Your Capex is My Opportunity: The AI Capex Cycle Has Replaced the Old Economy Business Cycle

[0:00] market continues to rally u and despite
[0:07] still more calls for bubbles. I was at a
[0:10] an event in LA and in DC and in LA I did
[0:15] hear the bubble talk uh and I even heard
[0:17] people say that it's obvious that it's a
[0:20] bubble. So we'll go through um how the
[0:24] market's being driven basically only by
[0:27] AI. uh but that is driving earnings
[0:30] higher, revisions higher, margins to the
[0:32] upside and we're in the early stages of
[0:35] this. GDP is stronger on the back of AI
[0:38] and semiconductors are basically leading
[0:41] everything. And when you went go through
[0:43] the earnings reports, it's not just how
[0:45] strong things are. There are now
[0:46] trillions in backlog uh for future
[0:49] growth. So again, this is the early
[0:51] innings of the AI situation. Now we
[0:54] start to get into the warning signs that
[0:56] are showing up. We are going to have
[0:58] despite earnings growing, GDP going
[1:00] higher, there's always negatives that
[1:02] show up. In the beginning of the year,
[1:04] we had the negatives of software, which
[1:06] those stocks are still sitting near the
[1:08] lows of the year. Private equity, those
[1:10] stocks are still sitting at least close
[1:12] to the lows of the year. We still have
[1:13] financials under the 200 day moving
[1:15] average, which is downward. So, don't
[1:17] forget the negatives from AI will show
[1:19] up again. And now we have a lot of the
[1:23] next year's uh earnings already built in
[1:25] with the pees rising. I'll show some of
[1:27] that in stocks like Caterpillar.
[1:30] And we've got inflation and rates
[1:32] flashing early warning signs uh for
[1:34] subscribers. I did a webinar this week
[1:37] and just so you guys hear it, I uh I on
[1:40] the website I do have the turbulence
[1:42] model for the week. We finally saw our
[1:45] first uh signal in turbulence. Uh it's
[1:48] not a trigger point yet. It's just an
[1:50] early warning sign, but that's because
[1:52] we're starting to see a breakdown in
[1:54] some correlations across assets and also
[1:56] the breath is starting to uh break down.
[1:59] I go through the five layer cake
[2:01] framework and how you guys should be
[2:02] thinking about it. uh for those of you
[2:05] who are working with me on uh your
[2:08] portfolios and how to navigate through
[2:11] this particularly uh some of the wealth
[2:13] channels and the RAAS. Uh I'm starting
[2:17] to release stuff called benchmark
[2:19] arbitrage. Uh it's an opportunity for
[2:22] you guys to think about how you can
[2:24] navigate through this and still make the
[2:25] money because the benchmarks are
[2:27] completely wrong for the AI world. They
[2:29] are still built around a world of
[2:31] software and partly the industrial and
[2:33] the consumer-based economy which is gone
[2:36] right now because this is being driven
[2:37] by AI and it's going to stay that way
[2:40] because we're not creating jobs as
[2:41] nominal GDP is growing higher. So the
[2:44] traditional framework between
[2:45] manufacturing and service or
[2:48] manufacturing and the consumer has
[2:50] broken down and a lot of the stocks and
[2:52] a lot of the business cycles that you
[2:53] guys have been associated with. They're
[2:55] just not in there. So, I'll go through
[2:57] the AI business cycle and then I'll talk
[3:00] about chemicals uh and highlight. I did
[3:02] a spotlight report on one of the
[3:04] chemicals, but I'm also going to give
[3:05] you more details. There's a new report
[3:07] up and then I'll go through Bitcoin and
[3:10] Ethereum. So, here's the S&P all-time
[3:13] highs. Uh, no divergences. We've got the
[3:17] NASDAQ or the NDX, the Q's also just a
[3:21] violent move up to new all-time highs.
[3:23] Small caps new all-time highs for the
[3:26] month. the S&P one of the best months in
[3:28] the last 25 years. Uh basically it looks
[3:31] like uh number four in the last 25
[3:34] years.
[3:36] Uh I highlighted the growing risk that
[3:38] whenever we do have a recession, you get
[3:40] the six-month rate of change negative as
[3:43] the first warning signal. Uh we've now
[3:46] obviously gone back higher and we're now
[3:48] up 5%. The one thing I would say is uh
[3:50] historically when you've broken down to
[3:52] levels like this and then you get back
[3:53] above the market's usually fine. Uh so
[3:57] these are kind of midcycle warnings and
[4:00] I think that's where we are because I
[4:01] don't think we're going to have a change
[4:03] in the earnings growth and I don't think
[4:05] we're going to have a change any of this
[4:07] stuff uh related to AI just because of
[4:10] what I'll go through today. Uh Mike
[4:13] Wilson put this out. I think this was a
[4:15] just a good thing. Forward earnings in
[4:17] the S&P. Look at look at this
[4:19] acceleration. I mean this is not a small
[4:22] move. This is the trillions of dollars
[4:25] going into AI that is driving this
[4:27] particularly stocks like Micron and
[4:31] other semiconductors but also
[4:32] Caterpillar and things like that. Even
[4:34] the hyperscalers contributing to it.
[4:37] Median stock no different. Small cap no
[4:40] different. So you've got year-over-year
[4:41] earnings forward earnings continue to
[4:43] move higher. Warren Pies put this out
[4:45] and just said estimate growing faster
[4:47] than they did in the mid 90s or late
[4:50] internet bubbles. The boom is unique
[4:52] because it is not coming off an EPS draw
[4:55] down. And that's really the reality is
[4:56] that normally when we get this kind of
[4:58] acceleration, it's coming out of a
[5:00] recession where you've had earnings down
[5:02] and we've had job uh losses. This is all
[5:06] happening because of the capex boom
[5:08] that's going on which I'll go through.
[5:10] And over the last 15 years, analysts
[5:12] have typically revised EPS down by about
[5:14] 2%. This is the normal average over the
[5:18] 2011 to 25 period of EPS revisions as we
[5:21] go through the year. This is what's
[5:23] happened this time. So this is the
[5:25] reason why stocks are going higher.
[5:27] Margins continue to go higher. And if
[5:30] you want to look at it again, margins,
[5:32] same thing. Um, forward up here
[5:36] trailing. I mean, we're just talking
[5:38] about something that hasn't happened
[5:39] before. revisions which had started to
[5:42] eek down. This is overlaid with the PMI.
[5:45] We got the PMI number. It went sideways.
[5:47] Didn't shoot higher. Uh but I'll show
[5:50] some other things on this. But the
[5:51] revisions, this is uh this white line
[5:54] here is the 10-week average, but this
[5:56] was the most recent week. We're back
[5:59] right up. So, PMIs are going to remain
[6:02] strong. France put this out this week in
[6:04] LinkedIn and just highlighted that the
[6:06] regionals are very strong. We didn't get
[6:08] a follow-through. This is the new
[6:09] orders. category. We didn't get anything
[6:11] uh on the PMI, but the the breadth of
[6:14] the regionals is very strong. We got
[6:16] capital goods uh the core capital goods
[6:19] year-over-year just exploding higher
[6:21] associated again with a PMI close to 60
[6:24] and here are the PMIs. So, capital goods
[6:27] continue to be there. I still have a lot
[6:29] of people that are fading the economy
[6:31] and that's because they're focusing on
[6:33] things related to the consumer. Uh
[6:36] here's Caterpillar. I don't know how you
[6:38] look at arguably the most important
[6:40] industrial stock and not see that this
[6:43] is doing the exact same thing as what we
[6:45] just looked at. And this really started
[6:46] after liberation day. But to be honest
[6:48] with you, it started in the second half
[6:50] of the year. And as I go through things,
[6:51] this coincides with when DRAM went
[6:53] higher. This is when the data center
[6:55] buildout really started to kick in. It
[6:57] is still early in this guys and it's
[7:00] going to continue. uh in their earnings
[7:02] report. We ended the quarter with a
[7:03] record backlog with excellent visibility
[7:06] into demand over the coming quarters.
[7:08] Backlog now stands at 62 billion. I'm
[7:10] going to show you backlogs for the
[7:11] hyperscalers, but I think when you're
[7:13] getting into Caterpillar and you're
[7:14] dealing with this kind of demand, you
[7:16] can just take it another level. Strong
[7:18] demand for large power generation
[7:20] equipment used in data centers
[7:22] supporting AI was a contributor. The AI
[7:24] boom is spurring additional
[7:26] construction. We now expect sales of our
[7:28] power generation equipment to be roughly
[7:29] three times larger by 2030.
[7:33] We are closely monitoring macroecon. So
[7:35] the reason I want to bring this up guys,
[7:37] historically if you were seeing these
[7:38] types of things and this is what you're
[7:40] going to see people talk about, well
[7:41] this is a problem and we're going to
[7:42] have overbuild and all of these things
[7:45] that happen. Um which traditionally I
[7:48] would say make sense and that's why
[7:49] Caterpillar the orange line here is
[7:51] their PE ratio
[7:54] forward. So, they're trading at 36 times
[7:57] next year's earnings. The green line
[7:59] here, though, is Nvidia. So, Nvidia is
[8:02] at 10-year lows in their forward PE. The
[8:06] reason I want to bring this up is
[8:08] historically, if you get one of these
[8:10] scenarios up here, it's either coming
[8:12] out of a recession, uh, or it's some,
[8:15] this is after oil went down to 10 and
[8:18] then you start to move lower. I don't
[8:19] think this is going to happen this time.
[8:21] I think multiples are going to stay at
[8:22] higher levels for what they just said
[8:24] which is this demand is going to
[8:26] continue into 2030 and that's because
[8:27] it's being funded by the revenues that
[8:30] are happening with inside AI which you
[8:34] have to remember everyone who said we'll
[8:36] get no revenues and now the backlog for
[8:38] three hyperscalers is 1.3 trillion and
[8:41] we're running at an annualized run rate
[8:43] of 30 billion for
[8:46] uh for Anthropic and most estimates are
[8:49] 50 to 100 by the end of the year. The
[8:52] Cosby one of the important signs of this
[8:54] whole thing mainly because they have the
[8:57] semis uh in terms of memory but they
[8:59] have tons of electronic side in there
[9:01] and they also have the heavy equipment
[9:03] which I showed. So the Cosby is directly
[9:05] related to this buildout. There's the
[9:08] machinery component. So it's not just
[9:10] semis of the Cosby
[9:13] Hyperscalers made new high. This is
[9:15] equal weight again. It's kind of lagging
[9:17] and it's just up here, but Google had a
[9:19] move 10% move higher. Meta had a 10%
[9:21] move lower. I still believe shorting the
[9:24] hyperscalers as a way and being long all
[9:26] of the thematic portfolio I have is
[9:29] still going to work out extremely well
[9:30] going forward. Uh because that's the
[9:33] spenders and then on the other side you
[9:35] have the buyers now or the receivers.
[9:38] Here are the numbers for the cloud
[9:39] backlog.
[9:41] 400, 460, 400. Uh you're dealing with
[9:46] $1.3 trillion of backlog for these three
[9:49] companies, Amazon, Google, and
[9:51] Microsoft.
[9:53] That's revenue. Those are contracted
[9:56] dollars that right now they can't get
[9:59] because they don't have the capacity.
[10:02] And that's where the issue is right now
[10:03] is the shortage. Uh hyperscalers are
[10:06] absolutely crushing it right now. and it
[10:09] goes through all the different
[10:10] components. I'm only bringing that up
[10:11] again because it's very hard if you're
[10:13] going to have on the one side all the
[10:15] semis and all the the receivers going,
[10:17] but at the same point you have the
[10:18] hyperscaler showing that adoption is
[10:20] happening at a much faster pace than
[10:22] people anticipated. It's very hard for
[10:24] the market not to have strength when
[10:25] these components are doing well. The
[10:28] buildout is uh the beta. The
[10:30] hyperscalers are just kind of market
[10:33] performers this year. They're actually
[10:34] underperforming by a couple percent when
[10:36] you equal weight them. Uh so we're just
[10:38] in that scenario. I did do a market
[10:40] update video this week that the
[10:42] subscribers saw which went through just
[10:45] the warning signs, a lot of which uh fit
[10:47] in with what I'm going to show here,
[10:48] which is we had a breakout in unled gas
[10:52] futures. So this is this week. So if you
[10:54] haven't been paying attention, I know
[10:55] Iran's move to the back burner. Forget
[10:57] about it being about Iran and just think
[10:59] about it being a signal for inflation, a
[11:01] signal for future growth coming down.
[11:03] Those are the two things that
[11:04] historically will impact stocks. When
[11:06] you add in the fact that we've had a
[11:07] massive rally and people are very much
[11:10] in the market, you end up in the exact
[11:12] situ uh situation that normally causes
[11:14] an issue because now people are in the
[11:17] market and it might not be all
[11:18] institutions but you do have a lot of
[11:21] people in this is and again I showed you
[11:24] gas futures that was six month or
[11:26] September. This is September as well. So
[11:29] it's not just that we had a move higher
[11:31] in energy. it's that the back months are
[11:34] starting to go up, which means this
[11:37] problem is going to last longer, which
[11:38] means the inflation component, I mean,
[11:40] just look at where this is. Go back to
[11:42] the gas side. You're talking about now a
[11:44] significant move that if we're in
[11:46] September and we've got prices up around
[11:48] $87,
[11:50] you've got a bigger issue because now
[11:52] you're going to have this year-over-year
[11:53] CPI.
[11:55] And again, I've talked about it being
[11:56] above 4%. I think the risks are growing
[11:59] the longer the straight remains closed
[12:00] that this gets up even higher. Here is
[12:03] uh gas at the pump which shot higher
[12:05] last week. And again, this is a a a
[12:09] straight line higher. And yes, it does
[12:11] come at a time when we're seeing a
[12:12] straight line higher in profit margins.
[12:15] So that's where the offset is. But
[12:17] again, this reminds me a lot when you
[12:19] have nominal GDP going higher and
[12:21] inflation going higher. This is very
[12:24] 1970ish in terms of that side. Uh we're
[12:27] not creating jobs. We've got no job
[12:30] creation. We'll see what happens when we
[12:31] get into stuff next week because it does
[12:33] look like we could have uh a surprising
[12:37] number in the payroll side. If for no
[12:39] other reason, just the claims data plus
[12:41] the ADP weekly. We're not seeing that in
[12:43] the manufacturing PMIs. Their employment
[12:45] component went lower. Uh but it's just
[12:48] something to keep on. Diesel prices
[12:50] sitting near the high. So this is diesel
[12:52] prices across the country.
[12:56] the inventory side is is now at that
[12:58] point that people warned about and so
[13:00] all the energy side this is from Goldman
[13:03] Sachs is basically saying under the
[13:04] three scenarios of when it starts we're
[13:08] past late April uh so let's go to miday
[13:11] this is where they're saying the global
[13:12] inventories would be are likely to reach
[13:14] all-time lows even if the straight opens
[13:16] by late May um to show you how the
[13:20] market is not prepared for this I wanted
[13:22] to use Chevron uh on the weekly call I
[13:25] did mention that one of the places that
[13:27] I'm starting to add is in the chemicals
[13:29] and the energy which I believe will are
[13:32] are lower beta uh parts of this and the
[13:35] reason is because they're not part of
[13:37] the buildout. They're kind of midcycle
[13:39] to late cycle. I'll show that in terms
[13:41] of the AI side. The early cycle is
[13:43] really the semis, the optical fibers and
[13:44] all the stuff that went through it. But
[13:46] let's just take this as what it is.
[13:49] You've got this when oil was down here.
[13:53] Uh, and again, we're using September oil
[13:55] prices, which matters the most for the
[13:57] energy stocks. You were around 66 bucks.
[14:00] Well, now you're up at almost 90.
[14:05] Chevron's basically unchanged from where
[14:06] we were before the war. So, just
[14:09] remember that. Um, there's been a lot of
[14:11] movement. Oil's gone higher, but the
[14:13] energy stocks haven't. And again, I
[14:15] think that's because right now the focus
[14:17] for everyone trying to make money is on
[14:19] the stuff that's working on the long
[14:20] side and chasing all of these names
[14:22] which are related to the semiconductors,
[14:24] the optical fiber. All of the names that
[14:27] are in my thematic portfolio just
[14:28] continue to rip higher to all-time highs
[14:31] and that's what's driving the market.
[14:32] And I think the earning side has been a
[14:34] lot of the enthusiasm. But now that
[14:36] we're through the bulk of it, I think
[14:38] maybe you should worry about the things
[14:40] that I mentioned. And like I said, I am
[14:42] now getting more turbulence uh signals.
[14:45] If I get any intraweek this week, I will
[14:47] put them up. If I get any signal that
[14:49] says that just like we had in early
[14:51] February that we've hit a trigger point
[14:53] across assets where deleveraging starts,
[14:55] I will send that out as a special video
[14:57] update. Um airlines would be an obvious
[15:00] candidate for something to go on. And
[15:02] remember, they've fallen off and they're
[15:04] sitting here. So, kind of the opposite
[15:05] of the Chevron thing. Uh the airlines
[15:08] would be the number one risk. you start
[15:10] to see things and BFA put out this week
[15:12] that air travel travel has slowed
[15:14] substantially.
[15:15] Uh we'll see where this goes. But again,
[15:18] when you start getting into the price
[15:19] moves in the inflation side, you should
[15:20] see a demand response. We got the uh PMI
[15:24] this week and here was the prices paid.
[15:26] I mean, we're basically in stratosphere
[15:28] levels associated with, you know, oil
[15:31] was up at 155 during here before the
[15:34] great financial crisis and then
[15:36] obviously we had the shutdowns and
[15:37] before that. It's very hard to find
[15:39] times where you get prices paid in there
[15:41] at these levels. Uh China,
[15:45] there's all kinds of stress going across
[15:47] there. Exporters are raising prices on
[15:49] swimsuits, skiuits. Basically, all the
[15:52] Chinese exports are starting to show it.
[15:53] Fuel inflation's about to hit the
[15:54] kitchen table. We've got crop prices
[15:56] that are breaking out. So, the futures
[15:59] are finally starting to move. their late
[16:01] cycle of the inflation side related to
[16:03] oil as the fertilizer prices are higher
[16:06] and then the crop prices eventually go
[16:08] higher. So you should start to see that
[16:10] um this will be the next uh I'm always
[16:12] looking for the next fear-mongering
[16:14] thing of you know scaring people and so
[16:18] now we've got not only the fertilizer
[16:19] issue from the gas but they're
[16:21] forecasting the strongest El Nino in 150
[16:25] years with droughts and all kinds of
[16:28] things. If we get any of that, just
[16:30] remember again, you're looking for
[16:31] things with inside the market that could
[16:33] have an issue and cause the next wave.
[16:35] And I do think this is the 1970s, so I
[16:37] think you have to focus on the
[16:39] government is shoving money in the
[16:41] system. We're growing. And even though
[16:43] we aren't creating jobs, if we start
[16:45] creating jobs, then I think they have to
[16:47] build rate hikes in. As of now, we've
[16:50] started to build in some rate hikes or
[16:51] at least one into next year uh after
[16:54] this week. But I think we have to pay
[16:56] attention. the inflation swaps. This was
[16:59] another part of the warning part of the
[17:00] video. I mean, we're we're moving
[17:02] higher. Um, last time we were at these
[17:04] levels, we had CPI
[17:07] kind of uh where you'd associate up here
[17:10] the fears of us getting back in the four
[17:12] or five handle. So, I think the
[17:13] inflation swaps are going higher and
[17:15] long-term yields, they're contained in
[17:17] the US. This chart just looks like
[17:19] eventually it's going to break higher.
[17:20] Um, 5%. Michael Hartnik came out and
[17:23] said that's where he expects uh problems
[17:25] to be. The government's been trying to
[17:26] keep this below 5%. Obviously, you've
[17:29] got the housing issue. If you get above
[17:30] five, well, right now, 30-year yields
[17:32] are sitting just below five and 10ear
[17:34] yields still sitting in this triangle
[17:37] formation. Uh, again, I think you have
[17:39] to watch this. That would be a risk that
[17:41] would come and we already have it
[17:43] spreading. So, this is 10-year JGB
[17:45] yields, which have broken out to new
[17:47] highs. Uh, we also have guilt. So,
[17:50] 30-year UK yields have broken out to new
[17:53] highs. Uh,
[17:56] the breath is bad. This is one
[17:58] measurement. Goldman Sachs warned about
[18:00] this. I think it's a good one. The
[18:01] median stock is currently still 13%
[18:04] below its 52- week high. Uh, there are
[18:08] about 50% of the S&P stocks that are
[18:10] that are down on the year. So, despite
[18:12] the all-time highs, it's not being seen
[18:15] with a lot of contribution. Uh this was
[18:19] an interesting one just because the
[18:21] NASDAQ 100 closed up more than more than
[18:24] half a percent. Fewer of the 29% of the
[18:27] stocks uh ended up being up last time
[18:31] that happened has not happened at least
[18:32] 20 years. These are all from Jason Gford
[18:35] and I think he does a good job of just
[18:37] highlighting when you get these issues
[18:38] of breath. On Monday the S&P closed at a
[18:41] record high. The next day at least 1%
[18:43] more stocks hit a 52- week low than
[18:45] high. Uh that is also not normal. just
[18:47] shows it's being led by very few
[18:50] companies and more importantly it's
[18:52] probably being led by a lot of small and
[18:53] midcap names uh in terms of what you're
[18:56] seeing on the screen and inside the S&P
[18:59] it is a lot of the names that are
[19:01] related to the semis and uh
[19:03] infrastructure side financials as I
[19:05] mentioned still below they've tried to
[19:08] work higher but they've stayed below
[19:11] and I just want to bring this up because
[19:13] this is where I think the issue is going
[19:15] to be before the the end of the year. Uh
[19:18] Elon Musk warns the US could soon be
[19:20] producing more chips than we can turn on
[19:22] and China doesn't have the same issue.
[19:24] So he said this um at the World Economic
[19:26] Forum at the beginning of the year. He
[19:28] said it on moonshots. I've reported it
[19:30] twice. This is becoming more and more of
[19:33] a reality. So early cycle, you hoard and
[19:37] buy as many chips as you possibly can in
[19:41] expectation of what you're going to
[19:42] need. And as the compute is ramping up,
[19:46] everyone's buying more because there's
[19:48] more compute that's needed. But again,
[19:50] we don't have the compute right now. So,
[19:52] we're starting to run into issues and
[19:55] semis have just accelerated. So, I just
[19:58] wanted to make sure that we get into
[19:59] this point of where we are. Uh, if
[20:02] you're not invested in semiconductors,
[20:03] they are now the biggest
[20:06] level two gick in the S&P at 10
[20:08] trillion.
[20:10] uh media, software, tech,
[20:14] diversified financials, and again, this
[20:15] is level two, so it strips it out. And
[20:18] this is why when you're looking at tech
[20:19] and how tech is driving everything, it's
[20:21] not tech that's driving everything, it's
[20:23] semis, which happen to be in tech when
[20:25] you have a level one gick. When you
[20:27] separate them out, semis are the biggest
[20:29] contributor. And if you don't have a big
[20:31] waiting in them and again if you equal
[20:33] weight this it's even bigger because of
[20:35] that 10 trillion about half of it is
[20:37] Nvidia which is one of the worst
[20:39] performing of the semis this year. So
[20:42] this is being driven by the other semis.
[20:44] This is something uh I presented at the
[20:48] event in California. So the prior 15
[20:51] years was about your margin is my
[20:53] opportunity. The next 15 years is about
[20:56] your capex is my opportunity. And this
[20:59] is the five layer cake that Jensen Yuang
[21:01] speaks about for the $90 trillion. This
[21:04] is why this won't stop, guys. It's being
[21:06] funded by these companies that were the
[21:09] B that made all the money. So everyone,
[21:11] Meta, Tesla, Amazon, Google, just go
[21:15] through the list. They are the ones
[21:16] funding this buildout, $90 trillion.
[21:20] Now, it's also going to involve Morgan
[21:22] Stanley, Goldman Sachs, Eli Liy, Merc,
[21:24] all of the big companies that need AI
[21:27] are also going to be funding this
[21:28] because they're going to need their own
[21:29] data centers as well. That's why the 90
[21:32] trillion is just the beginning. We've
[21:34] only started this. So, this is where you
[21:36] want your portfolio to look like. The
[21:38] problem is the way the portfolios are
[21:40] set up, these are the heaviest weights,
[21:42] the spenders. So for you RAAS that are
[21:45] looking for what you need to do, this
[21:47] needs to be a higher waiting than what's
[21:49] actually in the waiting. The business
[21:52] cycle has changed. So this is what
[21:54] everyone looks at. This is what had
[21:56] historically happened. What you're
[21:58] seeing right now is it doesn't matter if
[22:00] oil goes higher. It doesn't matter if
[22:02] the Fed doesn't cut rates, which had
[22:04] been expected. It doesn't matter if jobs
[22:07] aren't being created. The money is still
[22:09] being spent. And there's a new cycle
[22:11] that's happening. The early cycle is the
[22:12] semiconductors. The midcycle is the
[22:14] power, the data centers. We have to get
[22:16] this done. So, you're starting to see
[22:18] the transports go higher. And that's
[22:20] because now we're finally in the
[22:21] buildout stage. They've hoarded the
[22:22] semiconductors. And as Elon Musk said,
[22:25] if this doesn't come in, we got a
[22:27] problem. Well, once we finally do start
[22:29] getting this built out and we actually
[22:30] can meet some of those RPOs, which is
[22:32] going to take another year, two years,
[22:35] then we'll start getting into the late
[22:37] cycle, which will be the humanoids. This
[22:40] is when AI will be the strongest and
[22:41] this when we will start solving
[22:43] problems. This is why I'm writing a
[22:45] paper on the benchmark is late. The time
[22:48] where this matters is when I talk about
[22:50] the supersonic tsunami. When something
[22:54] changes rapidly, you end up with an
[22:57] obvious benchmark arbitrage. So
[22:58] benchmark arbitrage for traders ends up
[23:01] being the ad deletes uh around the
[23:03] Russell or the S&P. But this one is
[23:06] different. The benchmark is completely
[23:09] waited improperly for the future. Um, a
[23:12] lot of the winners back here, if we go
[23:15] back into the Dan Ies and Kathy Wood of
[23:18] what they've brought to the retail and
[23:20] the RAIA world, a lot of that has the
[23:22] software companies in there because
[23:24] those were the winners of the past
[23:25] cycle. This is a nontech story. That
[23:30] thing I showed you, semiconductors,
[23:32] those are not pure tech. We have they're
[23:34] hardware. We've got the hardware names.
[23:36] We have the commodities. All of this is
[23:38] a different story. Chemicals, those are
[23:41] not thought of as AI names, but they
[23:43] are. So again, here's the breakdown of
[23:45] the way that I did the five layer cake
[23:47] specifically for this. You have the
[23:48] rack. So up here, the rack, the pack
[23:52] advanced packaging is the best one. And
[23:54] then uh you've got the optical. So these
[23:58] three are the early cycle. This is what
[23:59] was hoarded. You had the deals with Meta
[24:02] buys six billion of Corning's fiber.
[24:06] You've got all these deals that Jensen
[24:08] has done with both optical fiber,
[24:10] coherent, light, but also Marll. All of
[24:13] these different places are the semi
[24:15] side. Then when you move down here, this
[24:17] is the chemicals. So chemicals have had
[24:19] a good run. And I'm going to go through
[24:21] the chemicals in more detail because
[24:22] they're a big part of this on both the
[24:25] tubing side for the fiber and also the
[24:28] advanced packaging. The issue is these
[24:30] guys are more midcycle. They're a little
[24:31] bit more steady. So this is lower vault.
[24:33] And then you have the power side. The
[24:35] power side does include Exxon and
[24:36] Chevron and Bloom Energy, but it also
[24:38] includes a bunch of different components
[24:41] and then some things that I think are
[24:42] going to be brought into the mix related
[24:43] to batteries. So, those are the reasons
[24:46] why I've chosen the five. And here's the
[24:47] S&P. So, if you're overweight these by
[24:50] even 10%
[24:52] and you're reducing the S&P by 10%.
[24:55] You're getting an outperformance. And
[24:57] this is what people need to do. This is
[24:58] why I'm starting to do swaps with
[25:00] clients so that they have the ability to
[25:02] invest in this stuff. This is just an
[25:04] important thing and the reason that I've
[25:05] set up those thematic portfolios for you
[25:07] guys. Um here is all of them as a
[25:10] thematic uh the thematic basket as a
[25:12] portfolio and I just wanted to highlight
[25:14] the white line here is the S&P. The
[25:16] yellow line here is the hyperscalers.
[25:18] They've still underperformed the S&P and
[25:20] then you have software. This is the way
[25:22] the world is weighted today. That's why
[25:24] the S&P's d is dragged down here. A lot
[25:26] of these names are in the midcap, small
[25:28] cap and overseas. That's why you've seen
[25:30] the US underperform international over
[25:33] the course of the last uh 16 months.
[25:36] Uh I've been pounding the table over and
[25:38] over again as this AI agent mega trend
[25:41] moves. So again, when you're looking at
[25:42] these parabolic moves, including the
[25:45] hyperscalers uh and anthropic, it's all
[25:49] because of AI agents. This all changed.
[25:51] And again, this is why on the one side,
[25:53] why did software get killed? Because of
[25:55] Opus 4.5. Why did all of a sudden the
[25:57] adoption start to increase? because of
[25:59] Opra's 4.5 because that's when the
[26:00] agentic world took over. This is not
[26:02] some speculation. This is triggered by
[26:04] an actual event and that is the reason
[26:06] why my thematic portfolio has gone
[26:08] straight up because it is geared towards
[26:10] the agent world. The thing I will say is
[26:12] that has just started. So, are we going
[26:15] to have corrections and will it stay
[26:16] this parabolic? No, we'll have
[26:18] corrections of course. The issue is for
[26:20] people that are looking for corrections,
[26:22] these numbers are just going to
[26:23] accelerate. We barely have that much
[26:25] adoption happening. It's going to
[26:27] increase dramatically. And if you think
[26:29] about where Anthropic is today and
[26:30] they're forecasting 50 to 100 billion by
[26:33] the end of the year and they're at 30,
[26:36] that's a compounding number. We are
[26:38] going to continue to have excess demand
[26:41] because as OpenAI said this week,
[26:43] there's not going to be enough compute
[26:44] in the world to meet the demand. So
[26:46] think about all of the people, Michael
[26:48] Bur, the people that wrote all of these
[26:51] papers about how it's a bubble. Just go
[26:53] back and think about it. That was six
[26:55] months ago. So, how could people be
[26:57] invested in here? I still hear people
[26:59] saying they've seen a bubble. They know
[27:00] what a bubble looks like. This is a
[27:01] bubble. Open eyes revenue chief says,
[27:03] "Enterprise business is accelerating." I
[27:06] have used chat GPT 5.5 more than Claude
[27:10] for this entire week while I was in LA.
[27:12] It is a phenomenal model. All of the
[27:15] visuals and the images you're seeing
[27:16] were done by by GPT5. All the new ones.
[27:20] I only used Nano Banana for about nine
[27:23] months. I have not used Nano Banana once
[27:26] because this imaging is off the charts
[27:28] compared to that and this is by far the
[27:30] smartest the most intelligent by far
[27:34] model that I've used. Um Thomas Curran
[27:38] the CEO of Google's cloud Google cloud
[27:41] regarding compute I think for the next
[27:42] 10 years there will always be more
[27:44] demand than supply.
[27:46] Take a picture of it save it before you
[27:49] go to bed every night read it again.
[27:51] This way you don't make the mistake of
[27:53] listening to the people who will
[27:55] continually tell you that there's a
[27:56] bubble being thrown. Some White House
[27:58] officials also worried that Enthropic
[28:00] wouldn't have enough access to compute
[28:03] to serve that many more entities. So the
[28:05] question that's becoming there when you
[28:07] see the benchmarks of 5.5 it's pretty
[28:10] much right on Mythos. So the question is
[28:13] was this advertising for Mythos? Was
[28:16] this a slowdown because they didn't have
[28:18] enough compute and they wanted to get
[28:21] some positive press out of this to say
[28:23] they're I'm going to go with the latter
[28:26] for sure. Uh why Google and Amazon are
[28:29] betting billions on anthropic. These
[28:31] deals are as much about compute and
[28:32] infrastructure as they are about
[28:33] capital. I don't think they have enough
[28:35] compute for what's going on. Uh just
[28:37] talked to a lab with a war chest balance
[28:39] sheet. They are actively cont actively
[28:41] contract compute across all major clouds
[28:43] in Neos. No one has any material
[28:45] capacity coming online between now and
[28:48] September. Hourly compute pricing is
[28:50] going much higher in the coming 5
[28:52] months. The cost of compute is far
[28:55] beyond the cost of the employees. Nvidia
[28:57] executive says right now, this was this
[28:58] week.
[29:00] I was quoted a couple times in the
[29:01] Atlantic article. Here we go.
[29:04] We went from AI is a bubble to there is
[29:07] not enough compute in the data centers
[29:09] in less than six months.
[29:11] It's agents. Uh here's the chart of
[29:14] those RPOS again. Just look where it
[29:16] started. Opus 4.5.
[29:20] Looking ahead, the strong results
[29:22] reinforce our conviction to invest the
[29:24] capital required to continue. Anyone who
[29:26] is worried that the capex won't come.
[29:30] I just showed you how much they are
[29:32] behind. The capex is not only coming, it
[29:36] will continually be revised higher.
[29:40] Capex consensus. Just looking at all of
[29:43] these numbers,
[29:45] if you think for a second we're near the
[29:47] end of this, um I will say I overheard
[29:50] commentary uh with people hedge funds at
[29:54] this event, major pension event, and
[29:56] just listening to the conversations, the
[29:58] number one reason that this is an
[30:00] overbuild now, is because magically
[30:03] we're going to come up with some
[30:06] efficiency gain that will allow I I
[30:10] don't know how you can look at these
[30:11] charts and not realize there's no such
[30:13] thing as an efficiency game which will
[30:15] stop this. In fact, what we have had
[30:18] only over the last three years is
[30:20] efficiency gains. That's all we've had.
[30:23] We've been able to use the compute we
[30:25] have and get the progress faster. And
[30:27] now the adoption is kicking in. And the
[30:29] adoption is kicking in because the
[30:31] agents are here. The agencies far more.
[30:34] So we're going to be constrained. So go
[30:36] back to the benchmark arbitrage.
[30:38] you're if you're thinking that this is
[30:40] an overbuild and you're using history,
[30:42] then you're using the industrial
[30:43] consumer cycle to make a decision on
[30:45] what's going on. And that includes
[30:47] the.com bubble, guys. The.com bubble was
[30:50] a quarter of a century ago
[30:53] before the iPhone. Do not listen to the
[30:57] bubbleheads.
[30:59] Severe shortages of Taiwan semis. Most
[31:01] advanced processes are pushing an allout
[31:03] effort to build blah blah blah. Taiwan
[31:05] semi one of the constraints. If you guys
[31:07] haven't drawn the map on it, you got
[31:09] three end results of what we need. We've
[31:12] got Intel, we've got Samsung, and we've
[31:14] got Taiwan Semi. Uh Elon Musk basically
[31:17] bought up all of Samsung's production a
[31:19] year ago, or not even a year ago, and he
[31:21] now is out there doing terapab because
[31:23] he realizes we're not going to have
[31:25] compute ever. Again, I can say it over
[31:27] and over again. We are not catching up.
[31:29] Semis are going to be needed. This is a
[31:32] completely unique situation now.
[31:36] Nvidia's stock has been able to go
[31:37] higher while I just want to highlight
[31:40] how much the multiples have gone down.
[31:42] So, at the end of this month, we'll get
[31:43] theirs. But I would say at this point,
[31:46] no one believes the compute story
[31:48] because Nvidia has gotten cheaper and
[31:50] cheaper and cheaper. This week, the
[31:54] White House admitted we are in trouble
[31:55] with compute. So they're putting in
[31:58] section 303 of the Defense Production
[32:00] Act to help with grid infrastructure,
[32:02] equipment, and supply chain capacity
[32:04] because we are falling behind China and
[32:06] we don't have enough compute. They don't
[32:07] want to use anthropic because we don't
[32:09] have enough compute. Uh US spending on
[32:12] power plant equipment expected to triple
[32:14] through 2030s. Again, power is the base
[32:18] of the five layer cake. Uh the Compass
[32:21] data center is pulling out of Virginia.
[32:25] We still have bottlenecks going on that
[32:27] are preventing this and it's going to be
[32:29] an election issue in the midterms.
[32:30] Voters do not want AI and they do not
[32:33] want data centers. Uh I released this
[32:36] this week on the power of semiconductors
[32:38] as the next scarcity trade.
[32:44] It was a reminder that this is the
[32:45] report I put up last week into the
[32:47] subscriber level about the edge as we're
[32:50] getting into and that power semis were
[32:52] in there. You got six companies uh that
[32:54] I highlighted in there. Uh this came
[32:57] out. Lead times for power semiconductors
[32:59] have become extremely stretched because
[33:00] of AI servers. Again, the reason that
[33:03] the power semiconductors are so
[33:06] important is because they fit both
[33:09] inside the edge side and the power side.
[33:12] This is really important from the
[33:14] efficiency, from the thermal, all of
[33:15] these different components. And here's a
[33:18] chart of those six names. um
[33:23] that I put out in the power semis that
[33:24] are in that report. So, there's two
[33:26] things I want you to look at. Yes, it's
[33:27] gone completely parabolic over the
[33:29] course of the last two weeks. But
[33:30] secondly, it just broke out of a 5-year
[33:33] high just now. So, massive up week
[33:37] again, but again, it just broke out.
[33:41] Now, here's chemicals again, midcycle,
[33:45] late cycle, whatever you want. Um, this
[33:48] is the European chemicals overlaid with
[33:50] one of those power semis. Texas
[33:52] Instruments which just broke out. If I
[33:53] showed you the the power semis, it would
[33:55] look very similar. Chemicals will go up
[33:59] with power semis. They fit into the
[34:01] exact same side. So, as a reminder
[34:03] again, back to the themes. Here are the
[34:06] the power names. Here are the chemical
[34:07] names. You want to be focused on the
[34:10] late cycle part. Uh, this is a report I
[34:13] just put up uh this weekend. You guys
[34:15] can see it already. Uh, the chemical
[34:18] layer of AI. I go heavily into the 17
[34:21] names, which I did as a thematic
[34:23] portfolio, but now I go one by one just
[34:26] like I did for the Kentucky Derby
[34:27] handicapping. Uh, and this breaks them
[34:30] down by exposure scores for early, mid,
[34:32] late timing, and the investment logic
[34:35] behind both. Again, these are not
[34:38] suggestions on which ones to buy. Same
[34:40] thing as when I release a spotlight on
[34:43] Integris. This is not a timing thing,
[34:45] but these are ones that to me are going
[34:47] to benefit the most over the course of
[34:49] the next few years. I'm not trading
[34:51] these things, but I'm buying them. I
[34:53] bought Integris this week after this
[34:55] report came out after the earnings
[34:57] knocked the stock down. Uh I bought the
[35:00] stock at that point and I will continue
[35:02] to look for particular names in there.
[35:04] Camores is another name within there
[35:06] which I've said to many of you I thought
[35:08] would double this year. Well, we're
[35:10] almost there. Uh I think we're up 50.
[35:12] Now, we might be close to double now at
[35:14] this point, but uh I still think it has
[35:16] a ways to go. All of these names on the
[35:19] chemical side are involved and you've
[35:21] got advanced packaging and you have the
[35:24] optical fiber and the tubing side and
[35:26] the polymers. My argument on this stuff
[35:28] and what I'm writing a report on is on
[35:30] the energy intensity for chemicals at
[35:33] this stage of AI instead of oil. Oil
[35:36] does not have the same importance in
[35:38] today's world for a variety of reasons.
[35:40] But one of the major things is how
[35:41] chemicals to me are going to become far
[35:43] more important because the size of the
[35:45] terra fabon
[35:47] Musk wrote out the amount of chemicals
[35:50] needed that is not an oil trade. So when
[35:52] you're building these things you're
[35:54] taking a lot. So the chemical side to me
[35:56] is a major theme that will be in a
[35:57] long-term bull market. Compound and
[36:00] friends had a great interview this week
[36:01] with my boy Adam
[36:04] uh Parker and uh I thought he did a
[36:07] great job. I just want to highlight some
[36:09] of the things that he said.
[36:12] He basically said here's the sequence he
[36:14] laid out from the opening software
[36:16] discussion. So first the market
[36:18] compresses the multiple. That's what we
[36:19] did. The market cuts the valuation first
[36:21] then the fundamentals start
[36:23] disappointing. What usually happens next
[36:24] is an earnings mess. He says analysts
[36:28] still have models that assume things
[36:29] like roughly 80% gross margins similar
[36:32] margins in 27 28 29 very little
[36:34] structural change and profitability. His
[36:36] argument that those are too static for
[36:38] an AI disrupted world after earnings
[36:40] misses come sales misses. Multiples
[36:43] contract then earnings miss then
[36:44] eventually sales miss. Why would earning
[36:46] why earnings would break first margin
[36:49] pressure before revenue pressure? His
[36:52] reasoning is that software companies may
[36:53] need to spend more to stay competitive
[36:55] in AI. Um, regardless of your view on
[36:58] this, you have to really start to
[37:00] remember this gets back into that
[37:01] benchmark arbitrage. If you're long the
[37:04] S&P 500, you are long Salesforce.com and
[37:06] Adobe and all of these companies as a
[37:09] much higher weight than the future says
[37:11] they should be. If for any reason these
[37:13] things were disrupted to the manner they
[37:15] are, you would see a very, very large
[37:18] fall. I didn't include this in this
[37:20] week's video, but Novel Ravocant, who I
[37:22] really enjoy. He has a podcast, I
[37:24] believe it's novel, um, he talked about
[37:27] Apple in a very similar way. He was very
[37:29] negative on Apple, and he basically
[37:31] talked about a lot of the things that
[37:32] have to do with terminal value. We are
[37:34] at a massive AI disruptive phase. The
[37:37] certainty is that we will need more
[37:38] power. We will need more optical fiber.
[37:41] We will need more compute.
[37:44] We don't need software companies the way
[37:46] we did. We don't need a company like
[37:49] Apple for the infrastructure as this
[37:51] changes unless they can find a way to go
[37:53] through this. Novel was very strong in
[37:55] this. I find him to be one of the most
[37:57] clear-headed thinkers on any of this,
[37:59] not making uh bombastic comments. You
[38:01] should go listen to it. I thought it was
[38:03] very good. Um here's IGV versus the S&P.
[38:06] So if this was like last year out of
[38:09] Liberation Day, you would have seen
[38:12] software trade with its normal beta. So
[38:14] here it is the collapse. everyone who
[38:15] told you we got to go in here, we got to
[38:17] buy these. This is wrong. Every single
[38:20] tech person, and I'm not going to go
[38:22] through the names anymore, that said
[38:24] this is a mispricing, this is panic.
[38:26] Well, now you've had a fall, you've had
[38:28] a rally. Adam Parker's right in my
[38:31] opinion.
[38:33] Uh what he said on semis for Adam
[38:35] Parker, he thinks investors
[38:36] underestimate how much estimates were
[38:38] wrong. And again, that's the point.
[38:39] That's the only way you can have Nvidia
[38:42] at the multiples it is. and Micron
[38:44] Micron the multiples were just wrong. So
[38:47] the fact that it still trades at a 5p
[38:49] off next year's earnings and you've had
[38:50] a massive beat, Micron's too cheap. And
[38:52] he makes his argument specifically on
[38:54] this. He thinks Micron is a perfect
[38:56] example of why it makes sense. Something
[38:58] like four and a half times next year's
[39:00] earnings. The product is in shortage.
[39:02] Earnings are being revised sharply
[39:04] higher. Micron is not a meme behavior.
[39:06] It is the market repricing to a much
[39:08] stronger earnings path. Again, benchmark
[39:10] arbitrage. Do not listen to the bubble
[39:12] talk. Bitcoin to finish off. Um, all
[39:16] right. The MACD. We're despite it being
[39:19] boring and having really no movement.
[39:22] Remember, we're just getting these MACD
[39:24] weeklies. We're just breaking through
[39:26] some resistance. We've had now one week,
[39:29] two week, 3 week, four-w week, and then
[39:32] this week kind of finished around where
[39:35] we are. Um, meaning we're up at uh we're
[39:38] getting close. We're unchanged from last
[39:40] week. We are trying to move higher and
[39:42] we're coming from a level that normally
[39:44] would be explosive when it finally does
[39:46] go. Uh Anthony Pompiano on our weekly
[39:49] this week did mention something in
[39:51] talking to someone about Korea. Korea
[39:53] which is a major trading hub and
[39:56] especially for Bitcoin. Uh this seems to
[39:58] be an issue. So this is the Asia growth
[40:01] story and basically this is Bitcoin
[40:04] overlaid with it. Right now everyone in
[40:06] Korea and Asia is trading the AI trade.
[40:08] Bitcoin has lost interest, but the good
[40:11] thing is it's found a way to move from
[40:13] 60,000 up to 80,000. Well, that's a 33%
[40:17] rally from the lows. So, it is moving.
[40:20] Uh, it's doing better than software.
[40:22] It's doing better than the Asian growth
[40:23] story. If for any reason we start to see
[40:25] any turn in these things, maybe we can
[40:28] get it higher. But then again, like I've
[40:30] said, my trigger point is the fact that
[40:31] I think CPI is going above three-month
[40:33] yields. The Fed's not going to be able
[40:35] to uh raise raise rates or cut rates,
[40:39] whatever you want. We're going to be
[40:40] stuck. You're going to have a new Fed
[40:42] chair who's coming into the most
[40:44] dislocated situation. You got PAL
[40:46] staying on board. This just argues for
[40:48] just chaos going forward and a period
[40:51] where they're going to run it hot with
[40:52] rates at lower levels and inflation up
[40:54] at higher levels. We get anything on the
[40:56] jobs picture that moves uh long-term
[41:00] rates higher. Short rates can't go
[41:02] higher. The Fed's not raising rates, so
[41:04] we'll see what happens. Uh, but I would
[41:06] keep it in. Clarity Act, some positive
[41:08] news. Also, that could be a very big
[41:10] catalyst for this. What really needs to
[41:13] happen is retail needs to just jump on
[41:14] board. And if they do jump on board, go
[41:17] through. I mentioned that I bought Micro
[41:19] Strategy calls uh into the end of the
[41:21] year. I wasn't targeting short term
[41:24] because I think this could take a little
[41:25] bit longer, and I'm targeting when
[41:26] inflation gets above three-month yields.
[41:29] But PMIs are moving higher. three-month
[41:31] yields are stuck where they are. The
[41:34] Fed's not move. I think all of these
[41:35] things gone. Um on that same podcast
[41:37] with Adam Parker, uh the other person
[41:43] I I brought this up just because I
[41:45] thought it was really interesting that
[41:47] you're starting to see a change and I
[41:49] believe that Ethereum is the main
[41:51] important story for the rest of the
[41:53] year. If the Clarity Act does go through
[41:54] and as I show you, there are positive
[41:56] developments that are happening on
[41:58] stable coins and you're starting to get
[42:00] more and more people talk about Ethereum
[42:02] because
[42:04] there is a way to value it. The usage of
[42:07] the network is what creates value. All
[42:09] of these things in here, programmable,
[42:13] so you get into the coding side.
[42:14] Everything with Ethereum gives people a
[42:17] chance to use discounted cash flow. And
[42:18] in a world where growth assets are dead,
[42:21] I will say this to RAAS again. And I
[42:23] will say this to the mutual funds that
[42:24] watch this. You guys that are
[42:26] benchmarked,
[42:28] Ethereum and Bitcoin need to be part of
[42:30] your portfolio and they need to be part
[42:31] of your growth asset basket uh or
[42:33] allocation. Uh, I wrote this in Substack
[42:36] about killing the float, the pro
[42:38] beginning of programmable, programmable
[42:41] money and the new financial guard rails
[42:43] where this goes through the Door Dash
[42:45] decision
[42:47] uh, and their decision to basically take
[42:49] stable coins in and what this means and
[42:51] how this is different than the legacy
[42:53] system which took forever and how the
[42:56] new codebase system of programmable
[42:58] money has started. Here's the Ethereum
[43:01] chart. We keep running into this 200
[43:04] week moving average. Uh we've broken the
[43:07] downward trend line, but until we again
[43:09] I've said it before, until we're above
[43:11] 2400, in this case that line is 2456. I
[43:14] want two closes above that level. Uh
[43:17] we've got a buy signal down there, but
[43:18] right now we haven't gone anywhere. Uh
[43:22] and that's it for this week, guys. So, a
[43:24] lot of stuff in there. Uh thanks to
[43:27] everyone uh who saw me in LA in DC. I
[43:30] appreciate it. For those of you who want
[43:32] more details on any of the stuff here,
[43:35] uh just go to the website for uh 22V.
[43:38] And if you want, if you always look in
[43:40] YouTube, I put the link in there that
[43:42] you go directly to the website and see
[43:44] some of the stuff on the subscriber
[43:45] lists. See you next week.