GoodAlexander & David Choi - USDAI, the Upcoming IPO Bubble & Cheap Chinese AI

WEBVTT
00:00:00.239 --> 00:00:02.720
And boom, David.
00:00:02.959 --> 00:00:04.240
Good Alexander.
00:00:04.480 --> 00:00:06.080
Part two for David.
00:00:06.400 --> 00:00:08.880
Part four, maybe for Mr.
00:00:09.359 --> 00:00:10.080
Alexander.
00:00:10.160 --> 00:00:10.720
How are you guys?
00:00:10.800 --> 00:00:11.519
Welcome, welcome back.
00:00:11.599 --> 00:00:13.199
Alex, this incredible view there, man.
00:00:13.279 --> 00:00:15.119
That's fucking I'm getting view moged, dude.
00:00:15.199 --> 00:00:16.320
I have this new lighting setup.
00:00:16.480 --> 00:00:17.760
I'm getting view mogged right now.
00:00:17.839 --> 00:00:18.800
It's insane.
00:00:19.120 --> 00:00:25.440
I feel like it's just like you're you're like that would be like the basement in this this apartment, you know?
00:00:25.519 --> 00:00:28.480
It's sort of just a different setting in the same uh same location.
00:00:28.640 --> 00:00:29.359
Yeah, there we go.
00:00:29.440 --> 00:00:31.920
Um, dude, it's good to see see both of you here.
00:00:32.079 --> 00:00:37.920
Um, I mean, first and foremost, David, you were on what, maybe in October, November.
00:00:38.799 --> 00:00:40.399
Uh you first came on, I first met you.
00:00:40.479 --> 00:00:41.759
I thought you were doing this really interesting.
00:00:41.920 --> 00:00:45.920
The whole USDI uh AI concept was was pretty wild.
00:00:46.159 --> 00:00:51.439
And after that, a lot of smart people, Tyche, Alex Good, a lot of people were telling me about it.
00:00:51.520 --> 00:01:06.879
And then you just had TGE, you just launched TGE on chip, which is I mean, look, I got kind of excited about it because it's like first hyped crypto TGE since I don't know, Aster that didn't go down in a straight red line yet.
00:01:07.040 --> 00:01:13.200
I'd have you know it's only been like a day, two days, but did not go down in a straight red line is pretty exciting.
00:01:13.280 --> 00:01:19.439
And there were some pretty cool crypto names, good Alexander good being one of them that were excited about it.
00:01:19.519 --> 00:01:20.959
So congratulations on the launch.
00:01:21.200 --> 00:01:22.400
Do you want to start with?
00:01:22.480 --> 00:01:30.799
I think there's like a most people here have some context, but give us like a high level on what you're working on, what you just launched, and then we can get into some of the fun stuff.
00:01:31.920 --> 00:01:32.560
For sure.
00:01:32.719 --> 00:01:41.040
Um I guess for those that weren't listening last time, um, we're effectively trying to solve a capital issue that AI has.
00:01:41.200 --> 00:01:50.159
So AI, I mean, you see yourself with uh with Claude getting like kneecapped every time you try to use it because it's not enough compute, you know?
00:01:50.400 --> 00:02:06.719
Um and OpenAI seems to have caught up in terms of quality, um, and uh even though they have this myth hosting kind of in the back, it's because they kind of knew in the very beginning, like it's of course you can have better and better models, but at the end of the day, it's gonna be access to how much compute that you have.
00:02:06.879 --> 00:02:08.719
Because you'll never have enough compute.
00:02:08.960 --> 00:02:18.080
Um, and people might say it's a power issue, people might say yada yada is like quality of the LLM, but really it's just gonna be like how many black wells do you have access to.
00:02:18.319 --> 00:02:23.360
Um and that's kind of why you you're seeing this flip as of like literally this week.
00:02:23.599 --> 00:02:35.840
Um, and the problem we solve is like we're trying to figure out how to enable this abundance by having chips easily um accessible by not necessarily just open AIs of the world, but anyone who wants to buy the chip.
00:02:36.080 --> 00:02:38.479
Um it's kind of like a mortgage, but for GPUs.
00:02:38.560 --> 00:02:41.199
Um and we solve this by using stable coins.
00:02:41.280 --> 00:02:46.960
So using stable coins to finance um this wave of um CapEx build out.
00:02:47.199 --> 00:02:52.800
Um, because I think a lot of people kind of like overlook the fact that T-Bills are like the world's fastest buyers.
00:02:52.960 --> 00:02:56.719
I mean, stable coins are the fastest buyers of T-bills in the world.
00:02:56.960 --> 00:03:10.319
Uh yeah, and right now one of the biggest spends in the US economy today, literally, if you weren't if we weren't spending this money on AI capex, our GDP would be negative three, negative four percent growth.
00:03:10.479 --> 00:03:14.319
Um, it's just like is that that big of an effect on the US economy?
00:03:14.479 --> 00:03:19.759
Um that should be finance with the fastest horse, which I believe is stable coins.
00:03:19.919 --> 00:03:22.000
Um and that's why we build USD AI.
00:03:22.080 --> 00:03:24.639
It's like, how do you you have two mega trends?
00:03:24.719 --> 00:03:32.400
One is, of course, the capital formation of stable coins, and secondly, the AI capex spend in the US, um, which it's gonna be like trillion to trillion dollars.
00:03:32.479 --> 00:03:41.120
If you if you like some the last 10 years of infrastructure spend, oil, highways, whatever, one year of AI capex is more than all of that.
00:03:41.199 --> 00:03:46.080
Um so it's just like the amount the amount of money we're spending, like it's just infeasible.
00:03:46.240 --> 00:03:49.280
Like uh I used this metaphor last time with um on those calls.
00:03:49.360 --> 00:03:56.159
Like if you were to have visualize the Las Vegas sphere, the most expensive stadium ever, if you did 30 of these, right?
00:03:56.319 --> 00:03:57.599
That's one data center.
00:03:57.759 --> 00:04:00.319
Um, so just so much fucking money kind of being put.
00:04:00.639 --> 00:04:01.199
Is that true?
00:04:01.520 --> 00:04:10.080
It's yeah, because it's like I mean, one one gigawatt is like 50 billion dollars, and then that build outs like two or three.
00:04:10.240 --> 00:04:18.959
Um, so just have and you have these being built around the world, like 50 hundreds of these every single like Project Hail Mary is being like literally done.
00:04:19.279 --> 00:04:20.160
Have you seen the movie?
00:04:20.319 --> 00:04:27.279
But instead of trying to save humanity, you're trying to actually just have the best model, um, and just let an insane amount of spent.
00:04:27.360 --> 00:04:30.240
Um, it's like the most we've ever spent since like the build out of China.
00:04:30.639 --> 00:04:39.680
Um, but it's hard to see because you see like people complaining about data centers, but I'm like, I don't think people actually realize the hyper object of this.
00:04:40.000 --> 00:04:41.680
Um yeah.
00:04:42.319 --> 00:04:45.360
30 sphere build-outs for one data center is nuts.
00:04:45.600 --> 00:04:48.639
Um, Alex, by the way, why are you like moging me right now?
00:04:48.720 --> 00:04:51.360
Like what like I don't you look fresh last time.
00:04:51.600 --> 00:04:52.160
What is it?
00:04:52.560 --> 00:04:53.360
What's going on here?
00:04:53.600 --> 00:04:57.920
People were trying, you know, people kept, you know, they were commenting on my background.
00:04:58.079 --> 00:05:00.000
They're like, you gotta leave your room.
00:05:00.079 --> 00:05:08.240
And I'm like, I don't want to leave my room because you know, you don't want to see what's behind me in real life because you'll feel bad about your location.
00:05:08.480 --> 00:05:15.519
So now, now, now that we're in the mogging era, I just thought it was appropriate to just uh you know reveal reveal the state.
00:05:15.680 --> 00:05:20.240
I mean, it's gotta be the double jar or I haven't pinpointed it yet, but I'm talking to Rasmer in the background.
00:05:20.319 --> 00:05:21.439
We'll figure out what it is.
00:05:21.600 --> 00:05:25.279
But give me like a primer as well before we get into some of the details.
00:05:25.519 --> 00:05:29.839
You're I mean, you're definitely one of my high signal proxies for AI stuff.
00:05:29.920 --> 00:05:35.199
If you say something about AI, I generally at least pay attention to what you're talking about, maybe except for the Bob Along.
00:05:35.439 --> 00:05:37.120
But in general, I do.
00:05:37.519 --> 00:05:42.800
And you were pretty excited and have been excited about USD AI, chip, the whole ecosystem.
00:05:42.959 --> 00:05:48.399
Like it talk, yeah, explain that to me, like why you're I mean, why you're here and why you're excited about it.
00:05:48.720 --> 00:05:55.199
I mean, David and I have lived, uh, we're neighbors and we've we've been neighbors since 2021.
00:05:55.360 --> 00:05:55.759
Wow.
00:05:56.000 --> 00:06:16.480
And um, so we are I've been super in it, like, you know, ever since David originally was a financing platform, a blockchain-based financing platform uh for NFTs, and he built this really elaborate, you know, credit mechanism for like the worst credit in the world, which is basically guys borrowing other people's you know profile pictures on the internet.
00:06:16.639 --> 00:06:20.959
It's like like it's like it's like the guy, like the exact dude you don't want to lend to.
00:06:21.120 --> 00:06:24.160
So he built like a really good system for that.
00:06:24.639 --> 00:06:47.920
And when Nvidia started talking about this whole sovereign AI thing, um, you know, what what you might just at a very high level realize is that uh the holders of Tether, like the UAE, Malaysia, like these sort of like international jurisdictions that are mostly the stablecoin holders are also extreme investors in data centers, right?
00:06:48.000 --> 00:06:52.160
And they're extreme investors in NVIDIA graphics chips specifically.
00:06:52.399 --> 00:06:59.360
Um, and then like this is like the perfect crypto financing use case, right?
00:06:59.519 --> 00:07:02.000
Because it's like weird credit.
00:07:02.160 --> 00:07:12.079
Um, there's warehousing issues, it's a completely new category uh that that like like you can invent, you know, a lot of the for example, Canton Network is a very hot coin.
00:07:12.639 --> 00:07:15.600
Um, and it's like, okay, that's the old economy.
00:07:15.759 --> 00:07:19.439
That's like, you know, trading Coca-Cola fixed income on a blockchain.
00:07:19.680 --> 00:07:30.000
There are a lot of dudes who are trading Coca-Cola fixed income without a blockchain, and you're kind of like, well, anon, like, check out this like cool new private network, and they're like, How much can I farm?
00:07:30.240 --> 00:07:38.000
Whereas with USDAI, it's more like, okay, like nobody really knows how to finance graphics cards right now.
00:07:38.160 --> 00:07:48.480
Like, like, like actually, like before David showed up on the scene, a lot of this was done via private credit and just really like unsecured uh lending.
00:07:49.040 --> 00:07:54.399
And uh what's kind of novel or what got me excited was like, look, like these are warehouses.
00:07:54.639 --> 00:08:00.240
Uh, you know, previously, back when I worked at Palantir, I used to work in like a trade financing.
00:08:00.319 --> 00:08:11.360
Like I'd study uh basically, you know, standard chartered and receivables and credit and trade financing, which is one of the biggest forms of financing in the world.
00:08:11.600 --> 00:08:16.240
And GPUs are actually like firmly warehouse-based credit products.
00:08:16.480 --> 00:08:19.680
And so I'm like, okay, like has all the this checks all the boxes.
00:08:19.759 --> 00:08:21.040
It's a new category.
00:08:21.279 --> 00:08:28.959
It's got uh an asset backed security element, which is superior to the current market's treatment of the asset via private credit.
00:08:29.040 --> 00:08:34.240
Uh, the existing holders align with like you know, the tether holder base right now.
00:08:34.639 --> 00:08:39.120
And um, you know, I counseled David like back, I think we're in a sauna.
00:08:39.360 --> 00:08:44.080
Like he's like, I'm thinking of pivoting into AI, and I'm like, yeah, like you should absolutely do that.
00:08:44.240 --> 00:08:59.279
I mean, obviously, like I was like younger and happier, and like, you know, I was uh like things things were things were like you know, very like optimistic, but I think as we've evolved in the market, now it's just like shit's real, like everything is no longer a meme.
00:08:59.440 --> 00:09:06.639
All these things are actually happening, and yeah, that's that's how I got into uh you know chip and USD AI.
00:09:06.799 --> 00:09:36.799
Um and yeah, I'm I'm I'm I think it's like I think right now is just exciting uh because uh I think even a couple of weeks ago, like the consensus in the market would have been that Opus and Mythos and uh Open AI are just the god tier models, and that just random people buying GPUs or a lot of these sort of things that the people in the crypto space were saying uh were kind of like misguided, right?
00:09:36.879 --> 00:09:44.879
Because it's like, okay, do you really need to like buy your own GPU if like open AI is like the best thing in the world and you're just buying it via an API?
00:09:45.200 --> 00:09:51.200
Probably not, but like in the last two weeks, there's just been this non-stop, like basically barrage.
00:09:51.519 --> 00:09:52.399
What's the shift there?
00:09:52.559 --> 00:09:54.159
Like, what caused the shift?
00:09:54.559 --> 00:10:01.519
It it was um, you know, I think a lot of people for a while were saying that China was just distilling Opus.
00:10:01.919 --> 00:10:02.159
Yeah, yeah.
00:10:02.320 --> 00:10:02.559
Right.
00:10:02.720 --> 00:10:11.039
So so the narrative was that these Chinese models are just Opus distills and they're stealing our shit.
00:10:11.440 --> 00:10:22.960
And um, you know, therefore it's not worth paying attention to because actually what's going to happen is that Mythos is going to uh be a thing and they're gonna stop, they're gonna shut down the distills.
00:10:23.039 --> 00:10:28.559
So Anthropic put out these these releases saying, like, no more distills, we're shutting this down.
00:10:28.799 --> 00:10:37.519
And then there are all these releases that are coming out that are better than Opus in a lot of different ways in terms of efficiency, in terms of like utilization.
00:10:37.840 --> 00:10:47.440
Like the fact that you can run like one of these models that has Opus like performance or sonnet like performance on a$10,000 graphics chip is a complete games changer.
00:10:47.519 --> 00:10:50.720
Like they're there guys who are doing this on like 3090s now, right?
00:10:50.799 --> 00:10:53.679
So it's like damn, um I got a nicer card than that.
00:10:53.919 --> 00:10:57.279
Yeah, and so it's like the local, and these are real, you know.
00:10:57.360 --> 00:11:09.919
I I didn't even believe it because I like, you know, actually like with Chinese stuff, like you know, they do have a tendency to overpromote um these releases and game benchmarks, but you know, I tested this stuff out because I didn't even believe it.
00:11:10.159 --> 00:11:12.720
And yeah, these models are amazing.
00:11:12.879 --> 00:11:30.240
And so I think that's like I actually think the market still largely hasn't um realized what this means because uh they're still people are still very much in this kind of like, we literally aren't going to be able to afford Opus.
00:11:30.720 --> 00:11:35.919
Like, you know, my Opus bill, I was saying, like, you know, it's$80 per hour per terminal.
00:11:36.159 --> 00:11:46.399
If you're actually using Claude without uh with a proper zero data uh protection, like a ZDR, it's called zero data retention.
00:11:46.480 --> 00:11:51.840
So your IP actually doesn't get transferred to anthropic or you know, there's like strong protections.
00:11:52.080 --> 00:11:59.679
If you're using that via an API without the Claude Max plan or whatever, if you're just paying per token, it's like$80 per hour per terminal.
00:11:59.840 --> 00:12:03.440
And you see a lot of guys, I run two terminals, that's$160 an hour.
00:12:03.679 --> 00:12:04.480
Jeez, right?
00:12:04.559 --> 00:12:06.559
And that's like extremely expensive.
00:12:06.720 --> 00:12:15.600
And to put that in perspective, like this, you know, this uh machine, like the new models onto the Quen 3.6, is$2 an hour.
00:12:15.840 --> 00:12:16.159
Right.
00:12:16.240 --> 00:12:27.279
So you're looking at something like a 40 to 80 X Delta, and there a lot of these models are state-of-the-art, uh, SOTA.
00:12:28.000 --> 00:12:34.240
And it's just like, yeah, I'm like, oh shit, like and no GPT 5.5, I haven't really used it a lot yet.
00:12:34.320 --> 00:12:34.559
Yeah.
00:12:34.720 --> 00:12:39.679
Um, but it's just, yeah, I mean, I think that's that's what's new in the market.
00:12:39.759 --> 00:12:47.759
I think everyone, soft bank and a lot of these anthropic, you know, secondary market, a lot of this stuff has already created, like, had this crazy run.
00:12:48.000 --> 00:12:57.360
And I just don't think actually the open source uh stuff is really like priced in uh or appreciated by a lot of people in the market.
00:12:57.440 --> 00:13:14.320
And it's really good for crypto AI because I mean, we are not capable of serving closed source models like at uh structurally, and especially with the mythos launch where they're like gating it to like imagine trying to buy mythos on Venice AI, for example.
00:13:14.559 --> 00:13:16.080
It just doesn't make sense, right?
00:13:16.240 --> 00:13:19.679
You have like bioauthenticated closed source models.
00:13:19.919 --> 00:13:25.919
Crypto AI would really be in a bad spot if that were like you know the future.
00:13:26.080 --> 00:13:28.720
And so, so yeah, I mean, I think this is great news.
00:13:28.879 --> 00:13:36.799
Um, I'm I'm like super excited for the first time in a while because I actually I thought like open AI is just one and like, yeah, that would suck.
00:13:37.120 --> 00:13:37.919
That would suck.
00:13:38.000 --> 00:13:40.159
Uh David, okay, you go ahead.
00:13:40.480 --> 00:13:52.000
Yeah, I just want to add, like, that's kind of like why I thought having both of us on would be kind of nice because we kind of like, of course, we both come from Puerto Rico and seem despair, uh, just like Jeff from Hyperliquid.
00:13:52.080 --> 00:13:56.080
I just it's a very miserable place, but I love it at the same time, to some degree.
00:13:56.240 --> 00:13:58.720
Um, who's terrible, but it's lovable.
00:13:58.879 --> 00:14:02.000
Uh, but it's kind of like this new wave that we kind of want to push.
00:14:02.080 --> 00:14:09.360
It kind of like the exciting times of like early DeFi, where um, I think there's just such a a rut um in the ecosystem right now.
00:14:09.440 --> 00:14:13.039
People are like surprised that you know our project did like decently well in the market.
00:14:13.200 --> 00:14:27.440
And I'm like, it's because there's kind of been a sepsis of like just horrible projects, obviously, um, post-Me coin era and all this like um venture-backed vaporware that just their goal is to just extract money from public launches and just pinch it on perps.
00:14:27.840 --> 00:14:30.720
Um, but there is like there still is a future.
00:14:30.879 --> 00:14:33.840
And I think a lot of people kind of lost that glimmer of hope.
00:14:34.000 --> 00:14:36.399
You've seen so many OGs leave in the space.
00:14:36.559 --> 00:14:42.000
Uh, but you know, crypto AI is like this definitely is kind of what we're seeing as uh interesting use cases.
00:14:42.080 --> 00:14:43.679
Would you, you know, you've seen them with Venice.
00:14:43.759 --> 00:14:53.919
People are starting to see with us and soon post-fiat, um, where there are actual users kind of coming about and organizing to this new wave of uh well of hope.
00:14:54.080 --> 00:15:08.080
Um and yeah, it's just it's just like what we what what can we do to kind of show to the world like the things that we're doing actually has an impact and not necessarily just in the sector that we're in, but out toward the frontier, um, toward actual what we see in AI.
00:15:08.399 --> 00:15:10.960
How big is the market to finance the GPUs?
00:15:12.080 --> 00:15:13.679
Um, I'll give you an example.
00:15:13.759 --> 00:15:18.159
Um, NVIDIA has$500 billion of sales in the next five quarters.
00:15:18.399 --> 00:15:20.720
99% of that is financed.
00:15:20.960 --> 00:15:24.799
Um, so out of the 500 billion, it's about 20, 30 percent.
00:15:25.360 --> 00:15:26.879
Uh sorry, next five quarters.
00:15:27.200 --> 00:15:28.480
Next five quarters, okay.
00:15:28.720 --> 00:15:29.120
Yeah.
00:15:29.279 --> 00:15:36.080
And about 20 to 30 percent of that is is there's those are emerging neo clouds, um, as in their smaller participants.
00:15:36.240 --> 00:15:39.039
They're not the hyperscalers like the Amazons of the world.
00:15:39.200 --> 00:15:43.279
Um, they're core we was small at one point, like one of their loans was like 30 mil.
00:15:43.679 --> 00:15:46.639
Um, so it's pretty much all these participants.
00:15:46.799 --> 00:15:49.519
Um, and you can you really can't finance anything under 200 mil.
00:15:49.679 --> 00:16:01.200
We were chatting with this pretty like emerging data center, they just had a five-page pitch by Guggenheimer and or Jeffries uh explain why they can't get financing for the 200 mil GPU.
00:16:01.360 --> 00:16:02.399
And they they had a call with us.
00:16:02.559 --> 00:16:04.159
And why can't they get financing for it?
00:16:05.519 --> 00:16:09.600
Because uh this is best explainatory metaphor.
00:16:09.759 --> 00:16:16.639
Um when you get a loan for your house, the bank that you get a loan from is not the one holding the loan.
00:16:16.879 --> 00:16:19.279
They don't want you know when they get a mortgage, it's crazy, right?
00:16:19.360 --> 00:16:23.200
You get 10% down and you get a 90% loan on your house.
00:16:23.440 --> 00:16:29.360
Um because they batch it up, they package it up with 10,000 other loans and then they make it tradable.
00:16:29.600 --> 00:16:31.759
Yeah, um, much like wrap stake ETH, right?
00:16:32.080 --> 00:16:32.799
With e-staking.
00:16:33.039 --> 00:16:34.639
E-staking your ETA stock.
00:16:34.799 --> 00:16:38.879
But if they have a derivative, then let's stick with the mortgage market, David.
00:16:39.039 --> 00:16:39.360
Maybe not.
00:16:40.159 --> 00:16:42.080
Maybe that's not the one that you want to go with.
00:16:44.240 --> 00:16:44.639
Yeah.
00:16:44.879 --> 00:16:47.120
Uh but I mean the big short.
00:16:50.639 --> 00:16:57.200
But the main idea is that if you can you can you can make that really liquid um for an illiquid asset.
00:16:57.360 --> 00:16:58.960
So houses are not liquid, all right.
00:16:59.120 --> 00:17:02.639
And like how many how many houses sales are there versus the value of the houses?
00:17:02.799 --> 00:17:03.919
Like very, very little.
00:17:04.160 --> 00:17:09.359
Um so that exists for airplanes, does it exist for houses, for cars?
00:17:09.519 --> 00:17:13.119
Some kind of you're gonna you can get a loan against a Honda Civic, right?
00:17:13.279 --> 00:17:16.640
Um, but you can't get a loan for a single GPU.
00:17:16.799 --> 00:17:18.000
And that's a that's a problem.
00:17:18.079 --> 00:17:22.799
If you can't get a loan for a single GPU, um, you can't get a loan against a larger cluster.
00:17:22.960 --> 00:17:24.319
Um now, why is this?
00:17:24.400 --> 00:17:31.519
Well, it's because you don't have this thing that I just described, which is a mortgage-backed security, something that like batches up all the loans and makes it tradable.
00:17:32.480 --> 00:17:42.960
Why do you have this for airplanes, you have this for cars and and houses, but you don't have it for the most important commodity in the world today, um, which is the basis of national sovereignty, right?
00:17:43.200 --> 00:17:43.759
AI.
00:17:43.920 --> 00:17:45.279
Um, why don't you have this?
00:17:45.359 --> 00:17:51.839
It's because it takes it that the time it takes to issue that instrument would be like two years, right?
00:17:52.079 --> 00:17:56.880
You gotta give 10,000 loans, cut it out tranche, go on a road show, and then trade it, sell it.
00:17:57.039 --> 00:17:59.440
Um I expend the full sweep.
00:17:59.599 --> 00:18:07.839
Um and uh yeah, that just takes too long because by if it's two years, you're past the hoppers, you're past the Blackwells, the B300s, right?
00:18:08.000 --> 00:18:17.119
B200s, you're past the Rubens, the Vera Rubens, you're now at like the Feynman, which is the ones they they haven't even announced yet, because of how fast tech improves.
00:18:17.279 --> 00:18:23.599
So because you don't have the secondary debt market, which is the basis of how you get that easy, simple mortgage.
00:18:23.680 --> 00:18:27.119
Um, this is like US housing in the 1910s.
00:18:27.359 --> 00:18:33.920
You didn't have um mortgages, you did a five-year loan and the bank could call that loan any time.
00:18:35.359 --> 00:18:36.160
That's how it worked.
00:18:36.240 --> 00:18:39.279
Um, you get 50% down um as opposed to 10%.
00:18:40.000 --> 00:18:43.200
Um this too is like the Bitcoin pizza dynamic, right?
00:18:43.279 --> 00:18:55.119
Like imagine if your cursor and you're raising at a billion dollar valuation to buy some GPUs, you buy$100 million of GPUs, you sell at a billion dollar val, you get acquired for$60 billion.
00:18:55.359 --> 00:19:04.720
Like you just you just basically paid$6 billion of expected value for$100 million of GPUs, right?
00:19:04.880 --> 00:19:15.680
So it's like like if you actually are bullish on your company and you're selling like right now, the venture market is very hot, but like the reality is there's a reason it's hot, and a lot of these companies are going up a lot.
00:19:15.839 --> 00:19:21.759
A lot of these like Korean stocks are up massive, like you know, just like like these equities are crazy.
00:19:22.079 --> 00:19:33.839
And the idea if you're if you're traded that you know, people are saying, you know, SK Heinrich and some of these other, you know, Samsung, a lot of them are going to earn their entire market cap in like the next three to five years.
00:19:34.000 --> 00:19:40.640
Um, and the earnings of a lot of these companies, unlike the 1999 tech bubble, are quite high, a lot of them, right?
00:19:40.799 --> 00:19:46.079
And so it's like, okay, do you really want to uh sell equity?
00:19:46.160 --> 00:19:50.559
At least you want to have the option to finance the GPUs raw, right?
00:19:50.640 --> 00:20:06.079
Like you don't want to necessarily sell equity um or other types of credit when you have this like alternative uh financing mechanism, which is sort of unfortunate because like for the most part, like people actually do have to sell equity uh to do these.
00:20:06.160 --> 00:20:15.200
And so it's it's sort of like a weird category because a lot of the companies like that would be going into the traditional debt market, like the the underwriters are like, who are you?
00:20:15.279 --> 00:20:16.400
Like what is cursor?
00:20:17.680 --> 00:20:22.160
You know, and so I think it's like a it's like kind of like a cost of capital thing, right?
00:20:22.240 --> 00:20:23.759
A very basic finance 101.
00:20:23.839 --> 00:20:30.000
You're bullish on your own company, you don't want to issue unsecured credit, you know that you have to make an asset purchase.
00:20:30.240 --> 00:20:34.640
Why wouldn't you do asset uh asset backed financing like every other industry?
00:20:34.799 --> 00:20:35.920
Like, like this is common.
00:20:36.000 --> 00:21:00.559
I mean, this is like a weird detour, but like, you know, I used to work in the e-commerce uh space, and a lot of a lot of like just really 101 like like lawn mowing or like like garage work or like stuff where you know that there's gonna be a return, all of that is asset financed because of not only like because it's efficient, but because of taxes, right?
00:21:00.720 --> 00:21:11.440
So I mean, like this is something that people I I think also don't understand is that these things are like fully depreciatable assets with huge tax implications.
00:21:11.519 --> 00:21:22.559
So so because of the new tax acts, I mean, David, I don't know if you want to talk about this, but it's it's a material part of the yield component of this like credit instrument, is some of the recent tax bill changes.
00:21:22.880 --> 00:21:29.680
Yeah, it's kind of why you see Meta have almost a perfect pairing of like say 20 billion in revenue and they have 20 bill of capex.
00:21:30.160 --> 00:21:34.559
It's kind of on purpose, by the way, because they can uh they can offset all of that.
00:21:34.960 --> 00:21:35.759
Offset it completely.
00:21:36.400 --> 00:21:38.880
Yeah, uh, because of funds depreciation.
00:21:39.039 --> 00:21:42.480
Um, so it's uh the these are capex assets.
00:21:42.640 --> 00:21:45.920
Um, and we're really trying to encourage the development of stuff.
00:21:46.000 --> 00:21:54.960
And I think us both being in PR um kind of uh reinforces that like you can't it's definitely change human behavior with that kind of incentive structure.
00:21:55.119 --> 00:22:07.359
Um but for those that don't understand, like um when you when you buy, I mean Yeah.
00:22:07.759 --> 00:22:12.240
But pretty much whenever you buy a hard asset, you can write it entirely off.
00:22:13.200 --> 00:22:23.680
And like kind of have these tax tax deferrals pushing into the future because you're effectively having a net operating loss which offsets the revenue that you generated.
00:22:24.240 --> 00:22:27.359
Effectively zeroing out your taxes for that year.
00:22:27.680 --> 00:22:30.480
Can you sorry, finish, finish?
00:22:30.960 --> 00:22:31.200
Yeah.
00:22:31.599 --> 00:22:45.440
I was just going to say too, like it's um we we we're like pushing whatever we can to like encourage the development of uh of a of AI CapEx because it is such a massive like black hole of money.
00:22:45.599 --> 00:22:47.599
There's just so much money that's being put into it.
00:22:47.759 --> 00:22:51.119
Um and we should definitely like put ourselves at the center of this.
00:22:52.160 --> 00:22:57.839
Can you explain, David, why crypto is necessary to solve this?
00:22:58.400 --> 00:23:00.240
And why is crypto uniquely positioned here?
00:23:01.200 --> 00:23:07.680
Um crypto is unique because DeFi specifically is the sector that I tackle.
00:23:07.759 --> 00:23:09.759
Um it is a fastest horse.
00:23:10.079 --> 00:23:13.599
It forms capital in a way that is hard to recreate even at RadFi.
00:23:13.759 --> 00:23:17.920
Um like you can just spawn markets out of nothing, um, which is difficult to do.
00:23:18.000 --> 00:23:20.640
Like you need to have a lot of infrastructure, a lot of participants.
00:23:20.799 --> 00:23:22.160
Um, I'll give you a good example.
00:23:22.240 --> 00:23:26.160
A core weave bond has traded like 50 mil since like August.
00:23:26.319 --> 00:23:28.160
It's like a$1.5 billion tranche.
00:23:28.319 --> 00:23:29.759
It's like a tradable debt instrument.
00:23:30.160 --> 00:23:48.400
Um we ourselves was much smaller, but in the last six to eight months, we've traded about 15 billion secondary market volume because the moment you create us um USDI or SUSDI, you can then put up, you can put on either swap handle to strip out the yield, and then you uh trading the implied notional value.
00:23:48.480 --> 00:23:53.359
You're doing looping on like uh fluid or these other things, and you end up creating this massive volume.
00:23:53.519 --> 00:23:59.279
Um, so what what the what the tech really develops is um is one perpetuality.
00:23:59.440 --> 00:24:02.880
Of course, you have um um hype, which is the features perps.
00:24:02.960 --> 00:24:04.400
We're really like a debt perp.
00:24:04.559 --> 00:24:12.400
Um we we can actually instantly price and compound things, um, which is why whenever there's a new loan, it gets added the moment it's originated.
00:24:12.480 --> 00:24:14.880
You don't have to like wait for 10,000 loans to be batched up.
00:24:15.039 --> 00:24:21.519
The moment a loan's created is immediately um priced on nav and given to our debt instrument or to just SUSD AI.
00:24:21.839 --> 00:24:30.480
Um so it's just like a really, really good financial tech um product, which is the way that it's the way that finance should be run.
00:24:30.559 --> 00:24:35.759
Um, because it's uh um it's a it makes things extremely fluid and and tradable.
00:24:35.920 --> 00:24:38.480
You don't need to have just um white-collar guys trading it.
00:24:38.559 --> 00:24:39.519
Anybody can do it.
00:24:39.680 --> 00:24:45.839
Um so yeah, it's just uh is TradFi struggling with with GPU funding, or is that one of the hottest?
00:24:46.160 --> 00:24:51.519
It is TradFi is struggling with it on the lower end, like the smaller contracts, like you said.
00:24:51.759 --> 00:24:53.599
It it doesn't even exist in the lower end.
00:24:53.680 --> 00:24:55.200
It doesn't exist like you can't get financing.
00:24:55.359 --> 00:25:00.960
Yeah, if it's under like if it's under 200 mil, like you might get it if you work with like a private credit fund.
00:25:01.119 --> 00:25:03.359
By the way, every private credit fund's in retreat.
00:25:03.519 --> 00:25:05.680
Um Oracle is fully tapped out.
00:25:05.759 --> 00:25:09.440
If you see Oracle all the time, like yeah, they they're like max levered out.
00:25:09.519 --> 00:25:17.599
Um, they max out their real like their credit card, um their term loans, um, because you know you have a revenue and you can borrow like six X six X against it.
00:25:17.920 --> 00:25:37.519
Um so you you're literally seeing the very um um edges of pretty much financial vehicle to raise money to build out AI capex reaching its limits because um it's like you're playing hot potato where you can't, where you're trying to obviously make sure you don't your hand doesn't get burned, but people are playing solo, right?
00:25:37.599 --> 00:25:38.880
They're literally just holding the hot potato.
00:25:39.359 --> 00:25:45.200
So do you just like get stuck with the suboptimal financing then because you're the only one that will like underwrite these GPUs?
00:25:45.839 --> 00:25:52.720
Oftentimes you have to do a convertible, um, uh, meaning that you get a loan, but then they can turn it into equity and you get diluted.
00:25:52.799 --> 00:26:05.599
Um, a good example is like if you're a seed investor in like Corweave or Crusoe, you were downloaded like almost 90% like 10 times more like equity that has to be issued because that's the only way to finance them.
00:26:05.680 --> 00:26:07.839
Um, because you don't have any other option.
00:26:08.000 --> 00:26:17.839
Um, so having like non-dilutive equity is like um uh non-dilutive financing is pretty pretty interesting, but generally it the the solution doesn't even exist.
00:26:18.079 --> 00:26:23.200
It's why we exist as a company, like it just doesn't even exist as an offering in traditional ecosystems.
00:26:23.359 --> 00:26:31.440
Um, and it's a solution that people have described as as a monopoly because we're like nobody else even comes close to offering this stuff, um, which is smaller than the monopoly.
00:26:32.079 --> 00:26:32.559
Yeah.
00:26:33.119 --> 00:26:35.119
Um, not gonna say who, but it was a good one.
00:26:35.279 --> 00:26:38.319
Before um I'll ask you one in a minute, Alex, but I'll go back really quick.
00:26:38.400 --> 00:26:44.240
It's like um, what is the win condition for the company at large?
00:26:44.319 --> 00:26:45.920
And then like, what is the purpose of the ship?
00:26:46.079 --> 00:26:50.240
Like, how how do you get 10x, 100x, not price, but just like overall size from here?
00:26:50.400 --> 00:26:51.599
Like, how does it get massive?
00:26:54.960 --> 00:26:55.920
To you, David.
00:26:56.720 --> 00:27:00.240
Um well, there's two angles for us.
00:27:00.799 --> 00:27:07.119
There's chapter one, um, which is next 12 months, which is we do like a like a billion dollars of loans.
00:27:07.279 --> 00:27:08.640
Um, and that's our ultimate goal.
00:27:08.960 --> 00:27:16.000
Like, we want to just do enough loans where people just start referring to us as like the the reference rate or what the interest rate should be.
00:27:16.240 --> 00:27:17.599
Because that's kind of what we're trying to do.
00:27:17.680 --> 00:27:25.119
Is like if you do enough loans globally, um, what happens that if you do enough of these loans, you kind of discover this interest rate.
00:27:25.279 --> 00:27:28.880
Um, and when I say interest rate, I mean that this is the cost of capital.
00:27:29.039 --> 00:27:31.279
This is how fast you can grow AI.
00:27:31.359 --> 00:27:33.039
Um, that is what you're trying to discover.
00:27:33.279 --> 00:27:37.279
It's really hard to know what the interest rate is of artificial intelligence.
00:27:37.359 --> 00:27:43.680
Um, it is literally the cost at which things can grow, like how much money you're willing to spend in for a return.
00:27:43.920 --> 00:27:54.079
Um, once you discover the interest rate, which is us doing you know a billion dollars of loans, how do you how do you then weaponize this to kind of penetrate to the rest of AI?
00:27:54.319 --> 00:27:56.720
Um and I'll I'll say this too.
00:27:56.880 --> 00:28:03.039
We are starting to see that every AI company we talk to will eventually own GPUs.
00:28:03.200 --> 00:28:15.200
Um we we started talking with these rental companies like a year ago, and then you know, the GPU rental platforms, they're starting to want to buy chips because then they rent it out to token factories, these guys who can like really amplify like the base tense of the world.
00:28:15.279 --> 00:28:18.160
Um, they amplify how many tokens you can generate per GPU.
00:28:18.319 --> 00:28:26.079
Now they want to buy chips, and then LLMs are already buying chips, and then the apps are gonna buy chips because it's just really expensive to rent it out.
00:28:26.240 --> 00:28:37.680
Um and um what we're trying to do next after we discover this interest rate and like offer financing for smaller participants and like applications or LLMs is how do we subsidize it?
00:28:38.000 --> 00:28:40.640
How do we lower that interest rate more than the market?
00:28:40.799 --> 00:28:42.720
Um, like a central bank in a sense, right?
00:28:42.799 --> 00:28:45.759
Like lower rates, things grow, people just spend more money.
00:28:46.000 --> 00:28:51.119
Um the the number the number one expense in all of AI is compute.
00:28:51.440 --> 00:28:52.559
I just want to clarify that.
00:28:52.880 --> 00:28:59.519
Like 80% of open AI's costs, salaries are very high, but compute's a way more expensive than obviously the salary part.
00:28:59.759 --> 00:29:05.680
Um this is expensive, of course, because NVIDIA chips are expensive, but also because all of it's financed.
00:29:05.920 --> 00:29:13.200
500 billion um of sales, like I said, it's actually 600 to 700 billion because of interest rates because it's expensive.
00:29:13.279 --> 00:29:14.319
Um, much like a house.
00:29:14.400 --> 00:29:17.359
You're paying$2 million for a million dollar house across 30 years.
00:29:17.519 --> 00:29:19.759
Um, in this case, it's just really high interest rates.
00:29:19.839 --> 00:29:21.839
Um, and it's uh everything's debt financed.
00:29:22.160 --> 00:29:29.519
So if you can lower this expense, which is very simple because you're just paying Wall Street, Wall Street's getting it's the second largest business in AI call it.
00:29:29.759 --> 00:29:35.359
It's open AI, Wall Street, and I'm sorry, it's uh Nvidia, Wall Street, then open AI.
00:29:35.680 --> 00:29:40.960
That's 20 billion revenue for versus 500, and then you have 150 right in between.
00:29:41.119 --> 00:29:46.160
Um how do you lower this second largest expense in the AI sector?
00:29:46.400 --> 00:29:50.319
Um we think it's by settling in a stable coin.
00:29:50.799 --> 00:29:52.799
I described uh a dead instrument, right?
00:29:52.880 --> 00:29:55.599
As USD AI, yeah, we're giving out loans against chips.
00:29:55.759 --> 00:29:59.680
But what if people started paying in a stable coin that I created?
00:30:00.000 --> 00:30:02.640
Which is this is what USD AI is, not the state version.
00:30:02.799 --> 00:30:10.960
Um, if you can start paying in my currency, I can subsidize interest rate because I earn 4%, 3% every dollar that's minted.
00:30:11.039 --> 00:30:13.440
Much like Circle Anther, they just pocket that, right?
00:30:13.599 --> 00:30:14.640
They just keep that money.
00:30:14.960 --> 00:30:22.720
Um I'm saying, why don't you have the entire AI sector settle in a stable coin by subsidizing their number one cost that they can control, which is the debt?
00:30:22.880 --> 00:30:27.279
Um so do you somehow um make money?
00:30:27.359 --> 00:30:30.480
I mean, there's we still earn um an origination fee on the on the debt.
00:30:30.559 --> 00:30:32.799
Um we also make a net interest margin.
00:30:32.960 --> 00:30:37.680
Um, but you know, eventually in the in the longer run, we're not giving 100% subsidies.
00:30:37.920 --> 00:30:42.400
Um much it's like the the growth pains of any um stablecoin at the beginning.
00:30:42.480 --> 00:30:44.720
You're always cutting deals, but eventually you just earn the yield.
00:30:44.960 --> 00:30:51.599
Um but if you can subsidize the interest rate of AI, which again is how fast AI can grow, the cost of capital.
00:30:51.839 --> 00:30:58.880
Um I think this is it it is a second line item in AI, but it's also the second largest business.
00:30:59.599 --> 00:31:03.519
And what is the can you talk real quick about the purpose of the chip token?
00:31:04.400 --> 00:31:06.880
Um that is kind of like the GP position in a sense.
00:31:07.039 --> 00:31:09.039
It's just like um this is the best metaphor I can give.
00:31:09.119 --> 00:31:17.599
Um we haven't really detailed um we will over the next few weeks, um, but it controls the the the fees that are related to the protocol and how it's routed.
00:31:17.680 --> 00:31:21.839
Um, it also determines what the interest rate should be, which I think is very powerful.
00:31:22.000 --> 00:31:27.519
Um like you don't want to know what the rates are for small deals, large deals, better off takes, um, worse off takes.
00:31:27.920 --> 00:31:39.440
Um, and there will be some prior prioritization um structures as well, where redemptions, I mean, that's usually where best token utilities are when it relates to prior prioritization, um, like transaction fees.
00:31:40.079 --> 00:31:49.440
Um, there's there's a number of use cases that governs, of course, but also has uh uh decision power and how the fees are routed.
00:31:49.519 --> 00:31:52.559
Um but that's up to the users, not necessarily just us.
00:31:53.279 --> 00:31:54.079
Um yeah.
00:31:54.400 --> 00:32:09.519
On uh Alex, on the crypto topic, I feel like broadly you are I don't know if this is surprising, but you're one of the more optimistic, forward-looking crypto specific people on my TL recently.
00:32:11.440 --> 00:32:18.079
Yeah, I mean actually, um I think Tyche got me into the thesis early on, you know.
00:32:18.240 --> 00:32:27.279
I I like I kept seeing these like Michael Saylor yoga instructor videos, and I'm like, this is like, you know, I I I think I pinged Tyche.
00:32:27.440 --> 00:32:31.279
I'm like, this is like this is just like late stage made off.
00:32:31.519 --> 00:32:33.680
Like it's just so bad, it's like so incredible.
00:32:33.759 --> 00:32:35.440
He's like, he's like, no, you don't understand.
00:32:35.519 --> 00:32:36.880
And I'm like, what do you mean?
00:32:37.039 --> 00:32:44.640
He's like, you know, you know, STRC can't go bankrupt because it's like a really smart convertible prep stack, which is based on mnav.
00:32:44.880 --> 00:32:45.759
And I'm like, uh, okay.
00:32:45.839 --> 00:33:06.319
And so I like looked into it, and Sailor has like cooked this inflow machine with STRC, and I know everyone like like I got excited because I like I know everyone's wrecked, everyone is like giving up hope, but Sailor can't give up hope maybe because he owns too much Bitcoin, or you know, he's an older man without a wife.
00:33:06.400 --> 00:33:13.759
Like, you know, who knows what the reason exactly is, like, but he has developed a very, very effective yield inflow mechanism.
00:33:13.920 --> 00:33:20.880
Like, I mean, actually, like there's like a funny economist chart, the silent generation is like substantially wealthier than millennials, right?
00:33:20.960 --> 00:33:34.880
Like, it's like it's pretty bleak, like like you know, millennials in Gen Z are pretty poor relative to older people, and so if you if you really want to get inflows into an asset class, you kind of need retirees to bid.
00:33:35.119 --> 00:33:44.720
Um and uh actually to be honest, like you know, not to derail on chip, but like you know, upbit volumes are very high.
00:33:44.880 --> 00:33:47.200
This is something I've clacked with XRP.
00:33:47.279 --> 00:33:55.599
A lot of the buyers are older, um, like older Korean women specifically are you know, it's like in women, yeah.
00:33:55.680 --> 00:33:57.759
It's a big retail market, the Hajima.
00:33:58.079 --> 00:34:01.759
Um, yeah, so so it's like it's very different from US retail.
00:34:01.839 --> 00:34:09.840
Uh, Japan and Korea both have a more uh gender neutral um like if you look at crypto Twitter demographics, it's 93% male.
00:34:10.079 --> 00:34:15.039
If you look at Japan and Korea in terms of the retail composition, it's closer to 50-50.
00:34:15.199 --> 00:34:22.639
And on a capital basis, on a trading basis, you might be seeing much even higher female skew because of cultural dynamics.
00:34:22.719 --> 00:34:32.880
Uh, Japan for the longest time and Korea have had very poorly performing stock markets over very long periods, and it led to Forex trading becoming popular.
00:34:33.039 --> 00:34:36.239
Both of them have famously bad work-life balances.
00:34:36.480 --> 00:34:42.800
Um, and it's led to women adopting financial speculation in a way that hasn't been true in the United States.
00:34:42.960 --> 00:34:50.480
And so, like, this is these demographic points are important because with Michael Saylor, the question was always like, who's the incremental buyer?
00:34:50.719 --> 00:34:52.960
Like, all the meme coin guys are cooked.
00:34:53.199 --> 00:35:00.559
Like, uh, no, none of the crypto AI protocols, at least as a you know, Betenser, you know, I've never been excited about it.
00:35:00.639 --> 00:35:04.000
I I was like, okay, like the AI buyers are not here.
00:35:04.159 --> 00:35:06.800
The Koreans are bidding AI stocks in Korea.
00:35:06.960 --> 00:35:10.079
So the question was always like, who's gonna buy this thing?
00:35:10.400 --> 00:35:17.519
And I mean, the stretch catalyst was big enough where you're like, all right, this is getting multiple billions of dollars of inflows.
00:35:17.599 --> 00:35:21.440
If you deploy a billion dollars into current tape, it's gonna move the market.
00:35:21.920 --> 00:35:29.280
Um, and furthermore, that you know, if you've seen micro strategies, mnav expand, which creates a flywheel.
00:35:29.440 --> 00:35:31.920
Um and this is just mechanical, right?
00:35:32.079 --> 00:35:37.519
Like a lot of the like the the recent bounce has been, you know, ETH BTC is like near the lows.
00:35:37.599 --> 00:35:41.039
This is a mechanical BTC driven pump.
00:35:41.840 --> 00:35:44.960
And um that's like one element, you know.
00:35:45.119 --> 00:35:59.760
The the other element is that uh a lot of the reason I think there was like a head fake conceptually, where everyone like bet on the institutions, like they bet on Trump, they bet on the banks, yeah, and it just was a rug, right?
00:35:59.920 --> 00:36:03.039
World Liberty Financial, like it kind of was a rug.
00:36:03.280 --> 00:36:20.400
Um and uh I think the the good news is that all of the reasons why crypto existed prior to that, like historically, if you study the crypto asset class and you're like, what is the driver of the price of Bitcoin or crypto market cap over time?
00:36:20.559 --> 00:36:23.199
It's mostly people moving money out of the system.
00:36:23.360 --> 00:36:26.559
Uh it's like Chinese whales mining Bitcoin.
00:36:26.719 --> 00:36:35.440
Um, you know, I think one of the things that informed like uh the lot over the last year, you've seen basically every exchange come after Monero.
00:36:35.920 --> 00:36:38.159
Uh the Europeans came after Monero.
00:36:38.239 --> 00:36:45.039
I mean, Spagni's done prison time and you know, Zcash has worked, Monero has worked, despite everyone coming after them.
00:36:45.360 --> 00:36:47.679
And now Zcash is on Robinhood today.
00:36:48.159 --> 00:36:48.639
Exactly.
00:36:48.800 --> 00:36:58.639
So so it's like you try to kill these privacy coins and you can't, which is exactly what you want to see for like the forward outlook of the asset class.
00:36:58.800 --> 00:37:11.760
Like, if you can't kill these things, if you try to kill them and then you fail, um, you have a picture that this is gonna last and be able to work in a regime which is uh not friendly to capital.
00:37:11.920 --> 00:37:12.719
And what does that mean?
00:37:12.800 --> 00:37:14.079
It means the Democrats winning.
00:37:14.159 --> 00:37:22.880
It means uh CBTCs, uh capital controls, wealth taxation, uh unfriendly policies towards capitalists.
00:37:22.960 --> 00:37:27.280
And I think like from here, it's like, okay, like what is gonna cause us to go incrementally?
00:37:27.440 --> 00:37:36.320
Right now, everyone is convinced the SP is like a moon machine, and uh that you know, like you don't need crypto because you have equities.
00:37:36.480 --> 00:37:47.280
Like in the near term, that's great because it means Sailor is gonna keep issuing STRC, selling 11% yields to 75-year-olds and getting billions of dollars to just max bid BTC.
00:37:47.440 --> 00:37:53.840
So, like in the near term, we're gonna be fine because Sailor is literally like just like an ATM that just props the market.
00:37:54.079 --> 00:38:03.440
And then in the long term, I also think it's gonna be fine because we have the privacy technology, and then we have the outlook of the G10 turning into China, right?
00:38:03.599 --> 00:38:11.199
So it's like, you know, you're gonna have extreme because of the fiscal situation, uh, that's not getting better anywhere in the world.
00:38:11.280 --> 00:38:19.199
And Iran is like making it substantially worse uh uh across Europe, across Japan, energy subsidies.
00:38:19.440 --> 00:38:25.119
Uh globally, the fiscal situation was quite bad before, and it's gone from bad to much worse.
00:38:25.280 --> 00:38:29.599
The way that's gonna be paid for is ultimately via taxation and money printing.
00:38:29.920 --> 00:38:34.559
And those are those are things which drive capital out of systems.
00:38:34.719 --> 00:38:37.360
And so I'm like, okay, like where's the money gonna go?
00:38:37.440 --> 00:38:40.400
It's gonna go into gold, it's been going into oil.
00:38:40.559 --> 00:38:45.760
Um, but you know, ultimately you can't easily leave a country uh with gold or oil.
00:38:45.920 --> 00:38:47.760
You can't get on a plane with a bag of gold.
00:38:47.840 --> 00:38:48.719
We saw this in Dubai.
00:38:48.960 --> 00:38:55.599
Saw this in Dubai, and there's gonna be a bid, uh, in my opinion, from the ultra-rich to move money out of the system.
00:38:55.760 --> 00:38:58.320
And that bid is gonna go into crypto assets.
00:38:58.480 --> 00:39:01.440
And so it's like that's sort of my my thesis.
00:39:01.519 --> 00:39:10.159
It's not, it's not like an optimism necessarily about uh, I think the stuff like chip is gravy on top of it, which is that we are in an internet economy.
00:39:10.320 --> 00:39:23.280
There are open source technologies and transactions that can be built on top of blockchains, and it doesn't make sense to build like, like, you know, it doesn't fundamentally make sense to build a new credit market without a blockchain, right?
00:39:23.360 --> 00:39:37.760
Like, why would you construct like a trend, like if you believe the system is kind of fucked, like why would you build a new credit system in this legacy technology stack when you know that you probably need to bounce to the UAE in three years or bounce to Latin America?
00:39:38.000 --> 00:39:44.000
And it's like it's like we know that the future is going to be built on crypto rails, which is maybe a five-year story.
00:39:44.159 --> 00:39:47.679
One year story, we have Sailor, three-year story, we have Capital Flight.
00:39:47.760 --> 00:39:52.079
So on a one, three, and five-year time frame, I feel very, very good.
00:39:52.239 --> 00:39:54.000
And right now it's quite contrarian.
00:39:54.079 --> 00:39:59.760
And so it's like you should be there pounding the table when other people are depressed because that's when you make money.
00:40:00.000 --> 00:40:07.599
How do you not drive yourself crazy with this bull thesis sort of cemented and everything is pretty fucked?
00:40:07.760 --> 00:40:15.199
They're gonna hyperprint late stage capitalism, but crypto, but risk assets in the interim are gonna pump really, really hard.
00:40:15.360 --> 00:40:17.519
Like, how do you think through that?
00:40:17.679 --> 00:40:20.239
How do you not drive yourself nuts thinking through that?
00:40:22.239 --> 00:40:37.760
I mean, I think a lot of the uh really exciting uh areas to speculate in such an environment are uh you know, I think the answer is you keep yourself busy, like things like Avis rental, right?
00:40:37.840 --> 00:40:40.000
This recent, you know, 850%.
00:40:41.039 --> 00:40:51.599
Like a lot of the retail feeding frenzies that you're seeing um are exactly what you would expect in like a late stage capitalist economy, right?
00:40:51.679 --> 00:41:00.719
It's it's kind of like all financial meaning has been reduced to meme stocks and sort of like meme-like asset price movements.
00:41:00.960 --> 00:41:08.639
I think SpaceX, OpenAI, Anthropic, you've got like four trillion dollars of equity coming onto the market.
00:41:09.360 --> 00:41:20.079
And um, so it's it's going to be the biggest period of spec IPO speculative excess in financial history by an order of magnitude.
00:41:20.559 --> 00:41:24.639
And like none of this shit's in the SP 500 right now, right?
00:41:24.719 --> 00:41:33.039
So, like, like if you think about it, the SP and the passive investing markets need to absorb$4 trillion of bids, right?
00:41:33.119 --> 00:41:38.559
And you're like, okay, like I don't know like if that's even possible.
00:41:38.880 --> 00:41:50.239
And so I mean, the the sort of bleak thing is that you know, I think you can predict that there are gonna be a lot of people pushing assets really hard.
00:41:50.400 --> 00:42:04.239
There are gonna be a lot of fundamental investors who get short squeezed on things like Avis, and you're gonna have a lot of basically dislocations where people are screaming and you know, like, oh, this shouldn't happen, like, you know, all kinds of GameStop like events.
00:42:04.559 --> 00:42:11.360
And then by the time it ends, when it when it does end, it's gonna be like the mother of all shorts, right?
00:42:11.519 --> 00:42:15.599
It's gonna be just like it's it's sort of like 2021.
00:42:15.679 --> 00:42:25.199
Like, it's like like we know that equities have turned into crypto, and we know like as crypto people were super jaded, and we see these like AI conferences, and we kind of know it's like, yeah, this is like 2021.
00:42:25.280 --> 00:42:30.239
Like, like I know what happens next, and like trust me, it's like it's not great.
00:42:30.400 --> 00:42:34.480
But it's like, but it's like it's like most people don't know that, right?
00:42:34.719 --> 00:42:57.039
So, so I think the thing that's like is exciting to someone like me is like, okay, like you're gonna monetize a lot of the insanity until the SpaceX IPO and the Open AI and anthropic IPOs, but like after these events, you're probably looking at like a generational top inequities where you're just gonna make so much money like shorting this thing into the fucking ground, right?
00:42:57.199 --> 00:43:09.840
It's like uh and like the like I think that you know ultimately like the terminal, the terminal uh, you know, my my sort of simple frame is that I think governance in the United States has declined substantially.
00:43:10.000 --> 00:43:16.000
I think the the government is not trustworthy and doesn't enforce property rights effectively.
00:43:16.159 --> 00:43:23.119
Even things like the supply chain risk classification towards anthropic shows a government which is capricious, random.
00:43:23.440 --> 00:43:31.920
And in the long run, we know what that model looks like in China and other countries that are not property rights, you know, friendly.
00:43:32.239 --> 00:43:41.840
And it's like 11 or 12 times or 13 times earnings, uh terminal, terminal earnings uh multiple, not like a 29 times multiple.
00:43:42.239 --> 00:43:49.519
And so like the very like, you know, the long term playbook for me is like assume that like that's why we live in Puerto Rico, right?
00:43:49.599 --> 00:43:59.039
Like we don't we don't uh we don't believe that the system in the long run is um sustainable.
00:43:59.519 --> 00:44:04.559
And You know, the reason I'm in crypto is because I don't want to have money in the system after it blows up.
00:44:04.719 --> 00:44:05.280
Yeah, yeah.
00:44:05.440 --> 00:44:07.039
Um, and so that's how I think about it.
00:44:07.119 --> 00:44:14.320
Is like I'm just building a life that's robust to the system going through a major transition to something which like isn't that awesome.
00:44:14.559 --> 00:44:16.719
Let me ask you one more on this, and I'll I'm gonna go to David.
00:44:16.800 --> 00:44:22.639
Let me ask you one more because you're kind of rolling, is like, okay, how does this play out in the interim?
00:44:22.960 --> 00:44:27.519
And I mean, something that's been fascinating to me is I've never really paid much attention to geopolitics.
00:44:27.679 --> 00:44:33.599
And I've spent the last two months trying to figure out how to, you know, monitor the situation, how to monitor these headlines.
00:44:33.840 --> 00:44:41.519
And you see fake news everywhere, you see a market whipsawing on Trump tweets, most of which that are fake.
00:44:42.320 --> 00:44:52.320
And you see a market that's basically just completely shrugged off, uh, a war and just a generational bid on the SP.
00:44:52.400 --> 00:44:53.599
Like what happened today was nuts.
00:44:53.679 --> 00:44:59.039
The bid, the bid is so insane, it makes you like, you know, start believing in a higher power almost.
00:44:59.280 --> 00:45:00.880
And at the same time, you see this.
00:45:00.960 --> 00:45:13.920
I almost compared it to the Daily Runner era of Solana, where there's a new, there's a hotball of speculative money that whips us from car to hymns to Texas instruments, the thing, the thing, the thing, the thing, the thing.
00:45:14.239 --> 00:45:31.199
And now we all know how this ended on Solana, but there's a difference where our majors are, you know, BTC and and and Chud ETH versus the SP that has, like you said, this passive forever, non-stop, aggressive flow, and they're printing and they're buying and they're printing, it's just non-stop up only.
00:45:31.599 --> 00:45:38.320
And so, like, how does this manifest in like positioning and price action over the short term here?
00:45:38.559 --> 00:45:41.679
How long can hop all the money whipsaw around?
00:45:43.599 --> 00:45:53.519
Well, I think the very near-term catalyst, like, I think SpaceX is going to suck a lot of liquidity out of the market in the US.
00:45:53.840 --> 00:45:58.079
My my crypto proxy is that SpaceX is the Trump coin.
00:45:58.320 --> 00:46:00.719
Like it is, it is, it's like, yeah.
00:46:00.960 --> 00:46:02.159
It's just obviously the Trump coin.
00:46:03.760 --> 00:46:09.280
It's like, it's like just like the cursor thing, like manufacturing the financials to be ridiculous.
00:46:09.440 --> 00:46:18.320
Like, like, you know, like the reason Elon acquired cursor, Elon is is other, you know, actually Sailor is is probably one of the best security promoters of all time.
00:46:18.480 --> 00:46:29.519
Like, but I actually think Elon is number one uh in terms of just his ability to come up, like he, you know, he's got a finance background, he knows how to pump stuff.
00:46:29.679 --> 00:46:59.519
And he acquired Cursor because it either manufactures$10 billion of revenue or capacity utilization for for uh SpaceX, or it juices, it allows him to like point to like an enterprise like billing curve similar to Palantir that he can he can hype up and um he's going to announce all these things that basically he's trying to front run Sam Altman and say that we have like a$600 billion AI research lab.
00:46:59.599 --> 00:47:07.599
So actually, like when you're buying Starlink, you're buying the future and space data centers, like like you never want to fade Elon's.
00:47:07.920 --> 00:47:18.960
Like, I actually think Trump is probably amateur hour in terms of uh Elon, in terms of but but but to be fair, like this is a 1.5 to 1.7 trillion dollar IPO.
00:47:19.039 --> 00:47:23.360
So so actually, in my opinion, this is the main show.
00:47:23.519 --> 00:47:31.599
Um, and everything leading up to it is basically like a dance, like before the the main DJ comes on the set, right?
00:47:31.679 --> 00:47:37.760
Like we're still like on the pre, like the the main TS Do on TS Dow's coming, and that will be the main set.
00:47:38.079 --> 00:47:40.000
And my drugs haven't even hit yet, fully.
00:47:40.320 --> 00:47:41.760
After that, it's over, right?
00:47:41.840 --> 00:47:47.519
Like, like, like I think he's he's going first, and you're gonna have like this series, then you have two more catalysts.
00:47:47.599 --> 00:47:52.000
You have the open AI IPO and you have the anthropic IPO, and they're gonna be systematically pumped.
00:47:52.159 --> 00:48:00.639
I'm more bullish on Sam Altman's ability to kind of engineer narrative and like manage uh manufacture a stock price than I am on Dario.
00:48:00.880 --> 00:48:10.559
I think Dario is kind of like a weird guy who like crashes out and sends weird email, like you know, like I think the the anthropic thing, there have been a lot of self-owns.
00:48:11.039 --> 00:48:23.039
And um, so I think like basically Elon's gonna go out first, Sam is gonna go out second, and then anthropic will be like the last kind of like oh, like that's it.
00:48:23.199 --> 00:48:36.800
And and I think after that, um the sort of so so to to your to your question is like you basically after the SpaceX IPO, you start getting very like bearish equities, right?
00:48:36.880 --> 00:48:46.719
Like, like that's sort of similar to Solana, like basically, you know the the SpaceX IPO is gonna be the Solana 300 moment, right?
00:48:47.199 --> 00:48:49.039
And that's where we are right now.
00:48:49.199 --> 00:48:52.320
We're in this like max bid, like anything's possible.
00:48:52.480 --> 00:48:58.880
Everyone is incentivized, every investment bank is incentivized to come out and upgrade every single AI stock.
00:48:58.960 --> 00:49:07.679
That's insane because they're gonna get so many fucking fees when SpaceX IPOs and open AI IPOs they don't want any of these comps to look bad, right?
00:49:07.760 --> 00:49:20.559
So, so so JP Morgan will do like a you know some insane enterprise deal with cursor, or you know, cognition will sell like a cajillion seats to financially engineer uh these IPOs.
00:49:20.719 --> 00:49:24.800
And so everyone's in on it, and everyone is incentivized to make these numbers look good.
00:49:25.039 --> 00:49:36.159
Everyone's in on it, and so like like when it when it but then after those fees have been extracted, nobody has any reason to to play anymore.
00:49:36.320 --> 00:49:42.880
And so that's sort of my mental model is like it's like bullish until you get four trillion dollars of liquidity dumped.
00:49:42.960 --> 00:49:57.440
And and like the the what like I think the the the wake up the realization that people are gonna have is like that Elon is is basically working with SP and the NASDAQ to fast track inclusion of SpaceX into these passive vehicles.
00:49:57.679 --> 00:50:08.960
And so this is like the really big moment where we finally see what we see in crypto, which is like we know that we know that high FDV scam in crypto, right?
00:50:09.199 --> 00:50:19.440
But the SP has historically been immune to the high FDV scam because there's like a committee, and they're like, ah, I can't list this, it's too spicy.
00:50:19.519 --> 00:50:28.639
But like in this case, it's such a big deal, and there's so much money on the line that they're gonna jam SpaceX into the SP like way faster than they should.
00:50:29.039 --> 00:50:34.400
And basically the entire retirement, you know is fucking SpaceX.
00:50:34.960 --> 00:50:41.360
Oh yeah, and then everyone will max bid SpaceX, they'll own a ton of it, they won't even know it, they won't have done any of the due diligence.
00:50:41.440 --> 00:50:44.239
And it's basically like an uh it's like EOS, ICO.
00:50:44.320 --> 00:50:49.920
It's like it's like imagine if EOS got added to the SP 500, like that's like the crypto equivalent.
00:50:50.079 --> 00:50:55.519
And so after that, everyone's gonna be like, so it's gonna be like us in crypto, but with equities, right?
00:50:55.599 --> 00:51:03.360
Everyone, like after Trump coin went down from like 20 to like seven, they're gonna be like, oh man, like this is all a scam.
00:51:03.599 --> 00:51:06.320
I can't believe Goldman Sachs did this to me.
00:51:06.480 --> 00:51:07.840
Why did they put a buy?
00:51:08.000 --> 00:51:10.639
It's like, well, I mean, you're a passive investor.
00:51:10.800 --> 00:51:16.079
You trusted the, you know, you trusted the market cap of the stock, and that's what you did.
00:51:16.239 --> 00:51:31.840
So um, so yeah, I think that's that's my my sort of thought process is that we're bull mode until SpaceX, SpaceX is the Trump ICO, and then after that, it's it's it's it's the beginning of of the downfall.
00:51:32.000 --> 00:51:34.079
Um, that's sort of my equity view.
00:51:34.159 --> 00:51:34.880
Oh, crypto.
00:51:34.960 --> 00:51:42.800
I'm like, is obviously kind of the opposite because it's like, okay, like after this kicks off, it's like they're gonna be a lot of people trying to get the hell out of Dodge.
00:51:42.880 --> 00:51:54.400
There are gonna be a lot of people moving to Latin America, there they're gonna be a lot of people in Argentina and Uruguay, and you're gonna start seeing like very rich people very, very soon starting to look for exits.
00:51:57.280 --> 00:51:58.480
Jesus, man.
00:51:59.360 --> 00:52:02.400
That was how could Goldman Sachs do this to me?
00:52:02.480 --> 00:52:06.159
Yeah, everyone just gonna hold fucking SpaceX in the SP.
00:52:07.360 --> 00:52:08.000
Yes.
00:52:08.800 --> 00:52:09.199
Damn.
00:52:09.280 --> 00:52:17.039
I I was telling the chat before, by the way, before you came on that you came on in 2024, and I was reading them my one-shotted tweets that I had 30 minutes after you got off.
00:52:17.119 --> 00:52:26.480
I was like debating if I should tell my waiter about truth terminal and shit, and everyone was fucking clowning me, and then now they're like, Yeah, yeah, yeah, yeah.
00:52:27.039 --> 00:52:33.119
Um David, can you explain to me why my uh it feels like the models are getting worse, by the way?
00:52:33.360 --> 00:52:34.079
Do you have a take on this?
00:52:34.159 --> 00:52:35.760
You talk you registered it briefly at the beginning.
00:52:35.840 --> 00:52:39.920
Why is uh my claw so horrific now?
00:52:40.320 --> 00:52:44.000
Like what is happening with these models just getting worse, seemingly getting worse.
00:52:44.480 --> 00:52:47.599
I mean, it they they even say it to you, they're rate limiting you.
00:52:47.760 --> 00:53:02.880
Uh they they rate limit you and then they it's kind of like the it's kind of like uh Big Mac over the years, like they slowly add in all these excuses why it's getting worse when they're just trying to just survive with the fact that there's just too much demand and um uh not enough compute.
00:53:03.039 --> 00:53:13.280
Um it's entirely, I mean, if you know there's there's a constraint resource and it's just access to how much um tokens they can generate and uh how much compute they have access to.
00:53:13.519 --> 00:53:17.840
Um they saw the users grow like this, but then their computer's only growing like this.
00:53:18.000 --> 00:53:22.960
Um you see this with the pricing of um with uh with GPUs as well.
00:53:23.119 --> 00:53:26.079
Like uh this uh installations are getting slower.
00:53:26.239 --> 00:53:39.920
Um there's the helium thing with laser hormoths, like you have all the you have a bunch of supply shocks, and what happens is that installations are stale, they're slow, and the demand is higher, so the equilibrium is fucked.
00:53:40.159 --> 00:53:42.800
Um, just don't have enough compute.
00:53:42.880 --> 00:53:45.840
Um, we're in a compute strained environment, the scarcity is real.
00:53:46.079 --> 00:53:53.119
Um if we had infinite compute, then you would have everyone would have a very good model, but um, they're growing too fast.
00:53:53.360 --> 00:54:00.480
Um technically, I like as much as much hate as open AI got, um, investing in compute was the right move.
00:54:00.719 --> 00:54:04.880
It was they knew very well that that was gonna be the constraint in the future.
00:54:06.079 --> 00:54:10.559
I feel like there's a uh open AI vibe shift happening like aggressively.
00:54:11.599 --> 00:54:12.000
Yeah.
00:54:12.559 --> 00:54:20.719
Yeah, I mean um even in uh for a lot of Chinese models, like there there are a lot of GPUs being built out in Asia too.
00:54:20.880 --> 00:54:22.719
Um so they're getting a little better.
00:54:22.880 --> 00:54:26.159
Um, but we are in a computers constrained environment.
00:54:26.400 --> 00:54:35.119
Um I mean, speaking of sovereignties, like Alex is saying, like eventually like they'll just people will be net importers and next net exporters of tokens.
00:54:35.199 --> 00:54:38.239
Um tokens meaning obviously related to AI and not crypto.
00:54:38.639 --> 00:54:46.239
Um and they're literally um developing the infrastructure to import or export how many tokens they could generate for each jurisdiction.
00:54:46.400 --> 00:54:48.719
Um so it's it's becomes a sovereignty thing.
00:54:48.800 --> 00:54:53.760
Um and uh I guess more toward like what what is the purpose of AI in the future?
00:54:53.840 --> 00:55:01.039
Like, obviously we have Petrodollar, and this was extremely evident of like what I think is a manufactured crisis that we made in the Strait of Hormuz.
00:55:01.199 --> 00:55:03.199
We just, yeah, we just take on a military and see what happens.
00:55:03.360 --> 00:55:04.719
Of course, they'll go to US.
00:55:04.880 --> 00:55:15.440
Um but um just to reinforce that the that you know the petro is still like the world's top concerns, whether it's like perfuse in Asia, um, because they don't have access to oil.
00:55:15.599 --> 00:55:28.400
Um but in the future, like having proper AI mechan um AI for your citizens, it's gonna determine the efficiency of your nation national well-being.
00:55:28.559 --> 00:55:34.639
Meaning, like if you don't have AI versus a country that does have AI, um you're gonna fall so far behind.
00:55:34.800 --> 00:55:37.360
You're like a backwater country if you don't have it.
00:55:37.440 --> 00:55:41.599
Um you explain uh petrodollar but the token dollar thesis.
00:55:42.559 --> 00:55:46.159
Um yeah, I guess I guess it's more AI dollar, uh compute dollar.
00:55:46.480 --> 00:55:46.639
Yeah.
00:55:46.800 --> 00:55:57.199
Um, but um, I mean, it it just like I mean, this was like uh what happened in the 1970s, um, obviously moving off the gold standard.
00:55:57.280 --> 00:55:58.639
Um they wanted to ease credit.
00:55:58.719 --> 00:56:08.639
Um this I'll try to explain this as high level as possible because once you ease credit, then you can uh um you can really start like enabling ownership of different things, whether it's houses, right?
00:56:08.719 --> 00:56:20.559
Um, whether it's like um credit cards and uh and other things, because you can actually have um some elastic control over like um how credit expands in your in your country, um, in your currency.
00:56:20.880 --> 00:56:31.039
But this the most important thing of what makes a reserve dollar globally is like if you control the most important resource that determines your ability to grow as a nation.
00:56:31.119 --> 00:56:32.239
Um, that was oil.
00:56:32.320 --> 00:56:39.119
Um this is why it was like a it's kind of like a techno like features um nation in the 70s, like you had to control oil.
00:56:39.440 --> 00:56:47.840
Like just compare a citizen in China that didn't have oil versus a guy who has a car and can and like else has heating and else has factories and everything like that.
00:56:48.000 --> 00:56:57.360
Well, oil was literally the determinant of what made a nation survive at the current moment in time and what made a citizen fall behind for the next 20 years, 30 years.
00:56:57.599 --> 00:57:02.800
Um and it was a commodity that was effectively the mortality of a nation.
00:57:03.039 --> 00:57:06.159
Um and the same thing is true today.
00:57:06.320 --> 00:57:14.880
If you have a citizen in a different country that doesn't have AI, versus you that you that does, um, and all things work, your capacity is 10x.
00:57:15.039 --> 00:57:17.679
It's like the same thing as like somebody with oil in the 70s.
00:57:17.840 --> 00:57:26.320
Um the entire purpose of the petrodollars that we had to control oil in order to become sovereignty, in order to have sovereignty, let alone.
00:57:26.480 --> 00:57:28.079
If you didn't have oil, you're not a country.
00:57:28.239 --> 00:57:32.719
Um if you don't have AI, the thesis being um you're not a country.
00:57:32.880 --> 00:57:34.400
You are significantly behind.
00:57:34.559 --> 00:57:36.239
Your citizens are fucking retarded.
00:57:36.320 --> 00:57:38.400
Um, if you don't if you don't have AI.
00:57:38.559 --> 00:57:52.000
Um and that that's kind of like the outputs and GDP growth and everything, all that is and the well-being of the nation is dependent on well, compute um and how much tokens you're able to generate per compute and the infrastructure you built in order to enable that.
00:57:52.239 --> 00:57:57.760
So um, and then you like if if you don't have enough, then well, you just fall behind.
00:57:57.920 --> 00:58:02.800
So your sovereignty as a result, like it is to subsidize um the AI wave.
00:58:02.960 --> 00:58:10.719
Like every dollar that we're spending in the US, yes, it might be really petrol, but at the same time, it's also making sure that we win this AI war, um, this AI arms race in a sense.
00:58:10.800 --> 00:58:14.400
It's why we're someone spending so much money just to build out these massive data centers.
00:58:14.480 --> 00:58:18.159
Yeah, because it really, like I said, it is kind of like a project hilmary, like it's it's an arms race.
00:58:19.039 --> 00:58:23.039
Um, so it's uh it's fetrodal over, all over again.
00:58:23.119 --> 00:58:24.800
Like it's it's it's a national resource.
00:58:24.880 --> 00:58:33.039
Um, and uh we're make and we have to we have to kind of make sure that it's continuously being built, um, either massive scale and also smaller.
00:58:33.679 --> 00:58:35.199
It was a good explanation, thank you.
00:58:35.360 --> 00:58:39.199
Um, I guess as a follow-up, and one of the last things I'll ask you, Alex.
00:58:39.519 --> 00:58:48.000
I'll let you guys go soon, I'll go back to David after is you posted an article, I think, today about how these Chinese labs are white pill.
00:58:48.800 --> 00:58:53.920
Can you give us uh I was gonna actually read it on stream, but I said instead of letting you guys wait, I'll just bring you on.
00:58:54.159 --> 00:58:59.679
Um, can you give us like a high level of your take and your thesis, like how you think through that?
00:59:00.480 --> 00:59:00.960
Yeah.
00:59:01.119 --> 00:59:14.639
Um, so essentially before the last two or three weeks, uh using these models on your local machine or on a cheap GPU.
00:59:14.800 --> 00:59:26.400
And when I say cheap, I don't mean an H200, I mean like a$10,000 GPU, uh wouldn't generate something which is acceptable for what people are using most of these models for, which is AI coding, right?
00:59:26.480 --> 00:59:33.519
So specifically, everyone is using quad code and everyone is using codecs, and Elon is acquiring cursor for 60 billion.
00:59:33.679 --> 00:59:40.480
So we know the PMF in the market, the reason why the software index is getting smoked is because everyone is writing code with AI.
00:59:40.639 --> 00:59:46.719
And prior to two weeks ago, doing that on commodity consumer hardware was not really viable.
00:59:47.280 --> 01:00:03.440
Um, and so the big shift is, of course, that that has become viable uh because of these efficiently like you know, distilled, uh whether they're distilled or not distilled versions of Opus, these Chinese models are a new a new thing.
01:00:03.599 --> 01:00:22.239
And so uh it's relevant to the crypto space um because uh if there were ever going to be widely distributed AI spend through crypto rails, it would almost certainly be open source models.
01:00:22.400 --> 01:00:36.000
Uh the likelihood of open AI or Anthropic agreeing, given some of their comments around national security, to uh uh deliver spend um through their platforms, it just doesn't make sense, right?
01:00:36.079 --> 01:00:48.239
Like a lot of the Tempest arguments or even these stable these stable coin arguments are like, oh, like we'll have these bioauthenticated, like KYC'd users, it's like, okay, you're gonna have zero fee capture with that.
01:00:48.400 --> 01:00:54.400
The the chunky spend, the stuff that's like leading big financing deals with with USD AI.
01:00:54.559 --> 01:01:11.199
If there was a world where we only had anthropic and open AI and they were the only buyers of compute, they would be doing debt financing deals with Goldman Sachs and you would never hear of chip, you would never be talking about BitTensor or any sort of distributed compute protocol whatsoever.
01:01:11.440 --> 01:01:17.679
And so, like basically the market is just totally offsides because no one thought this was possible.
01:01:17.920 --> 01:01:28.559
Um, and uh, you know, the result is gonna be like you're gonna see like explosive growth in like Hermes Agent, uh, or all these like you know, open source harnesses.
01:01:28.639 --> 01:01:31.360
I'm so used to using codecs that I was testing this on codecs.
01:01:31.760 --> 01:01:33.280
This is where you're long Baba right now.
01:01:33.519 --> 01:01:34.159
Yeah, yeah.
01:01:34.320 --> 01:01:39.840
I mean, I think like I'm also long a ton of I'm I'm long all of the Chinese, like you know, Baba is the worst performing.
01:01:40.000 --> 01:01:43.360
Baba came out with Quen, which is and it's just a you know, the stock is a piece of shit.
01:01:43.440 --> 01:01:45.039
So I was like, okay, like ever.
01:01:45.119 --> 01:01:46.320
Why is it worse talk ever?
01:01:46.639 --> 01:01:53.199
Uh it's it's generally part of, I mean, if you look at Queb and the Chinese internet index, like it's down-only tech.
01:01:53.599 --> 01:02:00.559
Uh um uh because of a lot of governance issues with the CCP.
01:02:01.039 --> 01:02:04.079
You know, it's like all the all the same shit as usual.
01:02:04.159 --> 01:02:04.480
Yeah.
01:02:04.719 --> 01:02:23.440
Um, but I think the the thing is uh like the if if a company is doing models like this, if they if they show that level of expertise, the likelihood that they have something closed source that they can deal to enterprise clients is really high.
01:02:23.599 --> 01:02:35.280
The the likelihood of them having like you're already seeing random D-gens popping up on Twitter being like, oh, I quantized Quen and now I can run it on like an even smaller thing at like 100 talks per second.
01:02:35.519 --> 01:02:41.280
Like the likelihood that Alibaba hasn't figured that out internally if they haven't developed the model is very low.
01:02:41.519 --> 01:02:48.800
And so it's like, okay, like suddenly this is, you know, actually people used to buy meta because they had llama, right?
01:02:49.039 --> 01:02:56.000
And so I'm like, okay, like this thing is like the most uh consequential AI release since llama.
01:02:56.559 --> 01:03:04.239
Alibaba stock is hated, everyone is balls long open AI and closed source tech because that was the current meta.
01:03:04.719 --> 01:03:10.400
I was long soft bank, you know, I was shilling it to everybody who I could, you know, who'd listen.
01:03:10.559 --> 01:03:12.000
And now I've had the catalyst.
01:03:12.239 --> 01:03:14.079
Now we've had the big blow off top.
01:03:14.159 --> 01:03:20.719
And I actually think like um the next big narrative is gonna be uh, I mean, obviously GLM 5.1.
01:03:20.800 --> 01:03:22.000
I'm not sure if you're familiar with it.
01:03:22.079 --> 01:03:25.920
Okay, Knowledge Atlas Technology Company, like 2513.
01:03:26.239 --> 01:03:28.960
Uh, a lot of these are HK stocks, which are weird.
01:03:29.119 --> 01:03:32.639
Uh, but you know, it's just an amazing model.
01:03:32.719 --> 01:03:36.159
It's it's I think it's a better chat model uh than Opus.
01:03:36.480 --> 01:03:42.639
Like, you know, with with our own cryptocurrency, we we uh you know, we have a community of people who are interacting with these these models.
01:03:42.800 --> 01:03:47.280
And so we we run our own benchmarks basically with real user data.
01:03:47.440 --> 01:03:52.320
It's like, okay, like which of these models are are good and what which ones are generating.
01:03:52.480 --> 01:03:56.960
Uh what you might have noticed on Opus is that it's just so fucking verbose.
01:03:57.119 --> 01:04:00.800
It it has a tendency to say like like four-page ill center.
01:04:01.360 --> 01:04:06.239
I'm like, bro, just please just tell it's oh just tell me they answer like it's like one of those women that sends you like.
01:04:06.639 --> 01:04:09.039
I'm like, bro, just answer the fucking question, please.
01:04:09.440 --> 01:04:10.400
Like, what is this?
01:04:10.559 --> 01:04:12.559
And and the Chinese models are not like that.
01:04:12.719 --> 01:04:16.559
They're like, they're like very, which is a good chat experience, right?
01:04:16.639 --> 01:04:24.079
And so it comes out in the data, and you're like, wow, like like out of the top 10 models, like seven of them are Chinese at this point.
01:04:24.159 --> 01:04:26.719
And you're like, okay, like like this is not.
01:04:27.119 --> 01:04:41.760
You know, it's funny the claw thing like worked backwards because at first it was very like wow, it's so personalized, and it's very like nuanced, and it's hitting these like details, but then it just the personality hits this pendulum too far out, and all that's it's a very fine line, right?
01:04:41.840 --> 01:04:45.199
And then all of a sudden it's like, what the fuck am I supposed to do with this?
01:04:45.360 --> 01:04:47.119
Every answer, what is this?
01:04:47.440 --> 01:04:58.639
They also hired a woman from OpenAI who um, you know, originally like people got too upset about what you call psychopancy, like uh, which is them kind of glazing you.
01:04:58.719 --> 01:04:59.119
Yes.
01:04:59.280 --> 01:05:11.280
Uh, and so then open AI kind of overcorrected and they made a model that's like, well, you might not be thinking of these seven nitpick points, and you're like, dude, like, okay, like I'm just talking to you.
01:05:13.199 --> 01:05:15.440
Yeah, like don't give me these fucking nuances.
01:05:15.679 --> 01:05:18.639
And they they hired that that woman at anthropic.
01:05:19.119 --> 01:05:25.039
Okay, this is my biggest pushback, by the way, with Claude, is that I I've sort of like, I don't know how important the instructions are.
01:05:25.119 --> 01:05:29.760
I should maybe update mine, but I I had sort of because it used to be way too glazy, right?
01:05:29.840 --> 01:05:31.840
Claude, not so much, but it's very frustrating.
01:05:32.000 --> 01:05:46.400
And so now it's every time I ask it anything, it's like, sure, but what you're obviously not considering, and I'm like, bro, like I I asked it uh about USD AI, and it's in it gives me an answer.
01:05:46.480 --> 01:05:47.920
I'm like, what do you think of like this question?
01:05:48.079 --> 01:05:51.760
It gives me an answer, and it's like, but what you're obviously not considering is that it's a crypto.
01:05:52.400 --> 01:05:54.320
I'm like, it actually hates crypto.
01:05:54.559 --> 01:05:55.840
I'm like, what just did it?
01:05:56.239 --> 01:05:57.840
Is it a good question or is it not?
01:05:57.920 --> 01:05:59.760
I don't need the the the your your life story.
01:06:00.239 --> 01:06:00.480
here.
01:06:01.760 --> 01:07:58.239
Yeah, I mean that's that's that's the uh have you you're a physiologist you know based on our earlier conversation have you seen Boris Cherney yeah I have actually he's the guy what are you situating exactly he's I mean if you if you look at him and you understand that the the system prompts are written by this man that you should understand exactly what's happening right like like like at an organizational level they're getting a bunch of of dudes who are are making this product like this weird it's SBF dude it's like SBF backed anthropic because they're a bunch of fucking weirdos who got along with him and his effective altruist friends right like like SBF is victory lapping that all the time but it's also like dude like it's cause they're weird people like you know these guys are weird they're in polycules you know they're they're just a group of they're they're yeah they're odd they're Bahamas like they're ripping the freak offs and stuff yeah I mean it's sort of like if those are the people who are generating your chat models like yeah it's gonna get weird like actually I think Opus 4.7 as a base model what you'll see about their corrections when they're putting out uh releases I they leaked fucking clawed code and you look at the clawed code and you look at like the stuff that got leaked and it's like there's like 20 script files about like generating different animal shapes for different users and you're just like like what is this talk yeah what do you like like like what were you doing when you made this and like why like and and you can tell it's all vibe coded too right and it's like like they're they keep they keep shipping all these products that are like clearly vibe coded and they clearly think they're sick but then there are all these guys who post these viral threads they're like yeah the UX doesn't work in like 15 different places and it's like okay like like I think what's actually happening is that they're a world class model company that's trying to be a consumer company and it's just turning out really strange.
01:07:58.320 --> 01:08:01.360
It's hard right it's it's strange is a very good word for it.
01:08:01.440 --> 01:09:22.000
And I think we overindexed on the Sama hate really he's coming out of this looking kind of based Elon can't build a model that works Claude's gotten fucking weird and then the only thing was open AI was too fucking nice GPT was too fucking nice and all this well everyone crashed out because he worked with Trump like you know I think I think there is like this nuance that people like like maybe miss is that you kind of don't want an effective altruist to be designing your work product like you kind of want the guy who says yes to the military to be building your coding tool because you it's like he'll actually do what you ask it right it's like if he if Heggseth is like blow up that building and anthropic will be like I don't know are you sure like that building is like very ancient and like has always the thing with the war by the way is like man uh I you know if you looked at it from the anthropic angle in the Department of War thing it like made sense like all right fine you know Dario says I don't want my air models to be used in automated you know with automated killing weapons but then at a certain point it's like all right you know I think uh Palmer lucky had a decent clip on this it's like all right like are we calling you know we're gonna call Dario to get approval on it's like if you're you know the guy in the jet in over Iran it's like are you calling Dar it's like a very fine line weird spot on this war stuff.
01:09:22.239 --> 01:10:07.520
Yeah I think that's that's you know I think Sam Altman I think he's a very opportunistic guy I don't frankly trust either Sam Altman or Dario and I think that's like another like really good thing for crypto AI is that like you shouldn't you also shouldn't trust the CCP right so it's like your your your kind of three options with the closed source versions are like you can trust Sam Altman who's like you know like he's Sam Altman he's he's a like people who know are close to him there's a New York Times article so saying he's like a pathological liar he's like he's like a killer and then you've got like Amode who's like just like an SBF type character and then you've got like fucking Xi Jinping and you're like dude I don't want any of my data going to all three of these guys uh because I just don't trust them like at all.
01:10:07.600 --> 01:10:18.000
And give me Eric Voorhees well I mean the crypto advantage is that the Eric Voorhees can serve Chinese models without you sending data to the CCP.
01:10:18.159 --> 01:10:21.439
That's the whole it's it's it's not that I trust Eric Voorhees.
01:10:21.920 --> 01:11:21.840
It's that you know it's that people like Eric Voorhees can make crypto protocols where you can actually say okay like did my data go to the you know you know mini max server that has like an agreement with the Chinese military it's like no it didn't exactly it's like it's like maybe I don't love the Eric Voorhees like you know maybe like I don't know actually I have I I tried looking into Venice um and like all their terms of service and how they're doing onion routing with their I it's very it's not transparent to me like I don't um I don't have a strong view about that but it's like on the surface crypto sh is likely going to be the rails for Chinese uh open source models yeah that was uh honestly sick discussion also I don't know if you froze but I was gonna ask David a question anyways David I sort of like a sign off now that you guys are live token is live you have a bunch of contracts like what is your number one focus going forward here like what are you most focused on?
01:11:22.399 --> 01:11:32.319
I just do more loans that's pretty much uh we're we're very much a product riven company um which might sound odd just like we're we really are trying to build something that Ratify hasn't been up to solve.
01:11:33.199 --> 01:11:46.079
And I think that really does come down to like um how do you actually deliver a service that the market that is really unique to something that traditional finance or other offerings haven't been able to capture.
01:11:46.159 --> 01:11:55.439
And it's kind of like hyperliquid right like hyperliquid really is a is an instrument or product that's hard to replicate um in traditional Rails which is why it becomes like a very unique offering.
01:11:55.520 --> 01:11:59.119
Like people come from outside the space specifically use that product.
01:11:59.680 --> 01:12:08.960
For us, yeah we have like um right now about 70 ish customers um kind of coming in um large neo clouds and it's only increasing every day.
01:12:09.359 --> 01:12:25.760
We're just making sure that we can actually service them by having the capital um which is why we're we obviously we launched the um the the token but um we also want to start building out more of the liquidity um for SUSD AI and USD AI as well that's also why we're launching a Solana next week.
01:12:26.800 --> 01:12:34.000
And once you actually start serving these then people to Alex's point people will just use crypto protocols out of need.
01:12:34.560 --> 01:12:43.439
Much like our customers they're they've never used wallets before like our uh 50 to 70 customers they've they've they've literally don't know don't even know what a wallet is.
01:12:43.840 --> 01:12:45.600
We introduced them to like Coinbase Prime.
01:12:45.680 --> 01:12:46.640
It doesn't work in Europe.
01:12:46.720 --> 01:12:54.000
So one of them we have to send three like ledgers and they're like what the fuck is it you know this full um and then and then once they do it they're like addicted.
01:12:54.079 --> 01:13:02.319
They're like you can literally just borrow against your GPUs against a smart contract get money and then buy more GPUs and get more money.
01:13:03.039 --> 01:13:05.600
So it just blows your mind that you can do this.
01:13:06.319 --> 01:13:13.760
Just like lever looping um on DeFi but for using GPUs as your lever loop um because it actually generates money.
01:13:14.000 --> 01:13:21.680
You know like I I don't think people realize how much money GPUs make um this is actually how we found this sector.
01:13:21.920 --> 01:14:08.720
I was actually mining BitTensor um with uh yeah with like uh with with a GPU and then I realized I was paying like three dollars an hour or three dollars an hour um for this chip and then you know you do basic math um there's eight thousand you know seven hundred hours a year or something somewhere around there and you're spending three dollars an hour ends up being like more than 25000 a year but then the chip itself costs$25,000 and I'm like I'm paying this much in a year to mine this and I'm like this makes no fucking sense like I should just buy the GPU um so I bought the GPU I actually I actually just bought the GPU but you're paying a hundred percent of the value of the chip imagine that for real estate like if you bought a house for a million dollars but your rent is a hundred K a month you should just buy the house.
01:14:10.640 --> 01:14:14.479
Yeah um but you you can't unless you have debt it's too expensive.
01:14:14.800 --> 01:14:23.199
Got it so that that's kind of like our aha moment like yeah we gotta just buy the chip and then you gotta get debt against it to lever it up and actually scale it.
01:14:24.239 --> 01:14:32.319
But yeah just uh getting more um AI participants to borrow against us um and most of our fees come from AI companies.
01:14:32.560 --> 01:14:43.760
Got it they don't come from crypto customers so um and it's it's also just like I guess like kind of why um me and Alex being on the same pod um it's kind of like to illustrate like there there is a use case out there.
01:14:43.840 --> 01:14:45.840
Yes there are no a lot of good use cases.
01:14:46.239 --> 01:15:16.479
Venice is one of them obviously what me and Alex are building um it really want I I really want this to be like a new wave of like going past the jaded like world of just the um uh just the jaded world of just meme coins and like unproductive things and actually start building real stuff um and I I do think it comes with where the growth sectors are which is probably not RWAs or whatever they call them like you're just taking dinosaur assets and bringing it on like that's fucking sick man good job.
01:15:17.119 --> 01:15:22.000
You should really be pushing an industry zeta at the frontier which really is just AI today.
01:15:23.199 --> 01:15:23.680
But yeah.
01:15:24.079 --> 01:15:24.640
Sick.
01:15:25.359 --> 01:15:41.279
Well look um thank you so much for coming on Alex was awesome you were incredible congratulations on the launch and a successful at least up to this point a successful one up until this point and I am looking forward to what you guys do.
01:15:41.359 --> 01:16:11.600
I mean it's about time we have some fucking some talent some sauce some excitement in the crypto hierarchy if you will so maybe until part three Mr David it's an absolute pleasure man thanks for coming on thanks for your time thanks for having me on brother have a good one I'm gonna lie that was Jenny Did you that was Alex is a lunatic and the the uh USD AI thing is really cool.
01:16:12.319 --> 01:17:13.840
Alex is Jenny bro he just has like I mean it's hard to honestly we should tomorrow we should sit down and we should do some math and really figure out what we're gonna do into the SpaceX IPO like I I don't feel like we're taking it seriously enough Jenny's like generational I don't feel like we're giving it enough credit I'm not gonna lie I don't I like David a lot smart guy really smart guy and yeah good Alex is one of the best if you want we could do a solo with good Alex in the future um I by the way just for context I don't hold any chip I did hold it I traded it made some money got out and I'm not the most confident in crypto alts but I I do like it but if you want I will do a solo Alex and we could talk more in depth loot and rob and a lot of this stuff because I think he has some really good takes good takes on privacy how it's by himself is too intense.
01:17:13.920 --> 01:17:30.239
Yeah maybe we'll like throw D nap in the corner or something he he is very intense do you understand why I got one shot now I I I would imagine that is now making sense to you why that happened okay I got fucking I lost it I lost my mind because he's talking about how AI is going to ruin we're gonna yeah you you get it