03):
undefined

Ejaaz: NVIDIA stock tanks 4% on the news, but I think that's a little too early to (03:05):
undefined

Ejaaz: judge. So firstly, what does this structure sort of look like? (03:10):
undefined

Ejaaz: Well, it's what you're seeing on the screen right now, which is essentially. (03:14):
undefined

Ejaaz: There's a problem in AI right now, which is all these hyperscalers, (03:18):
undefined

Ejaaz: all these AI labs, Anthropic, OpenAI, Google, Meta, you name it, (03:22):
undefined

Ejaaz: have spent a lot of money to buy GPUs. (03:26):
undefined

Ejaaz: The reason why they're doing this is to train and inference brand new AI models, (03:30):
undefined

Ejaaz: which they have a lot of paying customers for. (03:33):
undefined

Ejaaz: But the issue they're facing is the money they've invested, which is now to (03:36):
undefined

Ejaaz: the tune of $2.6 trillion converted over the next couple of years, (03:40):
undefined

Ejaaz: I believe, is not enough for them. (03:44):
undefined

Ejaaz: So much so that they're going into negative cash flow. So what happens when (03:47):
undefined

Ejaaz: you've spent all the money that you have in your company, in your balance sheet? (03:50):
undefined

Ejaaz: You need to go to Wall Street. That's exactly what NVIDIA, specifically Jensen, has brokered. (03:53):
undefined

Ejaaz: He's gone to Wall Street and he said, listen, we need more money to build more (03:59):
undefined

Ejaaz: GPUs to sell to these different customers so that they can produce their products (04:03):
undefined

Ejaaz: and services, their new models. (04:08):
undefined

Ejaaz: And Wall Street has gone back and said, I have an issue with this, (04:09):
undefined

Ejaaz: Jensen, which is GPUs aren't a versatile asset. like they can only be used for (04:12):
undefined

Ejaaz: one thing specifically, which is either training a model or inferencing a model, (04:17):
undefined

Ejaaz: and it's only customer-specific. (04:20):
undefined

Ejaaz: And Jensen responded to them and said, that's not true at all. (04:22):
undefined

Ejaaz: GPUs, specifically NVIDIA GPUs, are the most versatile asset out there. (04:25):
undefined

Ejaaz: You can use it for anything. You can use it for training, you can use it for (04:30):
undefined

Ejaaz: inference, and you can use it for any model, whether it's Claude, (04:32):
undefined

Ejaaz: whether it's GPT, whether it's Gemini, whatever. You can use it for it, (04:35):
undefined

Ejaaz: which means that it's a versatile customer base. (04:39):
undefined

Ejaaz: Earns a lot of money. And then Wall Street shot back at him and said, (04:42):
undefined

Ejaaz: well, hang on a second, these GPUs die after a couple of years. (04:46):
undefined

Ejaaz: And Jensen goes, that's not actually true. In fact, we have 10-year-old GPUs (04:49):
undefined

Ejaaz: that are being re-signed for another 10 years right now today at a higher price (04:54):
undefined

Ejaaz: than they sold earlier on. (04:58):
undefined

Ejaaz: So basically what he's pitched them is this is a new asset class and it can earn a ton of money. (04:59):
undefined

Ejaaz: And so Wall Street looked at this, some of the biggest financial powerhouses (05:05):
undefined

Ejaaz: in the world and thought, you know what? (05:08):
undefined

Ejaaz: He might be right. This is an asset class that can be likened to property or (05:10):
undefined

Ejaaz: railroads back in the day. (05:14):
undefined

Ejaaz: And that's why Larry Fink is comparing it to the 2008 mortgage-backed securities. (05:16):
undefined

Ejaaz: Now, if you're wondering, okay, well, this is like a financial crisis type thing, (05:20):
undefined

Ejaaz: you might be right, except there was like a few different things going on there, (05:25):
undefined

Ejaaz: which we'll unpack later in the episode. (05:28):
undefined

Josh: Yeah, it feels very much like AI compute is equivalent to revenue. (05:30):
undefined

Josh: And these are very now like durable, appreciating assets that can yield value over time. (05:34):
undefined

Josh: So when you think of like a bond, per se, the value of a bond is implied to (05:40):
undefined

Josh: go down over time, the underlying asset like the US dollar due to inflation, (05:45):
undefined

Josh: but the yield it will come up with is going to outpace that and then some hopefully (05:49):
undefined

Josh: the construction of a GPU is that not only do you get a yield in terms of the (05:54):
undefined

Josh: value creation off the back of token generation. (05:59):
undefined

Josh: But you now also have an asset that is likely going to appreciate. (06:01):
undefined

Josh: And even in the face that it doesn't, Jensen is giving plunge protection. (06:05):
undefined

Josh: So 25%. Yes. Depreciation insurance of up to 25% to help the banks get these (06:10):
undefined

Josh: marginal deals over time. So banks initially were concerned. (06:17):
undefined

Josh: They don't want to fund this because they don't want (06:19):
undefined

Josh: nvidia to come out with the new gpu that's a thousand times better than this (06:21):
undefined

Josh: one and it's going to knock all the margin out from underneath them (06:24):
undefined

Josh: who knows the roadmap about nvidia gpus better than anyone else it's jensen (06:27):
undefined

Josh: the guy who's building it so baked into this contract is the idea that (06:31):
undefined

Josh: jensen will ensure you he will make sure that hey your gpus are not going to (06:34):
undefined

Josh: fall drastically over time everything is going to be smooth and predictable (06:38):
undefined

Josh: and we'll work together to fund these companies that don't have (06:41):
undefined

Josh: the ability to do so so this comes in the form of like this long-term (06:44):
undefined

Josh: Debt and asset backed structures. (06:48):
undefined

Josh: And you think of it like if you're not a hyperscaler, if you're not Google who (06:51):
undefined

Josh: has a couple hundred billion dollars to spend off your balance sheet this year, (06:54):
undefined

Josh: but you still want to compete in the world of AI, you still need GPUs, these are who you go to. (06:58):
undefined

Josh: And they will offer you GPUs in exchange for interest on these GPUs. (07:02):
undefined

Josh: And in the worst case that it doesn't work out, they can just claw back the GPUs. (07:07):
undefined

Josh: NVIDIA can claw back those GPUs and turn it into their own NeoCloud, (07:11):
undefined

Josh: give it to another neocloud but the idea is that these assets are valuable they're increasing in money (07:14):
undefined

Josh: they are transferable in an easy way that you can just take the gpu and plug (07:19):
undefined

Josh: it in somewhere else or give the actual data center control over to someone else (07:24):
undefined

Josh: and it's a really lucrative kind of bizarre thing it's like okay if you're a (07:28):
undefined

Josh: bank 500 billion dollars (07:32):
undefined

Josh: you have insurance you are now able to allocate this (07:35):
undefined

Josh: incredibly valuable capital resource to anybody who you want and collect a pretty (07:39):
undefined

Josh: high rate of return on top of that (07:43):
undefined

Josh: and nvidia wants in on this too so initially it wasn't for nvidia nvidia now (07:45):
undefined

Josh: is given the option to backstop up to 25 of each opportunity so that is 125 (07:49):
undefined

Josh: billion dollars at the ceiling (07:55):
undefined

Josh: and basically now nvidia and co the avengers get to roam around choose who they (07:56):
undefined

Josh: would like to give these gpus to and wrap it up in a really interesting financial (08:01):
undefined

Josh: product that they can go off and, I guess, monetize. (08:05):
undefined

Ejaaz: And I want to stress that this is only for NVIDIA GPUs specifically. (08:08):
undefined

Ejaaz: Jensen brokered this deal for his company only. (08:12):
undefined

Ejaaz: And he has a reason to do that because GPUs for the longest time has been very (08:16):
undefined

Ejaaz: broad based. If you look at some of the GPUs that Google makes or that Meta (08:21):
undefined

Ejaaz: is making or that even OpenAIR and Anthropic are reportedly working on their own specialized chips. (08:25):
undefined

Ejaaz: These are exactly what I just said. They're specialized. They can't be used (08:30):
undefined

Ejaaz: for many other models. It's only used specifically for their things. (08:33):
undefined

Ejaaz: So it's a much more niche case to create a type of loan or credit-backed security for. (08:35):
undefined

Ejaaz: Jensen has the opposite issue, which is like, it's too broad. (08:41):
undefined

Ejaaz: But that makes it an amazing financial asset. So in effect, NVIDIA is sort of becoming a bank. (08:45):
undefined

Ejaaz: And I wouldn't be surprised if Jensen starts to make a lot of money from this. (08:51):
undefined

Ejaaz: Over the last couple of weeks, something that he's also started doing is backstopping (08:55):
undefined

Ejaaz: specific Frontier AI labs and saying, hey, don't worry, I got you. (08:59):
undefined

Ejaaz: I'll front up the money that you need to purchase my GPUs. (09:04):
undefined

Ejaaz: And in return, whatever money you make on the products that you're building, (09:07):
undefined

Ejaaz: you can give me a revenue split from that. (09:12):
undefined

Ejaaz: I think he signed like a reportedly 10% revenue split from Safe Super Intelligence, (09:14):
undefined

Ejaaz: which is Ilya Sutskiver's new lab for their breakthrough that they're launching pretty soon. (09:19):
undefined

Ejaaz: And I think he's gonna do the same for a lot of neoclouts like CoreWeave, (09:23):
undefined

Ejaaz: Nebius and such like that, that are reporting crazy earnings. (09:26):
undefined

Ejaaz: I think this morning, CoreWeave reported 464% increase in revenue year upon (09:29):
undefined

Ejaaz: year, which is just a precursor to like the insane demand that they're seeing right there. (09:34):
undefined

Ejaaz: But then a question that comes into mind is, where on earth is this money coming from? (09:38):
undefined

Ejaaz: And on the screen here, it's like the main claimants are pension funds, right? (09:43):
undefined

Ejaaz: So pension funds who have amassed a large amount of wealth and typically don't (09:49):
undefined

Ejaaz: invest in high volatile type assets. (09:53):
undefined

Ejaaz: They kind of stick to real estate, very low interest types of things, (09:56):
undefined

Ejaaz: are the ones that are going to be backing a lot of this new GPU asset class. (10:00):
undefined

Ejaaz: And you have the biggest, most (10:04):
undefined

Ejaaz: powerful financial people in the world that are kind of pushing this on. (10:05):
undefined

Ejaaz: And so I'm thinking, is this reckless behavior? (10:09):
undefined

Ejaaz: Well, if you take the word of Goldman Sachs CEO David Solomon, (10:12):
undefined

Ejaaz: he goes, $500 billion sounds like a lot, but there are $9 trillion in US money (10:16):
undefined

Ejaaz: market funds and more than $100 trillion in US equities. (10:21):
undefined

Ejaaz: He has a deep belief in this opportunity and Goldman brings its extraordinary (10:26):
undefined

Ejaaz: distribution network. Larry Fink, CEO of BlackRock, also says, (10:30):
undefined

Ejaaz: he said it's a very attractive opportunity with long-dated, long-term returns. (10:34):
undefined

Ejaaz: They will be talking to pension funds. So it seems like the two most powerful (10:38):
undefined

Ejaaz: financial connoisseurs in the world are convinced that this new asset class (10:41):
undefined

Ejaaz: is a very real thing, which means that they've probably looked at the balance sheets. (10:44):
undefined

Ejaaz: They've probably looked at the revenue demand that a lot of these frontier labs (10:48):
undefined

Ejaaz: that are meant to be purchasing these things are going to do. (10:50):
undefined

Ejaaz: And they're looking at it and they're saying, this is an obvious no-brainer. (10:53):
undefined

Ejaaz: Now, if you're listening to me and you're thinking, dude, this happened with (10:56):
undefined

Ejaaz: railroads and it didn't work out. (10:59):
undefined

Ejaaz: This happened with the housing environment, mortgage-backed debt securities (11:01):
undefined

Ejaaz: in 2008, that didn't work out. (11:05):
undefined

Ejaaz: I have to say, it's a very different story on our end. (11:07):
undefined

Josh: Well, that's what I was going to ask you. I was going to say like, (11:10):
undefined

Josh: hey, obviously they're going to come out and say these things. (11:13):
undefined

Josh: I mean, we've seen them manipulate markets for a long time. We just saw what (11:15):
undefined

Josh: Citadel did to Leopold. It's like everyone is very clearly out in their own (11:18):
undefined

Josh: best interest. So if we look at this deal, okay, they're not going to say it's (11:22):
undefined

Josh: anything less than exceptional. (11:25):
undefined

Josh: So how do we kind of vet this? How do we fit this into a specific piece of context (11:27):
undefined

Josh: that I guess we could reference? (11:32):
undefined

Josh: And there's an interesting example of like aircraft finance versus mortgage (11:32):
undefined

Josh: finance, because this is something that has happened in the past where (11:37):
undefined

Josh: when you have an expensive standardized asset that's transferable between operators (11:41):
undefined

Josh: and has a lot of demand for these secondary markets, it creates this interesting marketplace that (11:45):
undefined

Josh: I think is much more comparable to aircrafts than mortgages. (11:51):
undefined

Josh: And I'll explain. So like a Boeing 747 or 737 or whatever, that's been built (11:54):
undefined

Josh: like 20 years ago. And trust me, you've flown on these. The airlines kind of (11:59):
undefined

Josh: suck. You're flying in some old planes. (12:02):
undefined

Josh: That is just as valuable today (12:04):
undefined

Josh: As it was 20 years ago, because they're able to derive so much value from it. (12:06):
undefined

Josh: It does the same exact job. (12:11):
undefined

Josh: The same plane that was built today is doing the same job that was built 20 (12:13):
undefined

Josh: years ago. And sure, perhaps you would prefer to fly on the newer plane. (12:16):
undefined

Josh: But the reality is, is that tickets are sold out on the 2005 plane and the 2025 plane. (12:20):
undefined

Josh: And when you think of GPUs, they exhibit a lot of the same traits and characteristics (12:25):
undefined

Josh: as an airplane, where it's expensive, standardized, it's transferable. (12:30):
undefined

Josh: It has a lot of liquidity in secondary markets. (12:33):
undefined

Josh: And I think this is an interesting way of looking at it relative to mortgage (12:35):
undefined

Josh: finances, which is where we got in trouble. And this isn't the first time this has happened before. (12:39):
undefined

Josh: There is something that has been done similar to this with Broadcom, (12:43):
undefined

Josh: where like that Google Anthropic structure actually runs through this thing (12:46):
undefined

Josh: called an SPV, a special purpose vehicle that buys TPUs and leases them with (12:50):
undefined

Josh: Broadcom providing the residual value guarantees, (12:54):
undefined

Josh: and then Apollo and Blackstone supplying the private credit to fund all of this. (12:57):
undefined

Josh: So people have experimented with these structures, before. It has worked. (13:00):
undefined

Josh: We haven't seen it at this scale. I mean, the alarm bells are partially ringing. (13:05):
undefined

Josh: I'm like, just out of instinct, like out of an e-joke reaction, (13:09):
undefined

Josh: like, oh, wow, this is a lot of money. This is a lot of powerful people who (13:12):
undefined

Josh: can very much control and sway the way the market moves. (13:16):
undefined

Josh: So far, it seems like a pretty reasonable thing. It's like, hey, we need GPUs. (13:19):
undefined

Josh: GPUs are transferable. They're kind of like they're fungible, (13:23):
undefined

Josh: I guess. I'm like thinking of the word. I'm like, well, this feels kind of crypto adjacent. (13:28):
undefined

Josh: There's like these fungible assets that can be transferred that are valuable. (13:31):
undefined

Josh: So I don't know. There's a chance this goes over. OK. (13:34):
undefined

Ejaaz: And it's important to not extrapolate too far into the future, (13:37):
undefined

Ejaaz: like what we can feasibly attain from the data, which, by the way, (13:41):
undefined

Ejaaz: is publicly available. If you're listening to this and you don't believe anything that we're saying. (13:46):
undefined

Ejaaz: Maybe we should actually link to a bunch of sources. Maybe we'll link this artifact (13:50):
undefined

Ejaaz: that you're seeing on the screen right now. (13:52):
undefined

Ejaaz: The data is all available through quarterly earnings of every single company (13:54):
undefined

Ejaaz: that is leading at every layer of the AI stack. So you can see the data, (13:57):
undefined

Ejaaz: digest it yourself and figure it out for yourself. (14:02):
undefined

Ejaaz: But what I will say is when you look at the 2008 financial crisis, (14:04):
undefined

Ejaaz: when you look at the railroad crisis, when you look at the telecom crisis back (14:08):
undefined

Ejaaz: then there was a huge amount of oversupply. (14:12):
undefined

Ejaaz: Which didn't have the back demand that it stated it had so 2008 people assumed (14:16):
undefined

Ejaaz: that property prices would just keep going up and at some point (14:22):
undefined

Ejaaz: that got two head over heels if you look at the railroad they built too much (14:26):
undefined

Ejaaz: if you look at the telecom they built too many cables right in this case (14:29):
undefined

Ejaaz: we're constrained by a few things number one physically it takes so many different (14:34):
undefined

Ejaaz: substrate layers to build a gpu. (14:38):
undefined

Ejaaz: Every single layer right now in the world of physical atoms is incredibly constrained. (14:41):
undefined

Ejaaz: There's not enough. The GPU demand is overweight, the actual supply that is available. (14:45):
undefined

Ejaaz: Number two, the AI demand, which is driving GPU demand, is accelerating much (14:51):
undefined

Ejaaz: faster than the supply itself can. (14:56):
undefined

Ejaaz: So if you look at memory as a basis for this, they can increase capacity. (14:59):
undefined

Ejaaz: This is the top docs, top memory manufacturers can increase capacity around 20% per year. (15:02):
undefined

Ejaaz: But demand is compounding at 45 to 60% per year. (15:08):
undefined

Ejaaz: So if you do the math, if that continues, you're going to be in a constrained (15:12):
undefined

Ejaaz: supply state for like at least until 2028 or 2029 until some of these other (15:16):
undefined

Ejaaz: chip fabs can get increased. So we're still kind of in a holding period, right? (15:21):
undefined

Ejaaz: Now, if you look at the backlog for some of these companies. (15:25):
undefined

Ejaaz: You might be like, well, customers don't want AI as much as these guys are making (15:29):
undefined

Ejaaz: it out to be. They're just kind of like pushing their bags. (15:33):
undefined

Ejaaz: Well, look at Google's backlog. It doubled this year in a matter of months. (15:35):
undefined

Ejaaz: It is now at $460 billion, and we're at the halfway mark of this year. (15:39):
undefined

Ejaaz: It's probably going to increase even more. This is the case across Microsoft (15:43):
undefined

Ejaaz: and a bunch of other hyperscalers as well. (15:46):
undefined

Ejaaz: Then if you look at the GPU rental prices, my favorite thing, (15:48):
undefined

Ejaaz: which kind of came out this morning, Josh, or maybe yesterday, (15:51):
undefined

Ejaaz: CallWeave had their earnings, and they said, we recently signed an A100 contract that extends into 2029. (15:55):
undefined

Ejaaz: For those of you who don't know, and A100 is an NVIDIA GPU that was created in 2020. (16:00):
undefined

Ejaaz: And its lifecycle prediction back then was three and a half years. (16:05):
undefined

Ejaaz: Now they're predicting that it's actually going to be good to go until 2029. (16:09):
undefined

Ejaaz: That's because it's not just being used for bleeding edge training. (16:13):
undefined

Ejaaz: It's being used for inference and a bunch of other stuff. So the point is, (16:15):
undefined

Ejaaz: these GPUs are very versatile, NVIDIA specifically, and that's why they raised a crap ton of money. (16:18):
undefined

Josh: Yeah, and more valuable over time. It's this really bizarre thing in which the (16:22):
undefined

Josh: useful lifecycle of a hardware object is increasing instead of depreciating for the first time. (16:27):
undefined

Josh: And we've never really seen this phenomenon at scale before. (16:33):
undefined

Josh: But like you mentioned, there's not many reasons in which it's going to slow (16:36):
undefined

Josh: down in the near term future. When I look at this, I'm kind of looking at it (16:39):
undefined

Josh: like around the corner. And then you can't really see around the next corner, (16:43):
undefined

Josh: but we have an idea of the first corner. And that first corner is sold out supply (16:47):
undefined

Josh: for at least 2027, likely 2028. (16:51):
undefined

Josh: And then by the end of the decade, 2029, 2020, 2030, we start to run into larger (16:55):
undefined

Josh: constraints, mostly around energy and power. (17:01):
undefined

Josh: And we start to like run into resource constraints that we don't quite have now. (17:04):
undefined

Ejaaz: So are you basically saying there's like multiple corners, Josh? (17:08):
undefined

Ejaaz: Like I'm curious, like, for the GPU specifically, do you think it's like six (17:11):
undefined

Ejaaz: months? Do you think it's also like 12 months? Like, what's your guess if you had to... (17:15):
undefined

Josh: Well, there's a somewhat clear trajectory for the next 24 months, (17:19):
undefined

Josh: like 18 to 24 months in terms of... (17:23):
undefined

Josh: It seems fairly predictable where we know how many, like the lithography machines, (17:26):
undefined

Josh: we know how many chips they can create. (17:30):
undefined

Josh: Then we know how many of those chips can be packaged into usable chips. (17:32):
undefined

Josh: And then we know roughly how many data centers can be built that can actually (17:36):
undefined

Josh: turn those chips on and power them. (17:40):
undefined

Josh: And you can somewhat project that out up to 24 months loosely, (17:41):
undefined

Josh: very loosely, because there is only so much throughput for these machines. (17:45):
undefined

Josh: So if you assume everyone's operating at full capacity, you can kind of work (17:50):
undefined

Josh: those numbers backwards and understand like, OK, they're sold out and they're (17:53):
undefined

Josh: still not going to be enough to satisfy the demand, assuming the demand continues, (17:56):
undefined

Josh: which there is no signs of slowing down. (18:00):
undefined

Josh: All these use cases for AI, particularly around agentic AI, require a tremendous (18:02):
undefined

Josh: amount of inference and even including efficiency upgrades to the software, (18:07):
undefined

Josh: something similar to what we imagine SSI is working on. (18:10):
undefined

Josh: There's still a huge amount of demand that will fulfill that this is jevin's (18:13):
undefined

Josh: paradox which we really should name jensen's paradox because that seems to be (18:16):
undefined

Josh: a little more accurate in terms of how this is working (18:20):
undefined

Josh: but we can kind of project out till then and we know all right gpus fully sold (18:22):
undefined

Josh: out fully constrained after that things get a little (18:26):
undefined

Josh: more hairy right it's because you have to assume by that time we'll have something (18:30):
undefined

Josh: similar to agi asi self-recursive improvements we should be getting a lot of (18:33):
undefined

Josh: innovation breakthroughs around efficiency and software and we don't really know what (18:38):
undefined

Josh: the power market's going to look like. We're not sure if we're able to make enough (18:42):
undefined

Josh: Energy to satisfy the demand of the gpu centers that are being projected out (18:46):
undefined

Josh: into 20 to 30 so it seems like this is a very long duration thing that's going (18:50):
undefined

Josh: to need to play out but in the short term in the intermediary term it seems like i mean (18:56):
undefined

Josh: i'd like to find the steel man against this because this seems important and (19:01):
undefined

Josh: we should talk about like what are the possible ways in which this breaks (19:04):
undefined

Josh: but just looking at demand of inference and our capability of serving inference (19:07):
undefined

Josh: and there is a huge mismatch in the case of inference demand that's just not (19:12):
undefined

Josh: going to be met for a really long time. (19:16):
undefined

Josh: So those H100s from 2020, or the A100s, I should say, from 2020, (19:17):
undefined

Josh: are still going to be useful in 2027, 2028. (19:21):
undefined

Josh: And that is particularly valuable when you're investing in GPUs at this scale. (19:25):
undefined

Ejaaz: Well, actually, now that you say it, a lot of that inference demand, (19:29):
undefined

Ejaaz: at least in the next six months or so, is going to come from AI agents. (19:32):
undefined

Ejaaz: I'm in no doubt about that. (19:37):
undefined

Josh: You said the word agents. It's funny you should mention agents because we have (19:39):
undefined

Josh: something to say about agents from our sponsor of this episode, Ledger. (19:43):
undefined

Josh: If you're building with AI agents, you are probably worried about security and (19:47):
undefined

Josh: rightfully so because as we've seen recently, these agents have been kind of (19:51):
undefined

Josh: doing some funky things. (19:54):
undefined

Josh: So Ledger has this three-step approach to solving this. The first is that the (19:55):
undefined

Josh: agent proposes a change, then the human approves the change, (19:58):
undefined

Josh: and then the Ledger signer actually enforces this change. (20:01):
undefined

Josh: There's a three-step process to make sure your agent doesn't do anything you (20:04):
undefined

Josh: do not want it to do they have this thing called the ledger agent stack which (20:08):
undefined

Josh: is an open source software stack that gives you a series of tools to help you navigate (20:11):
undefined

Josh: your journey with your agents it works with cloud code it works with codex cursor anywhere that you use (20:15):
undefined

Josh: your ai models it's available now it is open source you can find the link in (20:20):
undefined

Josh: the description down below and thank you so much to ledger for sponsoring this episode the agent (20:23):
undefined

Josh: barrage? I guess we have to talk about the downside effects, (20:28):
undefined

Josh: right? What could happen if things don't go as well as planned? (20:32):
undefined

Josh: And where does that risk actually live? So, Ijaz, it seems like this has been (20:37):
undefined

Josh: fully prepared, lovely by our clawed artifact right here. So, (20:41):
undefined

Josh: what is actually the downside risk? What do we need to look out for when we're (20:44):
undefined

Josh: evaluating how to invest around this? (20:48):
undefined

Ejaaz: Okay. So, we are the Limitless Show. And as anyone who's listened to us for (20:49):
undefined

Ejaaz: a while knows, we are grounded or we are trying to ground ourselves a lot more from the bullish case. (20:53):
undefined

Josh: I mean, if you got to say it. Yeah. (20:58):
undefined

Ejaaz: So there are a few ways where this can obviously go wrong. (20:59):
undefined

Ejaaz: And I want to kind of like walk through some of these and get your take on this, (21:04):
undefined

Ejaaz: Josh. So number one, the thing that's like blaring to me is this is all based (21:07):
undefined

Ejaaz: on the fact that AI demand not only is sustained. (21:13):
undefined

Ejaaz: So you have paying customers to buy cloud subscriptions, GPT subscriptions, (21:17):
undefined

Ejaaz: companies paying tens to hundreds of millions of dollars a year for API access. But... (21:21):
undefined

Ejaaz: That it increases. Right now, it's increasing at a crazy rate. (21:26):
undefined

Ejaaz: We see all these quarterly earnings, revenues compounded between 100% to 500% year upon year. (21:29):
undefined

Ejaaz: It is insane, but that's not sustainable. It's not going to keep doing that. (21:34):
undefined

Ejaaz: It'll presumably eventually plateau. (21:37):
undefined

Ejaaz: So if that does plateau, or in worst case, if that plummets, (21:39):
undefined

Ejaaz: then these companies are going to need fewer GPUs, which means that Jensen's (21:43):
undefined

Ejaaz: $500 billion, these debt-backed securities are going to be in less demand. (21:48):
undefined

Ejaaz: And that's where you might see a default. (21:52):
undefined

Ejaaz: The second major thing here is that the GPUs themselves depreciate a lot faster. (21:54):
undefined

Ejaaz: And that has been the Michael Burry, the guy that did the famous big short back in 2008. (22:01):
undefined

Ejaaz: That's been his view this entire time. He says that the upgrade cycle for a (22:06):
undefined

Ejaaz: lot of these NVIDIA GPUs are actually a lot shorter than what Jensen NVIDIA claims. (22:10):
undefined

Ejaaz: However, in practicality, this seems to not be the case. (22:16):
undefined

Ejaaz: However, NVIDIA is now releasing a lot of GPUs at a much more higher frequency (22:19):
undefined

Ejaaz: rate, which means that they're going to replace more of the GPUs in the prior (22:25):
undefined

Ejaaz: market, in the prior cycle, and they'll start flooding the market. (22:29):
undefined

Ejaaz: My countess of that is simply you can't make GPUs that quickly. (22:32):
undefined

Ejaaz: It takes a lot of technical expertise, and it is limited by the likes of TSMC (22:36):
undefined

Ejaaz: and wafer capacity and a bunch of other technical stuff, which I don't want to get into on this show. (22:40):
undefined

Ejaaz: So I'm struggling to actually see how these two factors might actually be triggered. (22:44):
undefined

Ejaaz: But I don't know if you have a different opinion, Josh. (22:48):
undefined

Josh: Yeah, the thing that I'm looking out for most is the return on invested capital (22:50):
undefined

Josh: from the large hyperscalers. (22:54):
undefined

Josh: It feels like they just run the world. They're spending all the capex. (22:56):
undefined

Josh: They are basically floating the entire economy right now. (22:58):
undefined

Josh: And if the returns on that investment start to go down, for example, (23:01):
undefined

Josh: that seems like a very scary thing. So looking at Google's earnings reports, (23:05):
undefined

Josh: we see like, okay, They have $514 billion dollars. (23:08):
undefined

Josh: They're spending. Can they keep returning revenue on that on schedule? (23:15):
undefined

Josh: If the answer is yes, if there's still revenue to be made on AI spend, that is amazing. (23:19):
undefined

Josh: In the case that that turns and we start seeing earnings reports from companies (23:24):
undefined

Josh: who are spending huge amounts of capex saying, our margins are actually shrinking. (23:27):
undefined

Josh: Our revenue is not coming at the multiple that we expected. (23:31):
undefined

Josh: That seems to be a red flag because that will slow down spending significantly across the board. (23:35):
undefined

Josh: Basically, we want to make sure that all this stays profitable. (23:38):
undefined

Josh: So we want to make sure that Inference demand is actually continuing. (23:40):
undefined

Josh: Companies are actually able to meaningfully monetize this. (23:43):
undefined

Josh: The enterprises that are spending billions, hundreds of billions of dollars (23:46):
undefined

Josh: a year on AI spend, we need to make sure that they're actually getting value. (23:49):
undefined

Josh: Otherwise, they're going to cut those contracts. That is probably the most important thing. (23:52):
undefined

Josh: The second is just monitoring the rental rates. Like currently, (23:56):
undefined

Josh: a lot of companies are terrified to re-sign their long-term GP rental deals (23:59):
undefined

Josh: because the price that they're going to get them at this time around is going (24:02):
undefined

Josh: to be double the price that they got originally. (24:05):
undefined

Josh: That is a phenomenon that like no one was really expecting, but here we are. (24:07):
undefined

Josh: And if that trend continues as well that's something i'm kind of looking out (24:11):
undefined

Josh: for so i'm looking at what is the hour double (24:15):
undefined

Ejaaz: Like three years from now right i mean it. (24:17):
undefined

Josh: Might it might but i'm saying this this is just something that like you should (24:19):
undefined

Josh: keep a close eye on you know how long these (24:23):
undefined

Ejaaz: Contracts are for josh that they're signing i know. (24:24):
undefined

Josh: They vary quite a bit like some are short they're like a year some are longer (24:26):
undefined

Josh: out to like three years maybe um but i they're they're variable and i know that (24:31):
undefined

Josh: when the time has is coming to kind of re-sign this contract (24:36):
undefined

Josh: the price is higher not lower for the same supply so ensuring that this (24:40):
undefined

Josh: continues this trend continues and even if it doesn't continue making sure it (24:45):
undefined

Josh: doesn't flip negative because that might change things and granted nvidia has (24:48):
undefined

Josh: your 25 plunge protection service you're available but you don't really want to put that 125 (24:52):
undefined

Ejaaz: Billion dollars by the way for those of you trying to do the math. (24:58):
undefined

Josh: That's a lot of money man a lot of money um and then third is just like (25:01):
undefined

Josh: what the yields actually is from these gpus like how much the real yield yeah (25:05):
undefined

Josh: the actual real yield and (25:10):
undefined

Josh: these are also like those mous this is an assigned deal and we've had something (25:11):
undefined

Josh: similar to this before remember that crazy project back in the day called project stargate where elon and (25:15):
undefined

Josh: masa or not sorry not elon sam altman and masayoshi son and even donald trump (25:21):
undefined

Josh: all stood in an office together and they said (25:27):
undefined

Josh: we're going to spend x billion dollars on this data center build out (25:29):
undefined

Josh: it hasn't really happened as planned so this is not a contractual obligation (25:32):
undefined

Josh: to spend 500 billion dollars this is a hey dude (25:36):
undefined

Josh: We're all rich. We'll commit to like $500 billion and we'll see how it goes. (25:40):
undefined

Josh: And that's kind of what they have. It's a handshake deal to build this new financial (25:43):
undefined

Josh: economic instrument around the GPU, particularly as it relates to NVIDIA. (25:47):
undefined

Josh: So huge win for NVIDIA, probably a large win for a lot of the companies that (25:51):
undefined

Josh: are not able to afford this, and probably a huge win for the banks. (25:56):
undefined

Josh: At the end of the day, they seem to always win. And that's kind of what the deal is here. (25:59):
undefined

Ejaaz: I am really struggling to think about a world, an alternative scenario, (26:04):
undefined

Ejaaz: where AI doesn't require GPUs, specifically the monopolistic GPUs from NVIDIA. (26:10):
undefined

Ejaaz: They just have such a stronghold on the entire market. (26:19):
undefined

Ejaaz: And even if you have some kind of novel LLM architecture that gets created in (26:23):
undefined

Ejaaz: the future that completely disrupts the current paradigm, you're still going (26:28):
undefined

Ejaaz: to need hardware to run these things. (26:31):
undefined

Ejaaz: And that hardware is very much GPUs that are being designed and created by Jensen Huang. (26:32):
undefined

Ejaaz: So however way I skin this cat, I still think that you're going to need these GPUs. (26:37):
undefined

Ejaaz: You still need token generation. Gavin Baker has made this point across so many (26:42):
undefined

Ejaaz: other podcast episodes in the last two weeks that it's ingrained in my head at this point, right? (26:46):
undefined

Ejaaz: And then the other thing I think about is, okay, well, if NVIDIA becomes a bank (26:49):
undefined

Ejaaz: themselves and they start taking revenue splits from all these frontier AI labs, (26:54):
undefined

Ejaaz: that's a completely new revenue line for NVIDIA. (26:59):
undefined

Ejaaz: So when I think about this with my investing hat on, I'm thinking. (27:01):
undefined

Ejaaz: Okay, not only is NVIDIA supplying the foundational element that is required (27:05):
undefined

Ejaaz: to run and inference these GPUs, train these GPUs, but they're also (27:10):
undefined

Ejaaz: being the ones that are driving cost down per token, right? So like they're (27:14):
undefined

Ejaaz: doing this with their CUDA software mode. (27:19):
undefined

Ejaaz: And then I think about the financing side of things. So they're being the financiers (27:20):
undefined

Ejaaz: of this entire thing as well. (27:24):
undefined

Ejaaz: Now, that does sound like a house of cards if the demand wavers, if the demand plummets. (27:26):
undefined

Ejaaz: And I can easily see the market being very volatile and reacting to any kind (27:31):
undefined

Ejaaz: of headline like they did to this initial headline but i don't know it just (27:34):
undefined

Ejaaz: seems very bullish to me on nvidia at least and yeah i don't really know how (27:37):
undefined

Ejaaz: to think about it yeah yeah. (27:41):
undefined

Josh: And i mean like in this case like i do kind of lean on the opinions of people (27:42):
undefined

Josh: who are more in the know than me yeah someone like elon who is now exclusively (27:46):
undefined

Josh: committed to purchasing only nvidia gpus for the new data center build out (27:50):
undefined

Josh: and they are effectively the best data center builders in the world so i you (27:54):
undefined

Josh: have to like have a little bit of trust in the opinions of the (27:58):
undefined

Josh: true experts who are in the arena doing things when i look at that and i see (28:00):
undefined

Josh: like they exclusively want nvidia and they are building the best fastest most (28:04):
undefined

Josh: efficient data centers i'm like okay that's pretty good signal like micro hard (28:07):
undefined

Josh: the new data center that the spacex ai team is working on (28:11):
undefined

Josh: is i think like a third the footprint of macro (28:14):
undefined

Ejaaz: Hard right no micro hard. (28:17):
undefined

Josh: No no there is there's micro hard is (28:18):
undefined

Ejaaz: A micro hot. (28:20):
undefined

Josh: Yes and micro hard is about a third of the footprint i might be getting this (28:21):
undefined

Josh: wrong half or a third of the footprint of macro hard easy but it contains the (28:24):
undefined

Josh: same cluster of 200 000 gpus wow they just figured out how to do it much more (28:27):
undefined

Josh: efficiently and much more dense so (28:31):
undefined

Josh: there's a huge amount of innovation clearly they know things that the rest of (28:34):
undefined

Josh: the industry does not and when they come out the biggest purchaser right (28:37):
undefined

Josh: yeah and they're committed exclusively to the nvidia gpu wow and when vera rubin (28:40):
undefined

Josh: comes out at scale man oh my god i keep saying this for like holy smokes (28:46):
undefined

Ejaaz: Those models are gonna be insane those models are gonna be absolutely insane. (28:49):
undefined

Josh: So buckle up good time to be good time to be nvidia good time to be a GPU. (28:52):
undefined

Josh: Yeah, that's the update. So is this a house of cards? (28:58):
undefined

Josh: Is it all going to come tumbling down? Is this financial innovation in a new (29:02):
undefined

Josh: era of the United States of GPUs? (29:06):
undefined

Josh: Let us know in the comments down below if you enjoyed this episode. (29:09):
undefined

Josh: Don't forget to share it with a friend who might also enjoy this episode. (29:11):
undefined

Josh: Ejaz and I, cool story. We were walking down the street last night after dinner (29:14):
undefined

Josh: and two people walked up to us. (29:17):
undefined

Josh: They were like, hey, you guys are those podcast guys. You host the show. (29:19):
undefined

Josh: And we were like, yeah, we do. Like, cool. Thanks for watching. (29:21):
undefined

Josh: So that's like so cool whenever that happens. And that is because of you sharing (29:24):
undefined

Josh: with your friends, letting everyone else know that the show exists. (29:28):
undefined

Josh: And if you enjoyed it, you know, don't forget. Thumbs up. You could subscribe. (29:31):
undefined

Josh: You could give us a five-star review on your favorite podcast platform. (29:34):
undefined

Josh: Any parting thoughts, EJs? (29:37):
undefined

Ejaaz: Yeah. Homework for you guys. If you see us on the street, don't be shy. (29:38):
undefined

Ejaaz: Come and say hi. We want to meet you guys. (29:41):
undefined

Josh: Don't be shy. Say hi. (29:43):
undefined

Ejaaz: Yeah. It's lovely to meet you guys. We've now met people. We've now met listeners, (29:44):
undefined

Ejaaz: in New York and we've met them in San Francisco. (29:49):
undefined

Ejaaz: I have people calling in from Europe to my family talking about these random (29:53):
undefined

Ejaaz: guys that yap about AI. Turns out it's us. Like it's really cool to see the (29:56):
undefined

Ejaaz: momentum that we're getting here. (30:02):
undefined

Ejaaz: And it's all thanks to you guys. So if you're one of these people that care (30:03):
undefined

Ejaaz: passionately about what we talk about and tune in every day, thank you so much. (30:07):
undefined

Ejaaz: And if you're not, turn on notifications, please. Subscribe to us. (30:12):
undefined

Ejaaz: We bring the best news, hot, fresh out the oven, every single day, (30:15):
undefined

Ejaaz: or rather four times a week. (30:19):
undefined

Ejaaz: And we would love to hear from you. Leave us a comment, DM us on X. (30:21):
undefined

Ejaaz: And yeah, I think that's it. Thank you so much for listening. (30:24):
undefined

Josh: See you next time. (30:27):
undefined