Bessent Gets Drucked: Bitcoin, AI, and the New Macro Clock

[0:00] All right, as far as I'm concerned, uh
[0:02] I'm back in Brooklyn. So, uh summer for
[0:05] me is officially over. And for everyone
[0:08] else, if it's not yet, it will be. Uh
[0:11] let's go through it. Um
[0:15] Bessant gets drunk was the big story to
[0:17] start the week. Then you got Wars at
[0:19] Jackson Hole. Uh a bunch of podcasts for
[0:23] the week. I also uh I got so many people
[0:25] reaching out about Grockbot that I am
[0:28] uploading a uh a prompt. I highly
[0:31] recommend
[0:33] uh everyone who gets the opportunity uh
[0:37] to use it. They've talked about having
[0:39] free trials. I would definitely at least
[0:41] try it to understand the agentic side as
[0:43] that is going to grow. Um, and if you
[0:46] have the ability of finding a way to cut
[0:48] $200 out of your expense for one month
[0:50] and get it for a month so you can play
[0:53] with it for a month, I guarantee you it
[0:55] will be a uh
[0:58] a special time to learn about what AI
[1:00] agents actually do and how you create
[1:02] things. Um, there are so many good
[1:04] videos out there that you can basically
[1:06] use to make sure that you're getting the
[1:08] most out of it. Uh, I will spend time a
[1:11] lot more on uh the AA crypto macro
[1:14] collision. There's a lot of information
[1:16] and as I've talked about, I will be
[1:17] putting out a video specifically geared
[1:19] towards uh crypto. Uh in September, late
[1:24] September into October, um I'm guessing
[1:26] late September I should have what I want
[1:28] to get done. You guys have seen little
[1:29] bits and pieces of it. I'll go through
[1:31] some more. Uh it's been a big month for
[1:34] crypto and I will go through and just
[1:35] highlight some of the things in there
[1:37] and then um the Agentic AI mud trade
[1:41] continues. So, we will go through that.
[1:44] Uh, I highlighted it last week. I'm
[1:46] going to start showing you guys just
[1:48] like it appears everyone else uh the
[1:51] disclosure side of trying to make sure
[1:53] you guys understand I'm trying to help
[1:55] you guys educationally andformationally
[1:57] and I'm really trying to give you the
[1:59] opportunity to know that I'm outsourcing
[2:01] uh or you're outsourcing to me the
[2:04] ability to stay on top of AI and move
[2:06] things. So hopefully you're getting that
[2:08] feel. So let's go through this. Um it
[2:11] the week started off with uh Bessant
[2:13] basically saying he could tap the1
[2:16] trillion dollar general TGA.
[2:19] Uh and that was followed up quickly by
[2:22] I'd say the biggest story of the week
[2:24] which was Stanley Ducken Miller
[2:27] basically writing an op-ed. I think the
[2:29] reaction to it was in my opinion off
[2:32] base. uh Bessant
[2:36] basically the way everyone wrote it is
[2:39] he was blasting uh Duck Miller was
[2:42] blasting Bessant. I really if you read
[2:44] through it and I happen to agree
[2:47] completely with what Stan is saying on
[2:49] the surface which is the more
[2:52] manipulation especially into something
[2:53] like tenure yields uh it's the only
[2:57] fiscal disciplinarian the US has left
[2:59] and that's really the key thing is the
[3:01] fiscal disciplinarian line and that's
[3:04] because it's Congress that's the issue
[3:06] this spending problem has been going on
[3:08] for a long time the distribution of
[3:10] wealth problem has gone on for a long
[3:12] time the kshaped economyy's gone on for
[3:14] a long time. Um, and this cannot be
[3:16] resolved with slight interventions into
[3:19] yields and not allowing yields to go
[3:21] higher uh is definitely
[3:25] going to create other issues. But, and
[3:28] here's the thing that's the most
[3:29] important, we're at around 240 250%
[3:33] market cap to GDP. We're going to be at
[3:35] 300% soon. Uh, we're financialized. At
[3:39] the same time, the debt is enormous. So,
[3:41] you can disagree with what's going on,
[3:44] but there really isn't a solution that's
[3:47] easy, uh, other than allowing everything
[3:49] to head towards some sort of recession,
[3:52] which isn't an option. Now, I just want
[3:54] to make sure that you guys see that
[3:56] tenure rates are extremely mispriced
[3:59] relative to what we're doing in nominal
[4:01] GDP. And this is the issue is Donald
[4:04] Trump and Scott Besson have said they
[4:06] want rates lower. They're trying to
[4:08] handle the K-shaped economy. They have a
[4:10] belief that rates should be lower.
[4:12] Apparently, both of them do, or at least
[4:13] Donald Trump does. Uh here's where rates
[4:17] used to be. This is tenure rates, the
[4:19] orange line. And you can see that in the
[4:21] 80s, we were above nominal GDP
[4:24] throughout the entire time. And then at
[4:25] this point here in 1997, before the
[4:27] Asian financial crisis when I was in
[4:29] Brazil, that's when they really broke
[4:32] underneath. And they've been underneath
[4:33] since then, except for periods of
[4:35] recession, recession, recession. Other
[4:38] than that, we've basically been at a
[4:40] point where we've been keeping yields
[4:41] below nominal GDP. Well, we not only
[4:44] have nominal GDP right now at six and a
[4:45] half percent, but it's climbing. And so,
[4:49] the reality is for everyone involved,
[4:51] you can make the argument that rates
[4:52] should be at eight and change. They're
[4:55] not. So, this is where it gets into
[4:57] something. And I remember I was at a
[4:59] dinner
[5:00] uh not that long ago, I'd say two years,
[5:03] David Zervos was hosting it. And
[5:06] actually, I know when it was. It was it
[5:08] was 23. It was right after SVB. And
[5:12] every single time people at the table
[5:14] said this is what the Fed should do.
[5:17] This is what the fiscal side should do.
[5:20] And David Zervos would just stop
[5:22] everyone and say that's a normative
[5:23] statement. I don't care about it. The
[5:24] factual statement is what is happening.
[5:26] And that's really important for this
[5:28] stuff. As you read it, you can sit there
[5:30] and have the opinion that Stan Ducken
[5:32] Miller wrote in the article. And this is
[5:33] what I was going through. A a lot of
[5:36] what he's talking about is what should
[5:38] happen. We don't make money on what
[5:40] should happen. We make money based on
[5:42] what the administration is telling us
[5:44] they believe in. So I want to highlight
[5:47] there is another endgame and I actually
[5:49] believe this endgame has a a legitimate
[5:51] chance or is probably a better option uh
[5:55] if we actually hit this point than the
[5:58] other one. And this gets into the debt
[6:00] and entitlement issue which is based on
[6:03] human
[6:05] time. The debt is based on human time.
[6:08] It is the denominator
[6:10] uh or the the number on top basically as
[6:13] you go through this. You're going to
[6:16] have to find a way, and this is what
[6:17] they're trying to do, is grow yourself
[6:19] out of it, which is highly unlikely, but
[6:22] it's essentially the same debate as the
[6:24] AI infrastructure debt debate. When the
[6:26] hyperscalers are taking all this debt
[6:28] out, they're betting that they're going
[6:29] to get the revenues in faster than the
[6:31] debt is going to destroy them. Um, I
[6:35] think there's less of a chance of this
[6:37] working out well for them in a growth
[6:39] perspective and more likely at the
[6:42] government level because I do think GDP
[6:44] is growing faster than what's reported.
[6:46] The question is the tax system is not
[6:49] set up to benefit from it. So, this is
[6:52] where the issue comes in is it's great
[6:53] if you're trying to grow your way out of
[6:55] it. You still have to find a way to get
[6:57] the taxes in. And my guess is, if I'm
[7:00] right about humanoids and what's going
[7:02] to happen five, 10 years from now, 15
[7:05] years from now, whatever number you want
[7:07] to come up with, we're going to keep
[7:09] kicking the can down the road and see if
[7:11] at some point if humanoids are in there
[7:13] that were taxing corporate profits to
[7:15] pay for this stuff and the debt will
[7:17] grow at a place. But the one thing they
[7:19] don't want to do is have interest rates
[7:21] go higher. So again, it is virtually
[7:24] impossible, I agree, without exponential
[7:26] without thinking exponentially for us to
[7:28] get out of this problem. And that's why
[7:30] I keep saying every week, if you're
[7:32] listening to economists, you're
[7:34] listening to people who only think in
[7:35] human time by definition. They're
[7:38] academics. They've done all this work.
[7:40] The problem is if you're not focused on
[7:42] AI and crypto and you're not focused on
[7:44] how quickly things are going to change
[7:46] from a speed perspective, the question
[7:48] is can you outrun the debt clock? And
[7:52] I'm just saying there's no guarantee
[7:54] that you can't. It's part of the
[7:56] probability distribution. It is not
[7:58] zero. And this is what they seem to be
[8:00] attempting to do both the hyperscalers
[8:03] and these guys. So it is a huge bet and
[8:06] I think we just have to follow the
[8:08] facts. So the facts are yen intervention
[8:11] by Bessant
[8:14] late July August 1st. Then the quarterly
[8:17] refunding language changes.
[8:20] Then we get the buyback framework in the
[8:23] refunding statement. Then we increase
[8:26] the buybacks.
[8:28] Then he says the buybacks could be
[8:30] bigger on CNBC.
[8:33] And then this week possible TGA use
[8:35] discussed. So I this is a pattern. This
[8:40] is not what should happen. This is
[8:42] basically there. I'll get into WSH in a
[8:45] little bit. But there's a really really
[8:47] big problem with listening to what
[8:48] people are saying as opposed to what
[8:50] they're doing. This is clearly what the
[8:52] administration is trying to do. They did
[8:55] this yen intervention. Are they just
[8:57] letting rates go or is this something
[8:59] they're trying to manipulate? and can
[9:01] the markets for the first time beat out
[9:05] what the Treasury and the Fed wants to
[9:07] do?
[9:08] Um, this was a great um piece I thought
[9:11] by Andrea Stenol Lararsson. I'm not
[9:12] going to go through the whole thing, but
[9:13] I think it's important to understand
[9:16] that he's bringing up the repo markets a
[9:18] lot in this and heading us towards a
[9:20] situation where we're trying to
[9:23] effectively have the debt at the front
[9:27] end. And this gets back, in my opinion,
[9:29] into the speed side and growing things
[9:31] as opposed to allowing long-term rates
[9:33] where a lot of things are priced around
[9:34] the globe uh at a level where it doesn't
[9:37] just hurt us, it hurts everyone around
[9:39] the globe if 10ear rates go higher
[9:40] because everything is priced off of it.
[9:42] So, it's not just a US situation, it's a
[9:44] global one. But he goes through this and
[9:46] how important the basis trade is in
[9:48] here. You guys have probably at some
[9:50] point, especially when there's an issue,
[9:52] heard about the basis trade. I'm not
[9:53] going to go through every part of this,
[9:54] but if you wanted a visual and you
[9:56] wanted to learn, effectively, Besson is
[9:59] trying to find and making sure that
[10:01] there are buyers for long-term
[10:02] treasuries. The basis trade is one of
[10:05] the demand sides for it. You also have
[10:07] his goal to get stable coins going and
[10:09] getting trillions of dollars that way.
[10:11] There's other places that he's trying to
[10:13] do this in. He did it with bank
[10:14] reserves. He's going to do it with
[10:16] everything that he can possibly find,
[10:18] including trying to hopefully have the
[10:21] Japanese not have to sell uh their
[10:23] treasuries. And that's why he wanted to
[10:25] sit there and get involved with
[10:27] intervention in Japan. We'll see what
[10:29] happens in September. I think until we
[10:31] get to that point, you can expect the
[10:32] market to have back and forth,
[10:34] especially now that after the war,
[10:36] Jackson Hole speech, we're close to 60%
[10:38] chance of a rate hike and the BOJ is
[10:41] gonna tighten rates based on
[10:43] probabilities. What benefit has come if
[10:47] Besson's doing all this work and now
[10:49] Worsh is sending a signal to the market
[10:51] that inflation is his primary focus and
[10:54] they've built in now a 60% chance while
[10:56] the BOJ is trying to do something to
[10:59] stem the yen uh from falling any longer
[11:03] and the yen is at the weakest level in
[11:07] so far this month post the intervention.
[11:10] So if you go through the basis trade you
[11:13] can see where it is. It gives you what
[11:14] to watch out. I just say this as a
[11:17] point. Um, and this is the main thing
[11:19] for me is that I really do believe as
[11:22] people sit there and post when I or
[11:25] respond to me when I post this kind of
[11:27] stuff in X or if I even show it on this
[11:29] videos. You don't have to believe in
[11:31] anything related to AI. I know there's
[11:33] more skeptics than there are believers
[11:35] because people don't use it for much
[11:37] more than chat at this point and they
[11:39] believe that the doom and gloom side is
[11:41] going to happen. I'll get into more of
[11:42] the doom and gloom side, but you have to
[11:44] understand that if humanoids are
[11:46] building things in a decade, which seems
[11:49] legitimate based on what's already
[11:51] happening in China. Let's assume it's a
[11:54] decade. You've got a very different
[11:56] world. And I don't think you can assume
[11:58] that everything is going to play out in
[11:59] some linear fashion of human time again.
[12:01] And that's the reason why I keep going
[12:03] through these. I think you have to
[12:04] change your mind and at least open up
[12:06] the probabilities because it will help
[12:08] you make money as opposed to believing
[12:10] that every time that yields go a little
[12:11] bit higher and as I've highlighted the
[12:13] fact that 10 years are up whatever it is
[12:15] this year 40 50 basis points and people
[12:18] are taking this as being uh the end of
[12:20] the world it's just not they're way
[12:22] below where they should be that's not
[12:24] going to stop growth and it's certainly
[12:25] not going to change the dollar situation
[12:28] if they keep trying to get involved in
[12:30] it. They've sent a message they don't
[12:31] want yields higher. I think that's the
[12:33] most important thing that's going on
[12:34] right now. And I think trying to bet
[12:36] when this is going to collapse and end
[12:38] in a dalio style way, I just don't think
[12:40] that's going to make you money in the
[12:41] near term. And if for any reason rates
[12:43] do uh back up and you get Besson
[12:46] basically saying, "Okay, no moss." We'll
[12:49] deal with it then and we'll see where it
[12:50] is. Will you catch the top? No. But
[12:52] you're not going to catch the top
[12:53] regardless in terms of the market. The
[12:56] main thing is to figure out whether
[12:57] they're really going to let this thing
[12:58] go and cause a major calamity in in the
[13:01] United States or are they just doing
[13:03] what they've done throughout QE and
[13:05] everything else, which is the economyy's
[13:07] financialized, they can't have stocks
[13:08] fall in any major way. And if yields are
[13:12] the ones causing it, like we saw during
[13:14] the tariff on wine, as soon as yields
[13:16] went higher, they stepped in. They've
[13:17] continued that pattern. So AI is the
[13:20] main pressure point here because the
[13:22] capital needs are enormous. So the
[13:24] government is going to remain under
[13:25] pressure. This fight is not going away
[13:27] regardless of what inflation and jobs do
[13:30] unless we see a significant change in
[13:32] both of them. And right now I highly
[13:34] doubt I don't expect to see big job
[13:36] losses even if we stay on the weaker
[13:38] side. And I certainly don't expect the
[13:40] inflation to just fall off a map and go
[13:42] back down to where it was. Even though
[13:44] it is certainly more tame than the way
[13:46] people are talking about it to give you
[13:48] an idea of the numbers. I mean, this is
[13:50] just the hypers scalers in terms of as a
[13:53] percentage of all IG supply, it's
[13:57] already up to 9%. And we have a doubling
[14:00] from last year. We're still going to be
[14:02] growing. So, the numbers are going to
[14:03] continue to get bigger. Um, and now it's
[14:07] replacing in terms of bond purchases.
[14:09] You're getting net foreign purchases
[14:11] more on the corporate side than you are
[14:13] the treasury side. So you're getting
[14:14] this crowding out function in terms of
[14:16] making it harder and harder to find
[14:18] buyers around the globe of treasuries.
[14:21] Uh just to give you an idea, I mean
[14:22] Broadcom goes out and basically you got
[14:26] $70 billion deal that they announced and
[14:30] of course their CDS continues to go out
[14:32] and and all this CDX thing. Remember in
[14:35] the same way that we saw concentration
[14:38] we have about the amount of debt that's
[14:41] being issued by a few companies is
[14:43] enormous. It's all of the quote unquote
[14:46] circular finance companies. Um,
[14:49] basically the S&P 500, the rally from
[14:52] the great financial crisis on was
[14:54] dominated by the MAG7. Profit margins
[14:57] dominated by the MAG7. We live in a
[14:59] concentrated world, guys. Uh, so
[15:01] whenever you see the CDX, these single
[15:04] name should be going wider while the
[15:06] index should not be. Just think about
[15:08] what's been happening in the V of the
[15:10] equity side. We've seen equity V explode
[15:12] higher while index V hasn't. I think
[15:14] people need to understand this when
[15:16] they're dealing with the credit risk is
[15:17] not telling you a story. It's just
[15:19] because it's so concentrated and because
[15:20] there's a probability this won't work
[15:22] out and they should be buying hedges
[15:23] against it. Uh Nvidia $6 billion to
[15:27] build a open- source model to compete
[15:29] with China. So then we get Worsh
[15:32] concerned about inflation says the bank
[15:34] has has more work to do. Nothing he said
[15:37] was new. Um, I think people either went
[15:40] into this uh with the belief that if he
[15:42] gave anything inflationary, this wiped
[15:44] out what Besson was doing. I have no
[15:46] idea, but we did see the market respond
[15:48] to go from 35% to 60% chance based on
[15:52] something he said that wasn't new. I'm
[15:54] guessing based on his own words that he
[15:57] made some mistakes in the press
[15:59] conference that the reaction in the
[16:02] market remember two-year rates back in
[16:05] July 29th during the Fed meeting uh
[16:10] where we where they didn't raise rates.
[16:12] He came across as dovish and you saw
[16:14] two-year yields come down. He didn't
[16:17] want that message to come across. So now
[16:19] you have this. He comes out in Jackson
[16:21] Hole and basically as far as I can see
[16:24] that was the no I'm I'm not I want
[16:27] credibility. I am fighting inflation and
[16:29] so you get up to 57%.
[16:32] Again he's saying they want to fight
[16:34] inflation and until they do there's
[16:37] going to be more rate hikes. All right
[16:39] we've one and a half rate hikes. We had
[16:40] over one before the end of the year. So
[16:42] none of that is a major surprise. I
[16:45] think the bigger surprise out of this is
[16:48] when he said he was dovish, we saw
[16:50] yields back up on the long end.
[16:54] So now he's going to be hawkish and
[16:55] yields backed up on the long end anyway
[16:57] and we're back up to the level where
[16:59] starting here yen intervention. Okay,
[17:01] intervention, intervention,
[17:02] intervention. We'll see what happens
[17:03] next week, but basically 10ear yields
[17:06] despite him being hawkish. If that was
[17:08] the plan, if you be more hawkish, maybe
[17:10] this will maybe this will help the back
[17:12] end and we'll actually get what we want,
[17:14] which is long-term yields down because
[17:16] you're actually talking hawkish. Well,
[17:17] instead it did nothing. So, for everyone
[17:20] who thought if he just raises rates,
[17:21] long-term yields will go down. I'm not
[17:23] so sure that's the case. If this is
[17:25] being caused by the capital needs of AI
[17:28] and speculators believing that they're
[17:30] kind of caught here, which they are. So,
[17:32] here's what dollar yen did again. and we
[17:34] weaken to 160.
[17:37] Uh, important video to watch for
[17:40] everyone who's doubting AI. And I'll
[17:44] call this the best macro podcast of the
[17:46] week. Yes, this is Dylan Patel, the head
[17:49] of semi analysis. And this is his
[17:51] roommate, Dark Cash Patel, no
[17:53] relationship, who interviews all of the
[17:56] top Silicon Valley people. Uh, the fact
[17:59] that they're roommates mean they talk
[18:01] about this stuff all the time. And so
[18:02] whenever they speak, this is not the
[18:04] first time they've spoken, it's
[18:06] important. And they talk about compute,
[18:08] which I've brought up a bunch of times.
[18:10] They talk about the insatiable nature of
[18:13] compute. So let's just go through this.
[18:15] OpenAI and Anthropic are rapidly
[18:17] centralizing global compute. Now, I just
[18:20] want to highlight they don't talk at all
[18:22] in any way about open source being
[18:24] competition. So the first thing and I'll
[18:27] go through this at the end. I just want
[18:29] you to remember they're talking about
[18:31] OpenAI and Anthropic like they've
[18:33] already won. And the reason they've
[18:35] already won is because they basically
[18:38] have the highest margins. So when we go
[18:40] the reverse, oh, we're driving prices
[18:43] lower on the open source. Well, these
[18:45] guys are just collecting more and more
[18:48] money from the people that want to use
[18:50] the best intelligence and groups that
[18:53] are actually there. So this is the
[18:54] enterprises. So all of this stuff about
[18:56] them not getting money that Anthropic
[18:59] has reached a peak, these guys are
[19:01] completely saying the opposite. And I
[19:04] take them more than I I won't even name
[19:07] people in terms of what's going through.
[19:09] There is no way to take two companies
[19:11] whose ARR is growing the fastest in the
[19:14] history of the world and start trying to
[19:16] predict when the top of that is going to
[19:18] be when these guys are basically saying
[19:20] it all comes down to who owns the
[19:21] compute. And these guys have the highest
[19:22] margins, which means they can buy the
[19:24] most compute. Makes perfect sense to me.
[19:27] If they're securing all of the compute
[19:29] of the world, they will continue to have
[19:31] the best intelligence and they will
[19:33] continue to be in a position to make
[19:35] money. So what they're talking about is
[19:37] literally concentration of compute by
[19:40] these two companies. And again, I think
[19:43] when you go through this and you realize
[19:45] that why did Jeff Dean leave Google? Why
[19:48] did Demisabus reportedly try to leave
[19:50] Google? These companies are trying to
[19:53] get to AGI.
[19:55] Google is trying to find a way to make
[19:57] money. Meta is trying to find a way to
[19:58] make money. Everyone's trying to find a
[20:00] way to make money. They're public
[20:02] companies. They have no choice. Their
[20:04] free cash flows go negative. These are
[20:05] private companies that their entire
[20:08] fundraising and stuff is based not on
[20:10] them making free cash flow. It's on them
[20:13] being in a position to own AGI. So
[20:16] they're basically winning this and they
[20:18] go through the whole thing. They talk
[20:19] about regulation and safety pauses or
[20:21] the main breaks that could hurt the
[20:22] revenue per megawatt. But that's the
[20:24] main story is why can Anthropic pay so
[20:27] much money to XAI for Colossus Compute
[20:30] because their margins are so enormous.
[20:33] And so as you go through this they pay
[20:36] more well that makes it more difficult.
[20:38] They talk about how how little Kimmy K
[20:41] in terms of compute. All of these
[20:43] Chinese models they don't have the
[20:45] compute. you have to get the compute
[20:46] from somewhere and if anthropic and open
[20:49] AI can pay the most for the compute and
[20:51] we can't make enough compute relative to
[20:54] the demand then it prices everyone out
[20:57] of actually doing anything. So this is a
[21:00] completely different argument and this
[21:01] was brought up by Darkh on a podcast
[21:03] recently which Gavin Baker thought was
[21:05] interesting. This is them talking about
[21:07] that view. This is why it's so
[21:08] important. The US has pulled far ahead
[21:10] of China in a AI compute deployment.
[21:14] Then they start getting into something
[21:16] that I actually think is kind of funny.
[21:18] Um they start talking about interest
[21:20] rates and how the size of the capex
[21:23] which will go on from 2024 to 2029 will
[21:26] total about 11 trillion dollar 5
[21:29] trillion of which will be credit and how
[21:30] this is going to create a sovereign debt
[21:32] and interest rate shock. So, this is one
[21:35] way to know that when you're stepping
[21:36] out of your wheelhouse, I like all of
[21:38] this, but when they start getting into
[21:40] this and the interest rate shock, I'm
[21:43] sorry, we are also with AI going to lose
[21:47] a lot of debt. There's going to be a lot
[21:48] of defaults. There's going to be a lot
[21:50] of bankruptcies. There's going to be a
[21:51] lot of closures of companies. not in a
[21:54] disruptive synchronized way, but the
[21:56] reason these companies are winning is
[21:58] because the debt makes AI, which allows
[22:02] startup businesses to grow without
[22:04] taking debt. And we've seen this with
[22:06] the cursors of the world and a bunch of
[22:08] places until they get to the point where
[22:10] they get so big. But for companies that
[22:12] want to grow, they don't need people and
[22:14] they don't need debt. So, you got to be
[22:17] careful when you start stepping into the
[22:19] macroeconomic side of this. Now AI
[22:22] creates a higher return clock for
[22:24] capital. So this is the time disruption.
[22:27] This is where we're again getting back
[22:29] into the point where if you can't grow
[22:33] fast enough, if you're not fast enough,
[22:35] then your debt is screwed. You need to
[22:37] have returns coming on this. This is the
[22:39] reason why I said with the government,
[22:41] the hardest thing is going to be is if
[22:43] profit margins are growing rapidly. And
[22:44] that's what we're seeing right now. So
[22:46] GDI versus GDP. GDP and the taxes we get
[22:52] mainly from consumers with a small
[22:53] amount on the corporate side. It's just
[22:55] not going to work unless you go in and
[22:57] what do you end up seeing? Well, let's
[22:59] tax the wealthy and go through the the
[23:01] tax on the 182 trillion that's here.
[23:03] Well, that's not working. People are
[23:04] moving out of states and moving into
[23:06] other places. So, you're kind of left
[23:09] where do we get the money from? And the
[23:10] only thing I can see it on is a
[23:12] combination of the entitlement issue
[23:14] becoming less of an issue on Medicare
[23:15] and Medicaid due to the advancements of
[23:17] AI and at the same point finding a way
[23:19] to increase corporate taxes based on
[23:21] profit margins and profits continuing to
[23:23] go out. The old low rate economy gets
[23:25] crowded out and and again that's the
[23:27] crowding out that's going to happen is
[23:29] if you need low rates these guys are
[23:32] forcing rates to be higher and the
[23:34] reason is their margins are so enormous
[23:36] and rates are they're indifferent to the
[23:38] rates. doesn't change their ability to
[23:40] make money because they're paying 5%,
[23:42] 6%, 7% or 8% and they go through that
[23:46] the same way that it doesn't matter to
[23:47] them what the electricity costs are. 50%
[23:51] plus of a of a data center's cost is
[23:54] semiconductors.
[23:55] So, it's just not based on the credit
[23:58] side is not going to kill it. Um and
[24:01] again this is somewhat different for the
[24:04] hyperscalers and things like that where
[24:06] they do need to get the money in from
[24:08] the data centers being built. So higher
[24:10] discount rates crush long duration nonAI
[24:12] assets. AI compresses economic time. The
[24:15] old economy borrowed against future
[24:17] human productivity. AI infrastructure
[24:19] capital offers a faster claim. This gets
[24:21] back in in my opinion back to the
[24:24] drunken miller side. So again, when you
[24:26] go through this and you look at the
[24:27] centralization of labor and compute, I'm
[24:30] not going to read all of this, but
[24:32] basically one lab could hold more labor
[24:34] than humanity, you start getting into a
[24:36] very very
[24:38] uh AGI into RSI world. And the reason
[24:41] this becomes important is because we're
[24:43] really getting closer to that. And they
[24:46] talk a lot about how
[24:48] the amount of AI labors is growing by
[24:52] the week. It's it's an amazing amount of
[24:56] things that are happening with the
[24:57] agentic side and it's mainly happening
[24:59] right now at the places that have the
[25:01] most sophistication in compute. It's the
[25:04] reason why I want you guys to play
[25:05] around with Grockbot so you can start to
[25:07] understand what it means. Um the key
[25:09] point Dylan keeps coming back to is that
[25:10] compute will not be evenly distributed.
[25:12] It will flow to whoever can monetize it
[25:14] best. That gets back to the margins.
[25:16] Better models create more revenue. More
[25:18] revenue lets them buy more compute. More
[25:20] compute lets them train better models.
[25:21] better models let them monetize comput
[25:23] even higher rate. So again this becomes
[25:26] very circular not in only the financing
[25:28] side but the strongest flywheel ends up
[25:31] lining up here. You don't have to agree
[25:33] with this. These guys are far far more
[25:37] uh more uh in tune with what's happening
[25:39] than anybody at the banks than anybody
[25:41] you're reading on X. And they're young
[25:44] enough to basically be spending all
[25:45] their time. I read all the work that
[25:48] these guys do whenever I can and I
[25:50] listen to the podcast. They are very
[25:53] very on top of things based on
[25:55] everything that I've heard this year.
[25:56] I've heard them all multiple times. Um
[25:59] Dylan's counter. Oh, the open source
[26:02] thesis says model weights will
[26:03] commoditize AI. Dylan's counter is that
[26:04] compute doesn't commoditize when it's
[26:06] scarce. I've said repeatedly the most
[26:08] important story in the world is we don't
[26:10] have enough compute. We'll get into that
[26:12] later with the thematic agentic
[26:15] portfolio. So AI is real bottleneck.
[26:18] Open source can commoditize models.
[26:20] Scarce compute does not scarce compute
[26:23] is the key thing. Cheap capital wins the
[26:25] buildout. Higher dollars per megawatt
[26:27] can outbid everyone for compute. So the
[26:30] fact that we are scarce compute is what
[26:32] drives this factor.
[26:36] And here's what you get. The demand is
[26:39] growing exponentially.
[26:41] Everything over here, the design
[26:44] permits, all this, the compute side, the
[26:46] supply is growing slowly and these guys
[26:50] models are getting better and better and
[26:53] better and the disruption is taking off.
[26:56] It's just a reality of where we are. So,
[26:58] Vera Rubin getting uh getting it out
[27:01] there and getting Microsoft talking
[27:03] about using it. This is important. Um
[27:05] I'm I'm less worried about this from uh
[27:08] or I think the major story from this so
[27:10] you guys get it is
[27:13] open AAI this chip is developed by AI.
[27:16] This is starting to get into the side of
[27:18] where AI starts solving problems. So if
[27:20] you want to take it as a negative on
[27:22] Nvidia good luck on that uh Nvidia is
[27:24] just so massive and has so much capital
[27:26] and has just built so many of these that
[27:28] to be able to scale this is really the
[27:30] issue to compete. So um it's really more
[27:33] to say that there will be future
[27:36] situations where as AGI does come we
[27:39] will be solving even the physical side
[27:41] which is where I start to get into the
[27:43] terminal value the S&P and the fact that
[27:44] I think you will continue to see
[27:46] multiple compression and we will never
[27:48] get that quote unquote bull market that
[27:51] feels like a bull market where things
[27:53] just explode higher and the reality is
[27:56] that's because competition is coming
[27:57] three years down the road for
[27:59] everything. It's that easy to use
[28:01] Grockbot. I again I'm just highlighting
[28:03] the fact that this is becoming more and
[28:04] more of a story. Um more usage for it in
[28:08] terms of this part of what I'm going to
[28:10] give you is a prompt that will allow you
[28:12] right off the bat to analyze your
[28:14] emails, your calendar, things like that
[28:16] in a safe manner.
[28:19] Um
[28:21] Gavin Baker just talking about how much
[28:24] compute or the spend that he's doing.
[28:26] And then note that it is before Grockbot
[28:29] moves to consumption pricing which will
[28:31] likely create a step function when it
[28:32] happens. At least for me. I completely
[28:35] agree with that. And here's the latest.
[28:38] I showed you guys this last week uh
[28:41] multiple times. You can see I have a lot
[28:44] more agents than I did last week. I have
[28:46] a lot more routines that are being done
[28:48] every day. I have these two things
[28:51] displace and stride. This is a technical
[28:54] um indicator with three agents. This
[28:57] here is a probability uh it's a
[29:00] research-driven probability of based on
[29:03] earnings how vulnerable they are to AI
[29:08] disruption in a few years. So on a scale
[29:10] of one to 100 give me a level on that.
[29:12] So, I built those and this is what
[29:16] you're going to get. So, this will be
[29:17] uploaded to the subscriber site. You
[29:19] guys can read through it, but this is a
[29:21] chief of staff starter pack for
[29:22] Grockbot. The Grock part Grockbot
[29:25] starter pack. Um,
[29:29] and then just again, powering the
[29:30] deficit, AI's massive infrastructure
[29:32] needs now threatened the US national
[29:34] debt. This is becoming a bigger story.
[29:36] This is why again you can't have the
[29:40] rates go higher because of the K-shaped
[29:42] economy. My gut tells me that Kevin
[29:44] Worsh will not raise rates in September
[29:48] even though he may want to for his
[29:51] already legacy and for being an
[29:53] inflation fighter. Um, but I I just
[29:56] think the timing is not right for him to
[29:59] sneak one in at this point while tenure
[30:01] rates are up pressing against while his
[30:04] Treasury Secretary is trying to stop
[30:05] them and they just spent all this time
[30:07] trying to focus on the yen where the be
[30:09] they're trying to get the BOJ to raise
[30:10] rates. So, I think September could be a
[30:13] surprising thing. Is it possible that
[30:15] maybe Japan not only raises rates but
[30:18] then builds in that they're going to do
[30:19] more and they're really going to fight
[30:21] this? At the same time, Wars doesn't
[30:24] raise rates now that it's a surprise and
[30:27] we get a big move. We'll see what
[30:29] happens, but um I think the pressure is
[30:31] growing and I think it's worth spending
[30:33] time on. So again, it's virtually
[30:35] impossible to grow your way out of a $40
[30:38] trillion debt crisis. And although I
[30:41] would happen to agree if there wasn't
[30:43] this exponential innovation side, uh I
[30:45] do think it's worth keeping in the back
[30:47] of your mind as we get to a critical
[30:49] point leading into the midterms. That is
[30:51] the other thing that I think is
[30:52] important for people. Is it really
[30:54] likely that we're going to take any any
[30:58] risk with the long end at this point
[31:00] ahead of the midterms? Are they really
[31:02] going to let tenure rates break higher
[31:04] and go into the midterms that are
[31:06] already in a vulnerable position in the
[31:08] Senate? Um, I just think at this point
[31:11] you have to heir on the side of thinking
[31:12] what do they really want and this
[31:14] administration does not strive does not
[31:16] move too far away from trying to make
[31:19] sure they get what they want. So again,
[31:21] I wrote this paper. I'm just reminding
[31:23] you this was on August 10th. This was
[31:24] the end signal and AI agent macro nexus
[31:28] point getting back to the point that
[31:30] that's where this whole thing started.
[31:31] And the reason I wanted to bring it up
[31:33] is because Pomp did a video with uh
[31:36] Roman Yumplowski.
[31:39] It's worth listening to. Um my son had,
[31:43] ironically enough, reached out and said
[31:45] to me, "Have you listened to him speak?"
[31:47] Uh and he sent me a diary of a CEO, one
[31:51] that was old. And it was really about as
[31:54] much about how AGI is going to have a
[31:58] negative impact on the world. And I
[31:59] think he has publicly said 99.99%
[32:02] chance that humanity is destroyed by uh
[32:05] AI. Now he's a very respected person who
[32:09] is been writing about AI and talking
[32:12] about AI from like 2011 to 2013 from a
[32:16] safety perspective. So it's not
[32:18] surprising that uh I'm sure his public
[32:21] speaking things he's makes more money
[32:23] and does better if he talks about the
[32:25] negative side. What I thought was
[32:26] interesting was he's the only person
[32:29] I've seen connect as I did in the yen
[32:31] signal one. The AI
[32:34] needs, the AI growth, the capital needs
[32:37] to keep this intelligence going is
[32:39] putting pressure on yields. It's putting
[32:42] pressure on the old system and the old
[32:43] system is cracking. So now we've got not
[32:47] printing QE style but we definitely have
[32:49] twisting and we have using the repo
[32:52] facility and we have them clearly trying
[32:54] to find a way to not let yields go
[32:56] higher and that was the reason why in
[32:58] there I said I think this is the
[33:00] beginning of Bitcoin gold and everything
[33:02] else right after the yen side happened.
[33:04] Now when you go through what he said
[33:07] he's not mentioning any of the macro
[33:09] side. This is an AI researcher. If you
[33:12] listen to him, he speaks a language we
[33:14] don't speak. But he eventually gets into
[33:16] the exact same points that I make
[33:19] repeatedly in these videos that
[33:21] eventually because of AI abundance
[33:24] because of the lack of scarcity meaning
[33:28] things that can't be replicated very
[33:30] quickly that as we continue to
[33:33] accelerate in AI and the frontier models
[33:36] are much further ahead than what the
[33:37] public sees, it will fundamentally
[33:40] change labor in the economy. all
[33:42] cognitive jobs become automatable,
[33:44] including new jobs invented afterward.
[33:47] And he goes through the reasons why. And
[33:49] then he talks about
[33:51] the only things that'll be left for
[33:53] Bitcoin, waterfront property, and
[33:54] potentially human attention, meaning
[33:56] what has value.
[33:58] So, I didn't realize that he was that um
[34:01] as I was listening to it was a Bitcoin
[34:04] person. And he gets into the reason that
[34:08] it is. He goes through the importance of
[34:09] cryptography. And again, this is a
[34:11] safety person. The ability to copy
[34:14] things, how we need it verified. AI
[34:17] acceleration is fundamentally about
[34:19] compressing scientific and technological
[34:21] time. So, he's getting into the time
[34:23] thing the same way I did. So, I picked
[34:25] out the point that I wanted him to spend
[34:27] more time on, which was the AGI part,
[34:29] which is right here. And he talked about
[34:32] it where the implication is
[34:33] technological progress starts operating
[34:35] on a different clock from human society.
[34:37] This is the pressure we're seeing. It's
[34:39] not just the debt side again. It's this
[34:41] point that on the one side things are
[34:43] growing insanely fast like anthropic
[34:45] like open AI fact like the AI models in
[34:48] terms of the intelligence side facts
[34:50] like the capex side not just from open
[34:52] AI and them Nvidia spending money Oracle
[34:55] spending money uh Google spending money
[34:59] Amazon spending money Microsoft is you
[35:01] guys get it these are the winners of the
[35:03] technology world they see this coming
[35:05] they see the models that are already
[35:07] here and so we're getting to the point
[35:10] where accelerating drug discovery,
[35:13] improving our understanding of genomics
[35:14] and pro all of the things that I've
[35:16] written about and talked about and where
[35:18] we are. This creates a widening divide
[35:20] on digital time and physical time. We're
[35:22] in there. So AI acceleration is
[35:25] ultimately the compression of the
[35:26] discovery cycle. The defining threshold
[35:27] is reached when machines no longer
[35:29] merely help humans innovate but
[35:31] autonomously decide what to discover.
[35:34] He even gets into the difficult that
[35:37] that will be for
[35:39] companies in general and that you can't
[35:42] compete. Um, and then he gets into the
[35:45] physical uh bottlenecks such as chip.
[35:47] They're unlikely to permanently stop
[35:49] this process because he's saying as it
[35:52] gets smarter and smarter, we will keep
[35:54] getting more advancements.
[35:56] So
[35:57] the entire loop becomes automated. Now,
[36:00] what I did was that interview that my
[36:02] son sent me to watch had 21 million
[36:06] views. So, I'd never heard of this guy
[36:07] three weeks ago. And then Pomp does an
[36:10] interview. My son had already sent me
[36:12] something. So, what I do is I go compare
[36:16] what he said with Pomp with what he said
[36:21] a year ago and see if anything changed.
[36:23] Now, Bitcoin becomes a major theme. It
[36:26] was not even mentioned in his his
[36:28] podcast a year ago. Everything else is
[36:31] kind of in the same, but um it's like
[36:35] shifts. Nothing major. Humanoid robots.
[36:38] There's a little time in the in the in
[36:40] the time, but all of these things here
[36:43] are basically major things that he
[36:45] talked about a year ago. So, it's not
[36:46] like he's just saying this now based on
[36:48] what's happening. Now
[36:51] as you go through it and you look at the
[36:54] uh other parts in here, it's all the
[36:57] same. The shift is the government role
[36:59] has become more aggressive and the open
[37:02] source thing even surprised him in terms
[37:04] of how quickly it it it came. Now what I
[37:06] wanted to do was I released this this
[37:09] week for those of you who didn't see it.
[37:11] I've been saying that Bitcoin's been
[37:12] going through an IPO uh process that in
[37:15] my opinion started right after Donald
[37:18] Trump took over. even though that was
[37:19] not the peak and that's when I believe
[37:21] that the ideologues gave up. Um the ETF
[37:24] was not a good sign for people who
[37:26] wanted decentralization and not having
[37:28] the two systems merge but especially
[37:30] when the president endorsed it and then
[37:32] created a memecoin. I think that was a
[37:34] signal that this is no longer what
[37:36] people had gotten involved in and you
[37:38] saw a lot of people come out. So what I
[37:40] was saying is that Bitcoin AI and the
[37:42] collision of time and going through what
[37:44] I believe is now happening and why this
[37:46] is the critical moment for crypto and I
[37:49] said compare what I wrote to what he's
[37:52] [snorts] saying what he just said with
[37:54] Anthony Pompiano. These are all the
[37:56] different points. It's a time disruption
[37:58] human time. These are all things that I
[38:00] wrote about in the paper. Never seeing
[38:02] or hearing this guy. We're in agreement.
[38:04] The only reason I bring it up, you can
[38:06] disagree with me. You can disagree with
[38:08] Yman. uh with Roman Yampolski, but
[38:12] we're from two different worlds. I am
[38:14] not an AI person. I am not someone who
[38:17] just uh got involved with it uh 15 years
[38:21] ago from a safety perspective, but we
[38:23] are in heavy agreement on almost
[38:25] everything except for the way it ends
[38:27] the world. That's about the only thing
[38:29] that's not there. Now, for Bitcoin, just
[38:32] I'm going to do some more Bitcoin later
[38:34] on, but I I put this out because the 200
[38:36] day moving average slope
[38:39] turned up. So, these are the four times
[38:42] in history that we had the 200 day
[38:45] moving average pointed downward for at
[38:47] least 100 days. Uh, ironically enough,
[38:50] look at these declining observations.
[38:52] Last one took was the exact same amount
[38:54] of time. This one was only off by 13
[38:56] days. This one was off by 60. Here are
[38:58] the performances for Bitcoin over the
[39:00] next three months, six month, one year
[39:02] and two year.
[39:04] Uh in all cases, we never took out the
[39:06] lows uh from before that going down. So
[39:09] that would mean around 58,000
[39:11] would if this holds over history would
[39:13] hold. We got South Korea back in. Um
[39:16] I've mentioned before that AI had been
[39:18] taking everything in South Korea. Well,
[39:20] we had a volume spike in the crypto
[39:22] world over there. Um, I'm going to
[39:24] remind people the amount of people that
[39:26] basically were bearish at the lows. Uh,
[39:29] the most famous one being Jeremy
[39:31] Grantham saying it was less worth less
[39:33] than a bucket full of piss. That was at
[39:36] the lows. That was two days from the
[39:39] lows. Uh, and and really at the lows. It
[39:42] was basically 59,000ish when it was 58.
[39:44] Jim Kramer sold all his BTC two weeks
[39:48] ago. Uh Mark Cuban went on, I sold most
[39:53] of my Bitcoin. It's lost the plot. Uh
[39:55] more technical things in terms of a
[39:58] different way of looking at it. Uh this
[40:00] is a breakout signal that takes uh
[40:03] basically the rate of change and puts a
[40:05] zcore on it. And again, you had a big
[40:08] move,
[40:10] not shocking. Japan to start stocks and
[40:12] bonds tokenization development plans
[40:14] this year. They're racing towards
[40:15] moderniz. In Japan, if you go back and
[40:18] just ask what has cha changed in Japan
[40:20] with their views towards Bitcoin
[40:22] tokenization and do the same for South
[40:25] Korea, there is no way you can ignore
[40:29] that this is starting to become a bigger
[40:31] race just like the administration has
[40:34] said. Why AI forces tokenization?
[40:38] So why does AI force tokenization? Well,
[40:40] it needs programmable money to be able
[40:42] to go. It it compresses decision time.
[40:45] The tokenized assets become better
[40:47] collateral for AI. The fastest rails
[40:49] attract the capital. This becomes
[40:51] important as I get through with Japan.
[40:53] AI makes settlement speed a national
[40:55] competition uh competitive issue. So
[40:59] tokenization democratizes capital
[41:01] formation for the AI rebuild. This
[41:02] becomes critical if you want to raise
[41:04] the capital that's needed for AI.
[41:06] Tokenization helps that because if you
[41:08] guys haven't seen it, if the only people
[41:10] that can buy private credit are people
[41:12] that are
[41:14] wealthy, but you want everyone to be
[41:17] able to buy 8% yield bonds on things
[41:19] that are there or have the
[41:20] diversification to go through.
[41:22] Tokenization will democratize the
[41:23] ability to raise money from anyone. Why
[41:26] does Japan need to have a stake in the
[41:29] tokenization race? Capital moves towards
[41:31] the best operating system. Japan cannot
[41:34] afford to be slow. Global investors want
[41:37] speed, liquidity, yields are blah. If
[41:39] tokenized US assets trade and settle
[41:41] faster than Japanese assets, capital
[41:42] does not need to panic leave Japan, it
[41:45] can quietly migrate to the market with
[41:46] the better rails. Basically, if you're
[41:48] in a market where your stuff settles in
[41:51] a day while everything else settles in
[41:53] seconds, AI agents will continue to
[41:56] treat you differently. The same way we
[41:58] treat emerging markets differently in
[42:01] terms of wondering if their government
[42:02] is going to confiscate things or go
[42:04] through it if you haven't invested in
[42:05] the best technology. So, this is causing
[42:07] a race. Human time rails while or I'm
[42:11] sorry, tokenization compresses financial
[42:12] time. If Japan assets stay on human time
[42:14] rails, while global capital moves to
[42:16] software time rails, Japan does not just
[42:19] risk missing a technology channel. It
[42:21] risks lossing the marginal capital that
[42:23] sets the price of money. I really do
[42:25] think of Blackberry in this case as an
[42:27] example of something that just could not
[42:29] compete with the iPhone. Forget the
[42:31] iPhone and uh Android in terms of open
[42:35] source and closed source. Think of just
[42:37] what happened uh in terms of BlackBerry
[42:40] and not having the apps. tokenization to
[42:42] me and the ability to move things
[42:44] quickly, make decisions, having better
[42:46] liquidity, having democratization of
[42:48] capital. Okay, Ashe BD, I really liked
[42:53] this post in terms of u last week on the
[42:56] tokenization super cycle. Uh I think his
[43:01] analogy here in terms of what YouTube
[43:03] brought again is a democratization side.
[43:06] Distribution is valuation. the more
[43:09] distribution, the more people are
[43:11] involved and theoretically the more
[43:13] you're going to be able to trade things
[43:14] at least that day and know where the
[43:16] price is. Um, distribution was always
[43:18] one of the great moes of US capital
[43:20] markets and that monopoly has now met
[43:22] its final boss internet scale
[43:23] distribution. So the US has dominated
[43:26] capital markets not just because people
[43:28] are wealthy but because it work the
[43:30] deepest capital markets. If all of a
[43:32] sudden we have a competitor for capital
[43:34] markets which is tokenization
[43:36] creates an issue. Um again I'm showing
[43:40] you guys this because as I go through
[43:43] the next four things remember every week
[43:47] on the sub on the website I put up the
[43:49] video podcast info. All the podcasts
[43:51] that I reference in here those are not
[43:53] all the ones I listen to for the week. I
[43:55] go through I don't know 10 maybe 15
[43:58] different podcasts. I go through more
[44:00] than that in terms of transcripts to
[44:02] determine whether I should go look for
[44:03] it. Sometimes I find a transcript where
[44:05] I like the timestamps. I go listen to
[44:06] that 15 minute period. I use those in
[44:09] here. I will go listen to it. If there's
[44:11] more nuggets, this one is big because I
[44:13] have a lot of things in here that I
[44:15] spent a lot of time on. And this is
[44:16] starting to get into the crypto side.
[44:18] For those of you who want to learn more
[44:20] about tokenization, which I'm telling
[44:21] you is the most important thing to be
[44:23] spending time on at this point along
[44:24] with the payment rails in general. this
[44:26] token 49 Singapore event which was a
[44:30] year ago. This is 2025.
[44:33] Ash K ash um BD spoke and I watched it
[44:37] and I thought it was great. Um
[44:40] capital markets no longer work for
[44:42] ordinary people. Labor has been left
[44:43] behind while capital owners won. He goes
[44:45] through this whole thing on why
[44:48] tokenization is important based on
[44:50] what's happened to the world. He talks
[44:52] about Salana specifically, but the
[44:55] internet internet capital markets and
[44:57] what it's going to look like. For those
[44:59] of you trying to I I guess get to the
[45:02] point of number one, why AI agents are
[45:06] necessary for this to be the point. You
[45:09] cannot do this stuff without AI. And so
[45:12] in 2022 where these where you know we
[45:16] were talking about these types of
[45:17] things, it was still a dream. there was
[45:19] no ability to do this and we didn't have
[45:22] Salana and Ethereum to the degree that
[45:24] they are today. So all of this stuff in
[45:27] terms of what the end state is, a lot
[45:29] more assets, wallet based ownership of
[45:32] those assets, not just assets, but
[45:34] things like IP and everything. Um,
[45:37] productive assets. So again, anything of
[45:40] value that you can think of that right
[45:42] now you can't trade. Broader ownership
[45:44] could reduce inequality and social
[45:46] strain.
[45:48] Think about this and go back to the
[45:50] Bessant thing. If you're going to make a
[45:52] bet at this point, the new technologies
[45:54] are the place to make a bet. Not
[45:56] thinking about the world from a economic
[45:59] handbook that goes through the history
[46:01] of the world like Ray Dalio does and no
[46:03] offense to Dra who I love, but the same
[46:05] type of thought process. It's a you have
[46:09] to at least incorporate this in the
[46:11] possibility. I'm not saying it's going
[46:13] to work out. Governments could stop it,
[46:15] but this is not a US clarity act
[46:18] decision. This is a global decision.
[46:20] Salana is not a US thing. Ethereum is
[46:23] not a US thing. So when you go through
[46:26] all these things, I just want to make
[46:27] sure you understand that you have to
[46:28] watch these things to start thinking
[46:30] this way. The same way that a year ago
[46:33] when I was talking to people about
[46:35] micron and memory and why we need all
[46:37] this and I was writing about humanoids
[46:38] and embodied AI and then I wrote a paper
[46:40] on inference in May. Crickets
[46:43] tokenization crickets right now. Nobody
[46:46] wants to spend the time on it. And in a
[46:48] week where crypto is exploding, where
[46:50] Salana's up 40% this month, the only
[46:53] thing people ask me about this week is
[46:56] gold. I kid you not. Of the macro people
[46:59] that are reaching out to me, the only
[47:01] thing I heard this week was gold. Nobody
[47:04] is talking about this stuff. So, if gold
[47:07] is the big trade and people are saying,
[47:09] "Well, I think it's going to go up." I
[47:10] do, too. Gold's not going to be the
[47:13] fastest horse in the race. I don't even
[47:14] think it's going to finish in in in the
[47:16] top hundred of the fastest things. And
[47:19] that's because this thing is a part of
[47:21] the economy going forward and gold will
[47:23] not be a part of that. Gold will not be
[47:26] a part of that. So let's just remember
[47:29] that a store of value is good but the
[47:32] store of value and the tokenization and
[47:34] the stable coin and again
[47:37] if I compare what ash said and then I go
[47:41] through where stripe is
[47:44] again you guys can read this on your own
[47:47] what he was talking about a year ago and
[47:49] what Stripe is doing right now in terms
[47:51] of the companies they're buying and what
[47:53] they're trying to do what they said in
[47:56] their sessions and in their public
[47:58] reports. All of this stuff is based on
[48:01] what Stripe has said. Now, this is what
[48:04] he was talking about a year ago. And
[48:06] again, the reason this is important, AI
[48:09] agents were a dream a year ago. They
[48:12] were not supposed to happen the way they
[48:13] did. So, this has been the year of AI
[48:15] agents. If you guys go back and read
[48:17] what I wrote about or go watch what I
[48:19] wrote about last year on all the videos
[48:20] which are on YouTube, I repeatedly said
[48:23] AI agents will be the story for 2026.
[48:26] consumer agents and AI agents will now
[48:28] be the story for 2027. The reason that's
[48:31] important is because we're now getting
[48:33] into the application side, which is the
[48:35] agents actually consuming and doing more
[48:37] traffic on the internet, doing more
[48:39] things, which is why I want you to use
[48:40] Grockbot. That is the reason why I do
[48:42] these videos so you can learn. Tom Lee,
[48:46] I don't know if because Tom Lee has this
[48:48] kind of look to him that people don't
[48:50] pay attention once he started to get
[48:52] more involved in crypto. Um, I've done
[48:55] John Gillan
[48:56] interviews before. I think I have
[48:58] another one coming up in September. This
[49:01] one is a good listen. Uh, I think it's
[49:03] important. You should go understand
[49:06] Tom Lee says a lot of, uh, smart things.
[49:10] Uh, he says he has a lot of experience
[49:12] on Wall Street obviously, but he
[49:14] understands Tradfi, but he says a lot of
[49:17] very key lines. And I say that because
[49:18] when I'm listening to someone and during
[49:21] an interview I pause and stop the button
[49:25] seven to 10 times and I go into AI and I
[49:28] have a conversation. It took me about
[49:30] double the time this morning before I
[49:32] went to go uh work out listening to this
[49:35] for a second time where I had to go
[49:37] remember all the things. So when you go
[49:40] through this and you realize what he's
[49:42] talking about and how AI agents change
[49:45] everything, how AI agents demand this,
[49:48] they cannot work without the smart
[49:50] contracts. They cannot work without all
[49:53] of this. AI forces finance to move from
[49:56] human precision to programmable
[49:58] constraint. And again, it's the same
[50:00] thing as AI agents force memory. It's
[50:03] the same exact thing, guys. the
[50:05] inference needs, the memory needs would
[50:07] not be as high if it wasn't for the
[50:09] advancements we made with Opus 4.5. It
[50:11] wouldn't have led Nvidia to then very
[50:14] quickly go out and purchase Grock.
[50:17] All of this stuff happened in the final
[50:19] quarter of last year. We are now seeing
[50:21] the Agentic side and the consumer side
[50:23] go and we have an administration that is
[50:25] trying to figure out a way to get out of
[50:27] the debt problem in the old system and
[50:30] they're betting on this stuff working
[50:32] again. Maybe it doesn't work, but it's
[50:34] part of the distribution. The Ethereum
[50:36] network, he goes through that. Um,
[50:38] Ethereum is not just a blockchain. It's
[50:40] more like the root settlement layer for
[50:41] a new financial operating system. Um,
[50:45] there's a lot of things in here that are
[50:48] important. And again, financial systems
[50:51] are rarely won by technology alone. They
[50:53] are one by liquidity, trust, standards,
[50:56] developer activity, institutional
[50:58] comfort, and path dependence. This is a
[51:01] critical part because people on the
[51:02] crypto side that are negative on
[51:03] Ethereum, they see other places building
[51:07] and going through their own approach. I
[51:11] completely agree with this having worked
[51:12] at Wall Street. You have to remember the
[51:14] old system is not going away. We need to
[51:16] find a way for AI agents to work, but we
[51:18] also need a way for humans to feel and
[51:22] this is institutions too. At least for
[51:24] the near term, liquidity is there, trust
[51:26] is there. Maybe eventually it won't be
[51:28] there but for the beginning start I
[51:30] absolutely believe that Ethereum will be
[51:32] the big benefit from here. So Ethereum's
[51:35] edge may be less about speed and more
[51:36] about trust at scale.
[51:40] Salana may represent 247 speed of
[51:42] markets. Bit may represent protection
[51:43] against time debasement. Ethereum in
[51:46] Lee's argument represents the
[51:47] institutional settlement layer where
[51:48] tokenized assets and activity become
[51:50] legally financially and socially usable.
[51:52] He obviously has a vested interest in
[51:54] this because he has a DAT uh on
[51:57] Ethereum. So, Bit Miners is something
[52:00] that I own and I own it because I
[52:03] believe in Ethereum and I believe this.
[52:05] Uh I've listened to Tom a bunch of times
[52:07] and I actually think Ethereum should
[52:09] outperform Bitcoin. I like the way the
[52:11] chart has looked all year when Bitcoin
[52:13] was down. So, I just bring this up as
[52:14] something to go through it and I get
[52:16] back to tokenization as the super cycle.
[52:19] Lee's point is that tokenization turns
[52:21] assets into software. Once an asset
[52:23] becomes software, it can move faster,
[52:25] settle faster, be divided into smaller
[52:26] pieces, be embedded into apps, be used
[52:28] as collateral, be traded globally, and
[52:29] interact with smart contracts. This
[52:30] changes an asset from a static financial
[52:32] claim to a programmable object. Again,
[52:35] there's so much richness in this in
[52:38] terms of things to understand and to go
[52:40] through. But a stock becomes software, a
[52:42] dollar becomes software, a bond becomes
[52:44] software, a royalty stream becomes
[52:46] software, a membership becomes software,
[52:47] reput you get it. It's when AIs enter
[52:50] the picture. So assets become
[52:53] programmable. That's the tokenization
[52:56] side. Then we go into 247. I gave a
[53:00] presentation at the New York Stock
[53:02] Exchange uh to
[53:05] uh I think all of the Korean major
[53:08] Korean brokers. Um, just to give you an
[53:11] idea of again where Asia is leading on
[53:13] this. AI agents become the users of
[53:15] those markets. So, you need them to
[53:17] become programmable. That's what we've
[53:19] been working on and getting towards. You
[53:22] need this, but then you need AI agents.
[53:23] This is the catalyst point. Okay? That's
[53:26] why I'm talking about this now. That's
[53:28] why as much as I love the infrastructure
[53:30] names to outperform the S&P, I don't
[53:33] think you're getting 10 and 20 baggers
[53:35] from those in a year anymore like the
[53:37] firework show is over. But I do think
[53:39] you can find those things in crypto. We
[53:41] have the tape speaking to us. We have
[53:43] the action point of AI agents. We have
[53:45] the administration saying we're not
[53:46] going to let yields go higher. Maybe
[53:49] they'll change their mind. Maybe Wars
[53:50] will raise 50 base points in September
[53:51] and crypto will go back to the lows. I
[53:54] just don't think that's the place, but
[53:55] it's obviously part of the distribution.
[53:57] Now, connecting this to Grockbot, Elon
[54:00] Musk, Grockbot promised, we will make
[54:02] you whole if AI loses your money. I
[54:04] think this is an super important story.
[54:06] Number one, I'm using Grockbot. Elon
[54:09] Musk backed a bold Grockbot promise on
[54:12] Wednesday. If the AI agent loses money
[54:15] while running an investor's bank
[54:16] account, XAI will make the user whole.
[54:19] He's trying to get people to use it with
[54:23] agents. This is the way it starts, guys.
[54:26] And this is why he wants to own this
[54:28] framework, which is why I want you guys
[54:30] to use it.
[54:32] [sighs and gasps] Tom talked about why
[54:33] this crypto winter is different. Okay.
[54:36] Why this one is different. Price was
[54:39] acting like crypto was dead while
[54:40] fundamentals were acting like crypto was
[54:42] getting ready for an institutional
[54:44] adoption. So this is his big change
[54:47] right now is that other times when
[54:49] crypto winters were occurring.
[54:52] There was no fundamental side. It was
[54:54] all a way down a future dream. The
[54:57] fundamental side is happening. The value
[54:59] acrruel is happening. We have public
[55:01] companies that are showing what their
[55:04] earnings are. All of this stuff has
[55:06] changed over the last two years, but the
[55:07] fundamentals have been improving all
[55:09] year while the prices have been have
[55:11] not. Part of that was because AI was a
[55:13] better place to put your money. Whatever
[55:15] the case is right now, remember software
[55:19] was getting destroyed in the first
[55:20] quarter of this year. Then it reached a
[55:23] point where
[55:25] no one was in it and everyone was in AI.
[55:28] Then we got the momentum unwind in July
[55:30] which got people to cover their shorts
[55:32] in software at the same time reduced
[55:35] their longs and semis. We're still
[55:37] seeing that play out. And guess what
[55:38] guys? Software's gone back up towards
[55:40] the highs. Now Bitcoin miners led it.
[55:43] Palunteer is part of that. Palanteer is
[55:45] part of the thematic uh portfolio. It's
[55:47] what I've written about. It's what I
[55:49] own. All of these things in here as far
[55:52] as I'm concerned are part of the
[55:53] rotation of looking for the next thing.
[55:55] This is part of the software trade. So
[55:58] past crypto winners ended because
[56:00] speculation returned. This one may be
[56:02] ending because the use case matured.
[56:04] There was no FTX. There was no the
[56:07] system is showing a much better place.
[56:09] And meme coins are not leading this one
[56:11] out. This is being led as I created my
[56:13] thematic portfolio for crypto by quality
[56:17] fundamentally driven ones. You got this.
[56:21] This week, Coinbase and Better Mortgage
[56:22] announced general availability of
[56:24] Bitcoin backed mortgages.
[56:26] I could show you guys time after time
[56:29] how many countries are embracing crypto.
[56:32] It's happening.
[56:34] Uh figure more good news. The tokenized
[56:39] portfolio again broke out. Here we go.
[56:43] Bitcoin again. I'm showing you this.
[56:44] This is since October one. So, the
[56:47] tokenized index which is outperforming
[56:48] so far this year. It's up on the air
[56:50] while Bitcoin is still down. This is
[56:52] meant to represent that Bitcoin is the
[56:54] store of value for this and they're
[56:55] exactly the same trade. Okay. So, when
[56:57] you're trying to figure out the
[56:58] fundamental story for Bitcoin, even for
[57:00] crypto people who don't think this way,
[57:03] at the end of the day, it should be
[57:05] incredibly correlated to the ecosystem.
[57:07] So, when the ecosystem is getting the
[57:09] money from the Tradfi world, Bitcoin is
[57:11] going to be going higher as well. If you
[57:13] just want to buy the proxy, you go buy
[57:15] Bitcoin. I don't see them separating.
[57:16] Here are the verticals. Not going to
[57:18] take you through all of them, uh, but
[57:20] you can see how they're broken down
[57:22] sector-wise. This is what they are
[57:24] monthtodate.
[57:26] This is the verticals monthto date. So
[57:29] big moves
[57:32] all where everyone's complaining about
[57:34] AI movements. Here are the individual
[57:36] names that have moved so far. What you
[57:38] can see Salana up 46%.
[57:43] Figure public company. hype. one drug
[57:47] mentioned uh in his uh quarter remember
[57:50] pure or purr purr which is owns hype is
[57:56] on there another big performer and then
[57:58] you've got circle
[58:01] nvidia reported
[58:04] blowout numbers uh next year the
[58:06] expectations were for 44% growth they
[58:09] said it's going to be 70
[58:12] um here's the pe now uh continuing to
[58:16] just move lower. Here's the earnings
[58:19] estimates post the announcement. This is
[58:21] for 2029. I just want you to see how
[58:24] much it's up. That's how much of a
[58:26] surprise it was to analysts.
[58:29] But as I've said, we're in the Nvidia
[58:32] mudstorm. So the stock trades up 9% one
[58:35] day and then the next day it goes right
[58:37] back down four to 5%. This is what's
[58:40] been happening in it. this line here
[58:42] with the chart.
[58:45] It's almost exactly where we closed
[58:46] today. I think it's up $3 from that.
[58:48] That was October.
[58:50] The earnings have gone up dramatically.
[58:52] So, just get used to the fact that this
[58:54] multiple compression story has been
[58:55] happening to Nvidia. It's going to
[58:58] happen to a lot of the AI semiconductor
[59:00] names. They are overowned in the hedge
[59:02] fund world. People have already made a
[59:04] lot of money. They're just not having a
[59:06] fun time buying things that aren't
[59:08] working on a momentum basis. That's the
[59:09] way momentum ends up unwinding and
[59:11] selling off. I'll show more about that
[59:13] in a little bit. S&P for the week, as
[59:15] much as it was seemed like a crazy week,
[59:17] 40 basis point change. Same thing for
[59:20] the Q's. Russell was down 1.4%.
[59:24] The thematic portfolio was down 1.4%.
[59:29] Um, this is the breath analysis that
[59:31] I've shown you guys. This is what I'm
[59:33] waiting for for us to be out of the mud.
[59:35] I don't think we're going to be out of
[59:36] the mud till maybe the midterms. That
[59:39] would be my guess. I think we're gonna
[59:40] be in the mud. That doesn't mean you
[59:42] can't find a name that won't go up a
[59:43] decent amount. Pure Storage went up a
[59:45] decent amount. Marll went up from 180 to
[59:50] 240 and then pull back after good
[59:52] earnings report and after saying that
[59:54] things are going to be much better than
[59:55] what people expect going forward. But
[59:56] that's that's all different. Here's what
[59:58] we need. It's the 50-day moving average,
[60:00] not the 20-day moving average.
[60:02] Structurally 50-day moving average. We
[60:04] are still only have 29% above the
[60:07] 50-day. This need name this needs to
[60:09] start expanding. The slope of the 50-day
[60:11] is turning down. We only have 36 there.
[60:13] The 200 day still massively positive.
[60:16] We're 25% off the highs. That's the
[60:18] biggest amount we've been off the highs
[60:20] except for the situational awareness
[60:22] lows. So, we are getting some
[60:26] improvement in terms of the range, the
[60:29] volatility range. This is a good first
[60:32] step because the V is coming down. So
[60:35] this is the average volatility range
[60:39] for all of the names. So we are making
[60:42] some progress. Again, this is mud. So we
[60:44] bounced out of here. Ken Griffin buys a
[60:47] bunch. We go up and now we're basically
[60:49] no different in price than we were after
[60:51] the first gap day. We're sitting here. I
[60:53] expect a more of this. And let's assume
[60:56] this goes in the midterms. That's out
[60:57] here. It will be frustrating, guys.
[61:00] It'll be frustrating.
[61:03] Um, momentum, this is tech momentum. Has
[61:05] not stopped going down yet. Made new
[61:07] lows. Industrial momentum went back
[61:09] down. This part of the reason you're
[61:10] going through it. Remember, momentum
[61:12] does not mean that the AI trade is not
[61:13] working. It means that
[61:16] the long side is not moving. It's in
[61:18] mud. But the short side continues to go
[61:20] higher. That is what's driving this.
[61:21] Now, so software is going higher.
[61:24] The longs are not moving. So, what you
[61:27] get is the V's coming down. I mean,
[61:29] here's the 15-day V on tech momentum.
[61:33] The 60-day peak, but it's not going fast
[61:35] enough yet. Uh, factor neutral V still
[61:39] not moving. It's not coming down fast
[61:41] enough. It's going to take a little
[61:42] while. I think that's going to stay
[61:44] there. Bitcoin, you had this massive
[61:47] week. All those Sigma numbers, great
[61:50] consolidation week. Fantastic. Ethereum,
[61:53] a bigger week, over 30% consolidation.
[61:56] Salana even better up over 30 finished a
[62:02] week up eight. Salana seems to have some
[62:04] tokenization love going on right now. If
[62:07] you want to see something amazing in
[62:09] February semiconductors before they had
[62:12] had their big rally just to show you how
[62:13] quickly
[62:16] crypto can win.
[62:18] Salana relative to SMH
[62:21] is now back where it was in the early
[62:24] parts of February, guys. After that
[62:26] entire bare market, it takes one month
[62:29] for it to make up everything. That's how
[62:32] you can find five to 10 baggers with
[62:34] inside crypto. Here is the software
[62:36] thing I talked about. This is the UBS
[62:39] software index. This is not a cap
[62:41] weighted one. This is equate. This was
[62:42] the highest correlation with Bitcoin.
[62:45] again looks pretty good to me. Uh
[62:49] credit, we've got triple C's widening
[62:51] out. This is what I talked about before.
[62:52] I think the crowding out and the
[62:54] movement on all this stuff in yields and
[62:55] the CDS going out, you're going to see
[62:57] crappy companies that are not going to
[62:59] be able to survive. The junk spreads,
[63:02] they're hanging in there, not doing
[63:03] much, but the triple C's are widening.
[63:06] Just something to watch. Uh high yield
[63:09] CDS, nothing. Oh no, no. These are the
[63:13] disclosures that I mentioned at the
[63:15] beginning. I'll start including these at
[63:16] the end. You guys can watch them and
[63:18] read them on your own. Uh that's it for
[63:20] this week. It's good to be back.
[63:22] Although the humidity is horrible. Uh
[63:24] I'll be seeing you guys at events that
[63:27] I'll be speaking at. I'm sure in
[63:28] September and October, some in DC. I'll
[63:31] be at an event in New York in September.
[63:34] And then I think Pomp's event is uh in
[63:37] October. I look forward to seeing
[63:38] everyone there. Uh reach out and uh I'll
[63:41] see you guys soon. Bye.