The Edge Awakens: Why the Physical AI Upgrade Is Still in the First Inning [vGemma]

[0:00] Uh,
[0:01] I will be
[0:02] flying out Friday. So, these There's
[0:06] going to There's not I'm not going to
[0:07] recap the week since I won't be here for
[0:10] the the end of it despite where things
[0:12] are trading this morning. Uh, but this
[0:14] is an important one in my opinion for
[0:17] you guys because I am going to spend
[0:18] time on some themes that I've talked
[0:20] about but try to put them into context
[0:22] for the most part around one
[0:25] or two podcasts that I listened to this
[0:27] week. One with Craig Fuller, one with
[0:30] Dylan Patel. That I know I give you guys
[0:34] the podcast info. You can read the
[0:35] summary. I give you the YouTube side,
[0:37] but I highly recommend listening at
[0:39] least to the Craig Fuller one. The Dylan
[0:42] Patel one, if you've listened to other
[0:44] ones in the past, it's very similar. So,
[0:46] but this is really about the physical
[0:48] upgrade is here now. So, that was the um
[0:53] the theme of my my piece for this year
[0:57] was the physical world upgrade, which
[1:00] was all about leaving this world of the
[1:03] cloud, software, the hyperscalers,
[1:07] them spending money and needing
[1:10] according to Jensen Huang, 90 trillion
[1:13] dollars to upgrade the entire physical
[1:15] world with artificial intelligence. And
[1:19] we started that process early. It's very
[1:21] semi-heavy. It's power-heavy. It's
[1:24] optical fiber-heavy. That's what the
[1:25] themes are in my thematic portfolios
[1:29] with all the names. And by the way,
[1:30] thank you to everyone who has reached
[1:32] out, especially after last week when I
[1:34] went through a lot of the different
[1:35] details
[1:36] so that you guys can be involved in
[1:38] these names. They
[1:40] the thematic basket continues to just
[1:42] rip. I'll go through some of the slides,
[1:44] obviously, but this theme is in the very
[1:47] early innings. So, when you see these
[1:48] parabolic charts, yes, there's going to
[1:51] be pullbacks and they will be severe.
[1:53] But, if you continue to waste your time
[1:56] trying to pick bottoms in software, if
[1:59] you continue to waste your time trying
[2:01] to wonder when When is Nvidia going to
[2:03] outperform Marvell? Or when is it going
[2:05] to outperform Texas Instruments? Or when
[2:07] is it going to outperform Intel? It's
[2:08] not going to happen, guys. Um, this side
[2:11] is so massive. The concentration of the
[2:14] cloud, which was a 17-year trend, and
[2:16] then the IQ raising part of the
[2:19] pre-training of GPU, this stuff is so
[2:22] underinvested. And I through the the
[2:25] webinar I did this week for the
[2:27] subscribers, I highlighted that the
[2:29] semiconductors are now the largest
[2:32] weight at a level two gig in the S&P 500
[2:35] at 17% and growing.
[2:38] That is not a bubble. Last year at this
[2:41] time, software was bigger than semis.
[2:44] Software's gone down, semis have gone.
[2:45] And at 17%, that's about 10 trillion
[2:48] dollars. But of the 10 trillion,
[2:51] Nvidia,
[2:53] Broadcom, and Micron make up 7.5
[2:56] trillion. You are going to see the other
[2:58] names in there grow rapidly over the
[3:00] course of the next at least five years,
[3:02] but probably decades. So, as I go
[3:03] through this,
[3:04] really spend the time on making sure
[3:07] that you're prepared for this because
[3:09] the numbers are massive and the money is
[3:11] coming from the Mag 7. So, if you
[3:15] haven't read this, and again, if you
[3:16] listened to the Dwarkesh podcast, Jensen
[3:19] again reiterated the five-layer cake. If
[3:22] you didn't know what the five-layer cake
[3:24] was cuz I had some people reach out.
[3:26] March 10th, this is on a blog on Nvidia
[3:28] where he goes through
[3:30] This is where we are right now, guys.
[3:33] This is it. There's five layers to this
[3:34] with energy and chips as the base of
[3:37] this.
[3:38] This is where these three is where all
[3:41] of my attention is on making money. The
[3:43] models and the applications are more for
[3:44] the agentic side in terms of the use,
[3:46] the profit margins, the digital biology,
[3:49] the robot All of this stuff, but this is
[3:52] where the money's being made. And again,
[3:54] this is all underweight cuz this is
[3:55] industrial.
[3:58] He said it would be 90 trillion. 90
[4:01] trillion dollars again, the global
[4:02] economy is about 120 trillion. That is a
[4:05] massive infusion, and that is why when
[4:08] you go through and look at some of these
[4:10] semiconductor charts, and I'm going to
[4:12] highlight some more today that have just
[4:13] started, you start to realize why this
[4:16] is going on. And that is why I created a
[4:19] thematic basket, which is all related to
[4:21] this movement. So,
[4:23] here is my thematic basket now up 34%
[4:26] since November 24th. The reason I chose
[4:29] November 24th is cuz that's when Opus
[4:31] 4.5 was released, and that was when
[4:33] software started to truly get disrupted
[4:35] because we entered the agentic age. This
[4:37] entire basket is the agentic age. That's
[4:40] what it's for. It's for the buildout,
[4:41] it's for the infrastructure, it's for
[4:43] everything that you're going to need.
[4:44] These lines here, well, this is
[4:46] software, still down now 16%. It had a
[4:49] nice bounce. It pulled back. I still, if
[4:52] I have to guess, these lows will hold
[4:53] for the IGF. I don't think it will hold
[4:56] for all software names, and I want to
[4:58] make sure I I make that clear. I do not
[5:00] want to own a single SaaS software. I
[5:02] think you are wasting your time trying
[5:04] to spend time on software playing for
[5:05] this mean reversion. There is no mean
[5:07] reversion. Software is a dead thing in
[5:10] terms of built-on code. There will be
[5:12] some names that can survive, but just
[5:14] like Ford has survived for 35 years
[5:16] without the price going up. I think it's
[5:18] a waste of time to try and think that
[5:19] there's winners. Now, the problem with
[5:21] that is the S&P 500 is in here. Up 4%,
[5:25] 5% for the year. Heavily weighted into
[5:26] that. The white line is the Dan Ives
[5:29] ETF. Like I said, love Dan, got to know
[5:32] him in the last month.
[5:35] Unfortunately, a lot of the names he has
[5:37] in there are related to the web 1.0
[5:40] playbook, the 2.0 playbook, the software
[5:44] names. And again, that just acts as a
[5:45] dead weight just like for the S&P. The
[5:48] hyperscalers right here, dead weight. I
[5:50] want to use those as a funding vehicle.
[5:52] They're the spenders. You want to be
[5:53] long semis, long the infrastructure
[5:55] side, and you want to be short the
[5:56] hyperscalers. It is going to be very
[5:58] similar. If we have a problem where we
[6:00] run out of compute, the hyperscalers are
[6:02] going to get hurt as well because
[6:03] they're depending on the ROIC and the
[6:05] adoption. These are directly linked.
[6:07] They need to spend money on the compute
[6:09] side, the semis will have 30%
[6:11] corrections just like we saw with Micron
[6:13] from 460 down to 330, but then Micron
[6:16] went right back up. They won't always go
[6:18] up that fast, and I'm sure we're getting
[6:20] closer to a time where they start to
[6:21] sell off just because
[6:24] when Marvell goes from 95 when I write a
[6:26] paper,
[6:27] it closes that day at 87, and pre-market
[6:30] today it was at 170. Those types of
[6:32] moves are building in a lot of what is
[6:34] going to happen over the next year in
[6:36] earnings.
[6:37] Um, just as a reminder, and as we get
[6:39] into this industrial shakeout, this is
[6:40] the reason why I wrote this paper back
[6:43] in July. So, this is not a new theme for
[6:45] me. You can go read what it is, but
[6:47] basically,
[6:48] this was about not being like
[6:51] traditional demand cycles around real
[6:52] estate and autos, but by a new and
[6:54] powerful force, paranoia, across both
[6:57] governments and corporations, a
[6:58] deepening fear of falling behind in the
[7:00] global AI arms race. That's what is
[7:02] driving PMIs. I still read tons of
[7:06] people focused on real estate and autos
[7:08] and saying that we're going to have an
[7:10] issue because that's where growth is.
[7:12] That is not where growth is. And if you
[7:14] want to go find out, listen to this
[7:17] Craig Fuller
[7:19] interview on Adam Taggart Thoughtful
[7:21] Money back in November. This was before
[7:25] Opus 4.5 came out. And his thing, Craig
[7:28] is a freight transport expert. He was
[7:32] saying the real economy is very, very,
[7:34] very weak. And he was very bearish. And
[7:36] just to go through, he was describing
[7:38] that freight is a leading indicator is a
[7:40] better real-time leading indicator. The
[7:43] goods economy is much weaker than the
[7:44] headline data suggests. Current CapEx
[7:46] wave is still mostly hope-driven.
[7:50] Now,
[7:51] I love when you get smart people who are
[7:54] wrong,
[7:56] and they turn around and they do the
[7:58] exact same podcast five months later.
[8:01] Miracle turnaround. The US industrial
[8:03] economy is now booming despite high oil
[8:06] prices. And that's one of the things I
[8:08] want to make sure you guys recognize. If
[8:09] you've been listening to the oil
[8:11] doomers, we obviously are going to have
[8:13] an impact in inflation, which I'll go
[8:15] through. But in terms of the global
[8:17] economy stopping, AI is more important.
[8:19] I said it last year with tariffs. I said
[8:21] AI is the only story. AI is the only
[8:24] story. It is the only story story you
[8:26] will need. It's the reason why I'm doing
[8:28] what I'm doing and trying to bring you
[8:30] guys information that is not only
[8:32] real-time. There's a lot of information
[8:34] in this week's video that is actionable
[8:36] because it just happened this week. So,
[8:38] just keep that in mind is that it's
[8:40] moving so fast that you can't depend on
[8:42] sell-side research to stay on top of
[8:43] stuff. You have to be on the podcast.
[8:45] You have to be listening to Jensen
[8:46] Huang. You have to be going through X.
[8:48] And you have to be reporting it. So,
[8:50] here are the highlights. I'm not going
[8:51] to go through everything, but he said
[8:53] November, he flipped from November to
[8:56] now as bullish as he's been in years.
[8:58] The goods economy is now roaring. Rail
[9:00] shipments up. Chemical shipments at
[9:02] record highs. Remember, chemical moat is
[9:04] part of this. I'm going to go through
[9:06] chemicals again in this because that is
[9:07] a late cycle part of this. You are going
[9:09] to see chemicals. You need to be
[9:11] involved in petrochemicals, particularly
[9:13] US ones or ones that are going to
[9:15] benefit from lower natural gas prices
[9:16] versus the rest of the world.
[9:19] Truck tonnage at a three-year high.
[9:21] Truck loading postings at strongest
[9:23] level. I've read people that said that
[9:24] this was all about Avis in terms of
[9:26] transports. You have to stop listening
[9:28] to permabears and people that get caught
[9:30] in this situation. People like Craig
[9:32] Fuller that were bullish in November,
[9:34] and I could give you a list of people
[9:35] that were bearish in November that have
[9:37] not changed their view. They just move
[9:38] on to the next thing.
[9:40] Bonus depreciation is helping. Cheap
[9:42] natural gas is helping. It's no longer
[9:44] being led by imports but by domestic
[9:47] production. It means the middle of the
[9:48] country is what's driving GDP, and the
[9:51] coastlines are suffering. Keep that in
[9:53] mind when you're reading the surveys.
[9:56] Nothing in the freight data suggesting
[9:58] the war is hurting industrial demand.
[10:02] Insensitive, non-cyclical AI has the
[10:06] capital coming from
[10:08] free cash flow, and even if that free
[10:09] cash flow goes negative, these companies
[10:11] have massive debt they have massive
[10:13] equity, so debt to equity will rise.
[10:15] It's not good for them. That's why I say
[10:17] be short the hyperscalers, all of them.
[10:19] I don't care which one it is, including
[10:20] Google.
[10:22] Will they go up? I'm sure they will go
[10:23] up. Are they going to go up as much as
[10:25] the S&P? Not in my opinion, because
[10:27] semis are going to drive the S&P
[10:28] performance. Optical fiber is going to
[10:30] drive the S&P performance. Caterpillar
[10:32] is going to drive the S&P. You have to
[10:34] be long industrials, materials, energy,
[10:36] and semis. This is the 1970s again and
[10:38] again and again. This is a physical
[10:39] upgrade.
[10:40] Flatbed rejections hit the highest
[10:42] levels ever recorded.
[10:45] So, flat they have no capacity. There is
[10:48] zero slack.
[10:50] So, again, if someone has told you that
[10:52] trucking is not happening and this is
[10:54] fake, they're just wrong. And if you
[10:57] want actual factual data, here are the
[10:59] imports of capital goods relative to
[11:03] consumer goods. So, I want you to think
[11:05] about what this is. This is, "Hey, we
[11:07] need tons of semi, we need tons of uh
[11:10] uh of parts for the data center. We need
[11:13] things from Switzerland. We need gas
[11:15] turbines. We need all of this stuff."
[11:17] This is This is back to 1991, guys. This
[11:20] is the ratio of them. This is all about
[11:23] the opposite of what was going on from
[11:26] this period, where there was no capital
[11:28] goods relative to the consumer goods. We
[11:31] are not buying furniture. We are not
[11:33] buying things cuz there's no housing
[11:34] market at this point.
[11:36] Deal with We'll deal with that later,
[11:37] but there's confirmation that the
[11:39] industrial boom is happening. And if you
[11:40] don't believe that data,
[11:42] and you don't believe that um oil isn't
[11:45] hurting the economy, how about we go to
[11:47] Asia? Taiwan machinery, all-time highs.
[11:50] Kospi machinery,
[11:53] all-time highs. And then obviously you
[11:55] have the S&P machinery. They're all
[11:56] moving together.
[11:58] This week, United Rentals, a good proxy
[12:00] for not just the industrial economy, but
[12:03] also when the turnaround is starting. We
[12:05] had the largest day uh on Wednesday for
[12:08] United Rentals.
[12:10] This is in coming out of COVID. This is
[12:13] coming out of the worst of
[12:16] the housing GFC time, and this is coming
[12:19] out of 1998 with LTCM.
[12:24] Yes, Avis. Here we go. This is
[12:26] year-to-date. Avis is up. Who cares? The
[12:29] entire group is up. And this is after
[12:31] Avis had collapsed 70%. This is not
[12:34] about one stock, and don't let people
[12:37] uh
[12:38] try to direct you uh misdirect you with
[12:41] the bearishness.
[12:42] Prices are moving higher. There's no
[12:44] other way to say this. Um I will keep
[12:46] saying it. We are going to get CPI above
[12:48] 4%. If this holds, this little thing
[12:50] here, we're going to be above it very
[12:52] soon.
[12:53] When we get CPI above four, I don't
[12:55] think the Fed is raising rates, but that
[12:56] will be above three-month bills. So, we
[12:58] will have negative real yields again. I
[12:59] think that's an important story, but the
[13:01] most important part, this is from this
[13:03] week's S&P
[13:04] uh data. This includes the service PMIs
[13:07] and the manufacturing. This is not just
[13:09] manufacturing. It will spread a little
[13:11] bit into the service side, because
[13:13] technology is going higher. So,
[13:16] remember, semiconductors are in
[13:17] everything. When you walk into a
[13:19] doctor's office and you're getting a
[13:20] service done, if they need new
[13:22] computers, if they need new machinery,
[13:23] if they need new MRIs, it's all being
[13:25] impacted by commodities and especially
[13:27] by semiconductors. Here is the services
[13:29] part of the uh PMI. This is the ISM PMI
[13:35] for the service side. And again, I just
[13:36] want to highlight, we ripped to the
[13:38] highest level since 2022,
[13:40] when year-over-year CPI was up around 5
[13:42] and 1/2%. Here you can see the long-term
[13:44] trend of year-over-year CPI.
[13:47] Again, I find it hard to believe we
[13:48] won't get above 4%. There's fiscal
[13:50] stimulus that just continues to show up.
[13:52] Warren Pies put this out and basically
[13:54] said,
[13:55] "Tax receipts
[13:57] much lower
[13:59] than expected. So, we are seeing tax
[14:02] refunds tracking higher versus last
[14:04] year.
[14:05] Tax received for the government's much
[14:07] lower, so we're getting a fiscal push
[14:11] from the one big beautiful bill that's
[14:13] coming at the same time the gas at the
[14:15] pump has gone higher. So, for people
[14:16] that getting refunds that are much
[14:18] bigger,
[14:19] those are probably the same people that
[14:21] live paycheck to paycheck that are being
[14:22] offset by the oil prices. So, that's one
[14:25] of the reasons why we're not some seeing
[14:26] some big collapse. The other one is, as
[14:28] I've highlighted in previous videos,
[14:30] we are an exporter of energy, so it
[14:32] benefits the economy. Here's the raw
[14:34] industrial uh
[14:36] uh commodities. If the Iran situation
[14:38] was having an impact, you'd see the
[14:40] metals going down. They're not going
[14:41] down.
[14:43] In fact, Robert Friedland uh from
[14:45] Ivanhoe, the entire copper industry just
[14:47] returned from last week's leading copper
[14:49] conference,
[14:51] and basically
[14:53] uh saying
[14:54] they're even more bullish that all-time
[14:56] copper highs could be tested in the
[14:57] coming weeks. Uh there's all kinds of
[15:00] uncertainties. There's all kinds of
[15:01] bottlenecks. Uh if you want to get a
[15:03] sense as to how much people are
[15:04] investing now in trying to avoid the
[15:07] energy situation, uh solar batteries,
[15:11] EVs, through the roof from China in
[15:13] terms of exports.
[15:15] If you want to know what China is
[15:16] buying, remember, silver is part of my
[15:18] uh
[15:19] thematic uh portfolio.
[15:22] Uh
[15:22] it's lagging here, which means it's a
[15:24] great entry point in my opinion uh based
[15:27] on what I see and have written about on
[15:28] silver. Well, China just imported the
[15:30] most silver ever in March. And remember,
[15:34] they stopped exports December 31st.
[15:38] Surprising move here. I think this has
[15:40] implications if you're looking for
[15:42] something that could slow everything
[15:43] down.
[15:44] Uh this is the ADP weekly employment.
[15:47] We've had four weeks in a row now of
[15:49] kind of moving higher.
[15:52] If for any reason we get an unemployment
[15:54] report at any point, which is greater
[15:56] than 150, when everyone is now moved
[15:58] down to zero is okay,
[16:00] I think with inflation moving higher and
[16:02] a stable labor market, you will start to
[16:04] freak out in terms of whether the Fed
[16:06] may go.
[16:07] Um John Roque uh has been pounding the
[16:10] table on this chart, and this would
[16:12] argue for this being a risk, which is
[16:14] technically he does not like two-year uh
[16:17] notes. Uh and this is from a price
[16:19] perspective, this would mean yields
[16:20] going higher, and his first level is up
[16:23] at 4%, but his target is five. We're
[16:27] currently in the upper threes,
[16:30] uh around three and three quarters, so
[16:32] anything that moves in terms of the Fed
[16:33] possibly going could be something that
[16:35] kind of freaks out the market, even
[16:37] though I don't think the Fed is going to
[16:39] go. Texas Instruments reported. So,
[16:41] real-time data week. Uh
[16:43] they said, "Industrial demand keeps
[16:45] improving while data center segment
[16:47] flourishes." So, we are now getting the
[16:49] physical upgrade spreading
[16:51] into the industrial side. Texas
[16:54] Instruments is a demand barometer for
[16:56] the industrial economy. They cover so
[16:59] many different things, so
[17:01] recent commentary notes that industrial
[17:02] demand is recovering broadly. Comments
[17:05] about AI-related sensing and power needs
[17:07] suggest that AI workloads are starting
[17:08] to penetrate industrial and edge
[17:12] form factors, meaning robots, test
[17:14] equipment, building assistant, EV,
[17:16] industrial vehicles, not just GPUs in
[17:17] the cloud. This is critical, guys.
[17:19] Remember, we just increased the
[17:21] uh
[17:22] spending for defense.
[17:24] We are getting into the edge. The edge
[17:26] is everything. It's drones. It's
[17:29] humanoids. It's autonomous vehicles.
[17:31] It's computers. It's phones. It is
[17:33] enterprise edge. This is the beginning,
[17:37] the very beginning. This isn't even the
[17:38] first inning. I'm not even sure they've
[17:40] gotten through the national anthem yet.
[17:42] Texas Instruments expands
[17:43] microcontroller portfolio and software
[17:45] ecosystem to enable edge AI in all
[17:48] devices. And again,
[17:50] I bring this up and mention improve
[17:53] energy efficiency when processing at the
[17:55] edge and neural processing units. If you
[17:57] guys have my paper from NPUs, I would go
[18:00] bring it up and go read again. This is
[18:03] all happening now. So, if you don't like
[18:06] Texas Instruments, how about we go to
[18:07] Intel?
[18:08] They emphasize that AI workloads evolve,
[18:10] the importance of CPUs infrastructure is
[18:12] growing. I spent a lot of time on Intel
[18:14] last week, going from dead just seven
[18:16] months ago
[18:18] to
[18:19] breaking through all-time highs, and now
[18:21] we're ripping through them.
[18:23] Up this morning, 20-plus percent.
[18:26] You cannot put a GPU everywhere. We're
[18:28] moving into inference. Again, this is
[18:29] the very beginning. Intel and
[18:31] third-party analysis highlight that CPU
[18:33] demand is being pulled not just by
[18:35] cloud, but also by enterprise AI
[18:37] deployments
[18:38] that re-
[18:40] rely on CPU-heavy systems. Again, this
[18:43] is part of the edge device side. They
[18:46] mention Intel has flagged signs of
[18:48] improvement in PC-related CPU demand,
[18:51] with prior quarters already showing a
[18:52] pickup. One of the mistakes people are
[18:54] making,
[18:55] even if the prices go higher and PC
[18:59] overall PC demand is, let's say, lower
[19:02] than it was last year. Let's say
[19:04] handsets are lower than we were last
[19:06] year. Think Qualcomm. I'll be talking
[19:08] about Qualcomm more and more in the
[19:09] coming weeks. Remember, if you only do a
[19:12] certain this slightly less than last
[19:15] year,
[19:16] but of those, the percentage that are
[19:18] AI-related goes up, that increases all
[19:23] of these companies like Intel. It is the
[19:26] rotation. It is the, quote-unquote, as I
[19:29] wrote at the very beginning, the
[19:30] physical upgrade.
[19:32] Upgrade, upgrade, upgrade. You don't
[19:34] need to have GDP to have an upgrade. You
[19:37] can have zero GDP, and it can be a
[19:38] rotation. And when I mean zero GDP, I
[19:41] mean you can have zero growth, like you
[19:43] do in autos. You could sell 16 million
[19:44] autos this year, 16 million next year.
[19:47] And if this year none of them were AI,
[19:50] and next year 4 million of the 16 are
[19:52] AI. You had massive growth in AI. Well,
[19:54] that has enormous impact for
[19:56] semiconductor chips. That is what
[19:58] everyone is missing is they are focusing
[19:59] way too much
[20:01] on nominal GDP growth and not enough on
[20:04] the upgrade cycle.
[20:07] The Edge Awakens. So, this is going out
[20:09] Monday morning. I just wanted to give
[20:11] you guys a preview.
[20:13] The Edge Awakens. It's all on news that
[20:15] happened this week and it's not just the
[20:17] Intel and Texas Instruments, which you
[20:20] can see I highlight here, but it's also
[20:22] Nvidia's reported 800 V DC power push in
[20:25] South Korea that was going through the
[20:27] media this week.
[20:28] I wrote a paper on the whole rack. Intel
[20:31] is in the whole rack. So, it's one of my
[20:33] names in there. The Edge AI investment
[20:35] [clears throat] universe, which I
[20:37] highlighted last week and went on the
[20:38] subscriber website, which has 68 names
[20:41] in it. Many of them are in the overall
[20:43] portfolio. But in particular, the power
[20:46] semis, which is what this is on, there
[20:48] were six of them in there.
[20:50] If you didn't see what they've done this
[20:52] week, and again, I put this out the Edge
[20:55] thing last week, this is what the power
[20:57] semis did this week and this is the
[20:59] chart. So, 5 years ago, they were
[21:02] unchanged as of last week. Those six
[21:05] names have gone through the roof, of
[21:07] which one of them is Texas Instruments.
[21:09] So, I just want to remind you again
[21:11] because this was brought up in Craig
[21:12] Fuller.
[21:13] I highlighted this on the webinar this
[21:15] week. And again, I say this for all the
[21:17] FAs and RIAs that are out there. We've
[21:18] seen a lot of
[21:19] a lot of you guys signing up. Tell your
[21:21] other people that work for the firm, you
[21:24] should be paying attention to this. It's
[21:26] going to be very hard to get this kind
[21:27] of research. This is something that's
[21:29] worth the money and the reason is
[21:30] because I'm covering the space in a way
[21:33] that's real time. Power semis is a
[21:35] real-time story. It's in the very, very
[21:37] early beginnings. There's a lot of names
[21:39] on there
[21:40] that as far as I'm concerned, they
[21:42] haven't even seen their earnings grow in
[21:44] a meaningful way. You saw it with Texas
[21:45] Instruments, but the edge devices are
[21:47] going to keep coming because as I go
[21:49] through, we haven't even started
[21:51] humanoids. Like this literally is the
[21:53] beginning of a long-term process and
[21:56] last week I covered Terafab. Terafab is
[21:59] just your confirmation that Elon Musk
[22:01] believes we're going to run out of
[22:02] chips.
[22:03] You want to
[22:05] invest in scarcity.
[22:07] 6 months ago, post Opus 4.5 and open
[22:10] class. So, before we got there, these
[22:12] were all the things that people were
[22:14] worried about. This is what Craig Fuller
[22:15] was talking about. Here's where we are
[22:17] now. It has only been 6 months ago.
[22:21] All right, let's go through this
[22:22] quickly. The spending numbers, they're
[22:24] being revised higher.
[22:26] So, this is from BlackRock this week.
[22:28] Just look at these numbers. And again,
[22:29] this is the hyperscaler.
[22:31] These are still for the main companies.
[22:34] Remember, you've got other companies
[22:36] that are spending like OpenAI and
[22:38] Anthropic are each targeting now 30
[22:40] gigawatts of compute. They're not
[22:42] included in that hyperscaler number.
[22:44] Then you've got SpaceX not included in
[22:46] that hyperscaler number. They're going
[22:47] to do 25 billion or so
[22:50] 25 billion or so. Then you've got Tesla
[22:53] this week announcing increased to do 25
[22:55] billion. These numbers just don't stop.
[22:59] They are massive. And the way you can
[23:01] see them is ERCOT electricity demand.
[23:04] This is basically what we're talking
[23:06] about. So, here's where we are now.
[23:08] We're already at a problem.
[23:10] This problem gets worse and worse in
[23:13] terms of peak demand. Based on
[23:15] everything, and if you want to know
[23:17] what's causing us to be, look at these
[23:18] numbers. I mean, how are we going to get
[23:20] enough power?
[23:22] This is all related to data centers.
[23:24] This is the number out of this number
[23:26] total in terms of what is needed. And
[23:29] this point is being made. Remember when
[23:30] they screamed that Bitcoin mining would
[23:32] boil the oceans? Here's Bitcoin mining's
[23:34] expectations and this is where they
[23:36] were. We're dealing with numbers so
[23:37] massive. And again, look at the growth
[23:40] rate.
[23:41] This is all AI related. If you're not
[23:43] investing in these kind of CAGRs, what
[23:45] are you doing? That's entirely what my
[23:47] thematic basket is. Who cares if
[23:50] Salesforce is going to survive or not
[23:52] when you've got the opportunity to
[23:53] invent invest in those type of CAGRs.
[23:55] You just need to know the names. Reid
[23:57] Hoffman came out and basically said
[24:00] started talking more about something I'm
[24:02] I'm spending more time on, which is
[24:04] we're having a problem here with the
[24:06] data centers being built. You've got
[24:08] bottlenecks in in parts coming from
[24:10] China. You've got the politicians and
[24:12] the
[24:13] looking for votes and threatening rising
[24:16] electricity costs. So, they're spreading
[24:18] doomer news in terms of trying to get
[24:20] voters to turn.
[24:21] SemiAnalysis put this thing saying that
[24:24] now we've got the Blackwell shipment
[24:25] showing a slowdown. The issue is not
[24:27] demand, but infrastructure readiness.
[24:29] Some cloud service providers still do
[24:30] not have the data center infrastructure
[24:32] fully operational.
[24:33] This is what's happening. This is why I
[24:35] put the data center thing up.
[24:37] My analysis in the the subscriber
[24:40] website, which you guys can go see.
[24:42] Anthropic and Amazon expand
[24:43] collaboration for up to 5 gigawatts of
[24:45] new compute.
[24:47] Anthropic has troubles because they
[24:49] don't have enough compute. Meta and
[24:51] Microsoft looking to pay for their AI
[24:53] spending by trimming workforces.
[24:56] Trimming 10% of workforces not trimming
[24:59] um
[25:00] 7% for Microsoft. They have to finance
[25:03] this somehow. The spending will happen
[25:05] regardless of whether it comes from
[25:06] bonds or it comes from firing people,
[25:08] but it's going to come. The question is,
[25:10] will they be able to get the ROIC? The
[25:12] problem is for those companies, it's
[25:13] very hard to get the ROIC unless your
[25:15] data centers are built. That's the
[25:17] issue. So, the progress is happening.
[25:20] Demand is there from enterprises, but
[25:22] the question is, is the performance
[25:23] there? And as I go through, the adoption
[25:25] is screwed up as well.
[25:27] Why energy is becoming a boardroom
[25:29] priority. This was a great piece this
[25:31] week came out from
[25:32] Pure Storage, now Everpure, a company
[25:35] that is also on my
[25:37] portfolio thematic portfolio.
[25:41] I would go read it. The potential power
[25:42] shortage for AI infrastructure. Just
[25:44] look at these numbers, the shortfall
[25:46] available supply. Do I think they're
[25:48] going to find a way?
[25:50] Yes, but not in the near term.
[25:53] Data centers growing electricity
[25:55] consumption is set to double. There's
[25:57] just no way to build the infrastructure
[25:58] fast enough, so it's going to have to
[25:59] come through efficiency and things like
[26:01] this. So, this is what I was saying
[26:03] before about South Korea media report.
[26:06] Nvidia has reportedly visited Korean
[26:08] power equipment companies and asked them
[26:10] to design data center infrastructure
[26:11] around an 800 V
[26:13] class DC architecture.
[26:15] This is being interpreted as an effort
[26:17] to minimize power conversion losses. So
[26:20] again, everything is focused on power
[26:21] efficiency.
[26:23] This is the reason why on the optical
[26:25] side, you've got so much demand and this
[26:27] is the reason why the power semis are
[26:29] there. So, optical necessary for
[26:31] inference. This is why we've got
[26:33] Coherent and Light and Corning. It's
[26:35] also why we have Marvell, which is one
[26:37] of the leaders in silicon photonics and
[26:39] has purchased companies including one
[26:40] this week. All of these companies fit on
[26:43] being able to somehow or another find a
[26:45] way to get the compute through without
[26:48] needing more gas turbines, more gas,
[26:51] more everything by being more efficient.
[26:53] There's a race on all of these sides and
[26:54] this is where this fits in and this is
[26:56] why those South Korean heavy-duty power
[26:59] things go, but it's also where the power
[27:01] semis go. So, we've got shortages. The
[27:04] problem is the demand's growing so fast,
[27:05] we have run into shortages. We've got
[27:08] component shortages in many things, but
[27:10] in particular in this, the optical
[27:11] transceiver
[27:13] market.
[27:14] Naphtha has become a big issue because
[27:18] of the Middle East conflict. Critical
[27:20] semiconductor materials including
[27:23] ceramic capacitors, power devices,
[27:25] helium. They're expected to face rising
[27:27] prices, shortages and geopolitical
[27:28] disruption throughout this year. Lead
[27:31] times already there.
[27:33] Dylan Patel gave an interview with
[27:34] Invest Like the Best with Patrick
[27:36] O'Shaughnessy. Again, highly recommend
[27:39] listening to it.
[27:40] The critical parts in there. The economy
[27:43] has flipped. Ideas are are now abundant.
[27:45] So, get back to the abundance side. You
[27:47] want to be short abundance. Ideas are
[27:49] abundant. Anything that you can come up
[27:51] with an idea like a software, it comes
[27:53] to market immediate now. You can utilize
[27:55] it immediately. The death of software
[27:58] that is built on code.
[28:00] He described SemiAnalysis' own AI
[28:02] spending exploding from tens of
[28:04] thousands to 7 million annualized run
[28:06] rate. So, I want you to think and do the
[28:07] math in your head. If a little company
[28:10] like SemiAnalysis has gone to a 7
[28:13] million annualized run rate on AI
[28:16] spending,
[28:17] think about what a Goldman Sachs, a
[28:19] Morgan Stanley, a JP Morgan, Eli Lilly,
[28:21] how much money they have to spend and
[28:23] you start getting to the 90 trillion
[28:25] dollars, guys. You have to start really
[28:27] doing the math and realize how big this
[28:29] is. On token demand, he says the key
[28:31] story is not falling cost per token, but
[28:33] exploding demand.
[28:35] He expects meaningful robotics
[28:37] breakthroughs in the next 6 to 18
[28:39] months. The whole AI stack is
[28:41] tightening. Remember, the whole rack is
[28:44] the whole AI stack and that gets back to
[28:46] the five-layer cake. So,
[28:49] margins are expanding, memory is
[28:50] severely constrained, and even upstream
[28:52] items like copper foil PCB materials.
[28:54] Again, PCB material
[28:56] you're going to run into the same thing
[28:58] in chemicals, guys. He does highlight a
[29:00] public backlash against AI and they need
[29:03] to have the This is going to be an issue
[29:05] all year. The bottlenecks are going to
[29:07] be an issue.
[29:08] Um
[29:09] People will say, and I've heard this, is
[29:11] why I wanted to highlight this is, "Oh,
[29:12] the memory story is old news. Everyone
[29:15] gets it." No, no, no, you don't get it
[29:16] all. DRAM capacity will still double or
[29:19] even triple because the market needs
[29:21] such massive production capacity and
[29:24] they have to steal capacity from
[29:25] somewhere else. In a capitalist economy,
[29:27] the only way to steal capacity is by
[29:29] driving up prices to destroy demand.
[29:31] Elon Musk is building a Terafab. It's
[29:34] all related to the same thing. We have
[29:37] years ahead of a shortage, guys. Years.
[29:39] Here's another way to look at it. This
[29:41] is just the AI training and what is
[29:43] needed or the change that is going uh uh
[29:49] basically from inference. So, we're
[29:51] going from
[29:53] pre-training into AI inference. This is
[29:57] where we are now. Every year it gets
[29:59] bigger and bigger. That's why when you
[30:00] see these CPU charts and you start
[30:02] realizing why we need so many CPUs not
[30:04] just today. The agentic AI shifts the
[30:07] latency bottleneck from GPUs towards
[30:09] CPUs, just what I said. So, again,
[30:12] chatbots,
[30:14] that's what most of you use at this
[30:16] point. Well, let's go through everything
[30:18] and realize that complex orchestration,
[30:20] which really gets into the enterprise
[30:22] side, especially
[30:25] we need an enormous amount of memory.
[30:27] Morgan Stanley's report on agentic CPUs
[30:29] has a TAM expansion forecast in line
[30:30] with mine from a month ago. I thought it
[30:31] was crazy for saying it would reach 100
[30:33] billion in 2030 and now it's becoming
[30:35] consensus.
[30:36] >> [sighs]
[30:37] >> I mentioned the Intel,
[30:39] which happened last night.
[30:42] Also, Open AI released GPT 5.5,
[30:47] bringing the company one step closer to
[30:49] AI super app.
[30:51] On many metrics better than Mythos.
[30:54] China's Deep Seek released a preview of
[30:56] the long-awaited V4 model. It looks like
[30:58] they used Nvidia chip, so that reduces a
[31:02] risk for Nvidia.
[31:04] Oh, yeah, that's right. Uh nothing has
[31:06] been solved in Iran. The only thing
[31:08] that's not happening is we're not seeing
[31:09] bombings right now, but I wouldn't be
[31:11] surprised if that picks up again, but we
[31:14] do have the disruption. The oil shock
[31:16] that dwarfs history won't have a rapid
[31:18] recovery.
[31:20] This to me is the story. The doomers
[31:22] saying that this is like COVID have
[31:24] already been proven wrong. Um this is
[31:26] not a US, you know, stock bubble thing.
[31:29] The rest of the globe is going through
[31:31] this. This is an AI story. Again and
[31:33] again and again, I'll keep saying it. If
[31:34] you keep getting caught into these and
[31:36] go. Now, that doesn't mean that the S&P
[31:38] 500 won't finish down this year. That
[31:40] doesn't mean we won't see a 10%
[31:42] correction over the course of the next 3
[31:44] months. I think it's very possible we
[31:46] will. Again, I think the issue is the
[31:48] inflation and the possibility that rates
[31:49] will go higher. The bottlenecks are
[31:51] there. We could have disruptions where
[31:53] semi production doesn't meet what was
[31:54] expected. We miss earnings. There are
[31:56] tons of things that can go on in a
[31:58] 1970-style bottleneck. Do not get caught
[32:02] in looking at parabolic charts and not
[32:04] realize they can fall 30% and still be
[32:07] within a parabolic move. That is what
[32:09] you have to get used to. We are used to
[32:11] these nice linear staircases and not
[32:13] these elevators up, elevators down type
[32:17] moves, but that's what you're going to
[32:18] see. And the way that I know that is if
[32:20] you go back to the 1970s and you look at
[32:22] oil,
[32:23] if you look at it on a quarter basis
[32:25] where it strips out the daily moves,
[32:27] you're going to see something that's
[32:28] almost a straight line, but if you look
[32:29] at it daily, you have plenty of big
[32:31] corrections in this stuff.
[32:33] In this case, you can see the numbers.
[32:35] It is completely different. Forget the
[32:37] price side, the regional supply,
[32:40] I mean, these are big numbers. The
[32:41] demand has not been impacted and that's
[32:43] because again, most of the demand that's
[32:44] happening globally
[32:46] is happening in AI, which does not need
[32:48] oil. Kuwait declared force majeure.
[32:50] Force majeure is going to be a story for
[32:52] this year. It's going to be across many,
[32:53] many items. Do not think this is just
[32:56] going to be on the oil side. This will
[32:57] be in a lot of places.
[32:59] Russia extended their fertilizer export
[33:01] quotas until December as the global
[33:03] deficit deepens.
[33:05] Here comes the food inflation. I mean,
[33:07] these numbers speak for themselves. You
[33:09] don't think people are going to be
[33:10] complaining and there's no way to get
[33:11] around this. The shipping has gone
[33:13] higher in terms of diesel prices.
[33:15] Fertilizer prices have gone higher.
[33:17] Everyone forecasting or anyone
[33:18] forecasting that we're not going to see
[33:20] any kind of an increase in inflation, if
[33:22] we don't see increase inflation, I will
[33:23] be shocked. UAE asks about a wartime
[33:26] financial lifeline.
[33:28] I think the disruptions that have macro
[33:30] implications from what has gone on,
[33:32] people need to pay attention to. Okay, I
[33:34] put this slide in the subscriber thing.
[33:36] I'm going to emphasize it twice here.
[33:39] Again,
[33:40] what I want you to think about is just
[33:42] the massive nature of all of this in
[33:45] terms of the data center infrastructure
[33:48] moving into the Blackwell side and the
[33:52] Vera Rubin and where we are and the
[33:55] build-outs of more data centers
[33:56] necessary and what this all leads to.
[34:00] Again, we need more tokens per watt.
[34:02] We're going to have grid overload, so we
[34:04] could have blackouts. We could have
[34:05] situations. We've got power capacity
[34:07] limits. When you're running things hot
[34:10] and you don't have the investment in the
[34:11] physical infrastructure, expect there
[34:13] for there to be problems. At the same
[34:15] time, if the models get so powerful,
[34:17] you're going to have hackings. We don't
[34:19] have the security ready. When you move
[34:21] too fast and you run things too hot, it
[34:23] is a different kind of scenario than
[34:25] after the GFC where unemployment was 10%
[34:27] and we were gradually taking it back to
[34:29] four. Now we're fully employed. We have
[34:33] rates below inflation and we have way
[34:36] too much demand relative to supply and
[34:38] the money to fund it. It is not done by
[34:41] debt. It is not going to be a systemic
[34:43] event. Even though I'm still negative on
[34:45] private credit, I'm still negative on
[34:46] software and I think the credit cycle is
[34:48] in the early stages of deleveraging.
[34:51] It's not going to be fast because
[34:53] there's too much capital flying around
[34:54] and too many industries that are fine.
[34:56] This is where the edge is. We haven't
[34:58] even gone to the edge yet, guys. That's
[35:00] why I wanted that visual there. So,
[35:01] let's go through the analysis on the
[35:03] S&P. We've seen the earnings amazingly
[35:05] go higher despite the situation Iran.
[35:07] Well, if you want to break down where
[35:09] they're coming from, they're coming from
[35:11] technology, energy, and materials.
[35:14] This is the entire
[35:17] estimate revisions.
[35:19] And out of that, it's basically one
[35:22] company is 51% of 100 of it. Micron
[35:25] Technology is 51% of the earnings
[35:27] revision. Now, what else do you have on
[35:29] the list? Exxon, Chevron, Broadcom.
[35:31] These three alone make up 85%. This is
[35:35] not a broad thing. This is isolated to
[35:37] those areas that are part of all of
[35:39] those names, by the way. Those four
[35:41] names
[35:42] are part of my thematic portfolio.
[35:45] Those have been there all year. That's
[35:47] what we're seeing is the earnings growth
[35:49] again from this side. I thought this was
[35:51] uh I showed this this week in the
[35:53] subscriber webinar. I want to make sure
[35:55] that everyone gets to see it because
[35:58] the S&P earnings growth for the Mag 7,
[36:00] 23%.
[36:01] Here's the rest of the S&P 500. But,
[36:07] if you strip out
[36:09] Nvidia,
[36:11] then all of a sudden it goes to 6.4.
[36:15] So, if Nvidia were excluded, the other
[36:18] 495 would be required would be reporting
[36:20] higher earnings than the remaining Mag
[36:22] 7. So, the Mag 6
[36:24] uh or the Mag 7X
[36:26] Nvidia is only growing earnings at 6.4%.
[36:29] Think hyperscalers, guys. Think funding.
[36:32] All right, this is where the adoption
[36:33] problem gets in. AI is everywhere, the
[36:35] agentic organization isn't. I think you
[36:37] should read this report. It's a McKinsey
[36:39] report. I'm not going to show you the
[36:40] details, but basically,
[36:43] having worked at Morgan Stanley, knowing
[36:45] how archaic the entire infrastructure
[36:47] is,
[36:48] to make a wholesale change is
[36:50] impossible. And Aaron Levie put this out
[36:52] this week. It's remarkable how often you
[36:54] need to be dramatically upgrading your
[36:56] AI architecture given the pace of
[36:58] progress in AI models right now. For the
[37:00] fourth time I upgraded my open claw
[37:02] today.
[37:03] Now, not all today, but during the time
[37:06] since I started it, GPT-5 is now been
[37:08] uploaded into my open claw. So, again,
[37:11] and I'm working on getting more Mac
[37:13] Minis and more Mac Studios so that I can
[37:15] use either Deep Seek or Kemi K 2.6,
[37:18] which came out this week, which is
[37:19] equivalent to Opus 4.6 or Opus 4.7 as an
[37:22] open-source model that is anywhere from
[37:24] 70 to 90% cheaper.
[37:27] We're moving too fast for companies to
[37:29] know what to do.
[37:31] Amazon boom is creating a massive
[37:33] duplicate tools and data inside the
[37:34] company. As someone who uses this all
[37:36] day long, I don't know how you separate
[37:38] the two. You've got two separate
[37:40] components, the personal workflow, and
[37:42] that's what this duplicate tools inside
[37:44] the company are. Meaning, if I build
[37:46] something, but the person next to me
[37:47] doing the same job builds a completely
[37:49] different tool, which one do we use? Why
[37:51] do we have two tools? We're using more
[37:52] tokens. Maybe one's more efficient than
[37:54] the other. Then you've got the
[37:55] orchestration level, which is more from
[37:56] a top-down basis. How do we replace 80%
[37:59] of the people when we need to make sure
[38:00] that AI can do It's a very complex
[38:03] thing. The adoption is going fast from a
[38:05] place of zero, but don't be surprised. I
[38:07] still don't think the ROIC is going to
[38:08] come in as fast and that's the reason
[38:10] why I want to be short this. So,
[38:13] many of you on the RIA and FA world,
[38:16] this is going to be a problem. So, this
[38:17] is the chart. The white line here is
[38:20] Russell growth.
[38:22] The Russell 1000 growth. So, the
[38:24] overweight position that many, many
[38:26] people, including my stepfather, have is
[38:29] in growth. Uh there needs to be a
[38:31] reweighting and it needs to be not
[38:33] necessarily towards value, but just have
[38:35] to know that growth is not happening the
[38:37] way it was. This is about
[38:38] semiconductors. This is about energy.
[38:40] This is about materials. That's where
[38:41] you need to have everything. This chart
[38:43] is overlaid with the hyperscalers
[38:45] relative to the S&P. So, since 2016, and
[38:48] if I took this back, and the reason I
[38:49] haven't is because of the hyperscalers,
[38:51] but if you took this back to 2007, this
[38:53] has been all of the performance. You
[38:54] just wanted to be long growth relative
[38:56] to the S&P or the hyperscalers. Well,
[38:59] now I think that's going to be in a
[39:00] permanent reversal the other direction.
[39:03] Salesforce came out with their headless
[39:05] 360
[39:07] and then immediately
[39:09] uh were billy clubbed uh and down 8, 9%
[39:12] yesterday on the back of IBM's earnings.
[39:14] This is something I talked about a
[39:15] couple weeks ago.
[39:17] The bar for the software companies is,
[39:19] if you beat earnings, which is what's
[39:21] been going on, that's not a surprise.
[39:23] Unless you can actually show that your
[39:26] AI stuff is going to offset the rest of
[39:28] your seat-based model, that's the only
[39:31] way your stock's not going to fall. If
[39:32] you miss for any reason or it's slightly
[39:34] disappointing, you're down 8 to 10%.
[39:37] That's why it's a waste of time to spend
[39:38] time in the software side. This spreads
[39:40] into the private credit side. It's a
[39:42] very large default uh this week in
[39:45] Medallia. Probably didn't read people's
[39:47] radars, but wipes out 5.1 billion in
[39:50] equity, imperils the lenders, too.
[39:53] Blackstone's public BDC has massive
[39:57] size, nearly 400 million, and complies
[40:00] as 5% of the net assets.
[40:02] If you're in private credit,
[40:04] good luck.
[40:06] Uh again, this chart just kind of bring
[40:08] it in home with the themes. Here, since
[40:11] the lows in October, I wanted to show
[40:13] that this is my thematic portfolio. It's
[40:16] been up. This is the daily numbers every
[40:18] single day so far this month except for
[40:21] one.
[40:22] Uh and it was down slightly that month
[40:24] and
[40:25] slightly that day. Here is the basket
[40:28] relative to the hyperscalers. This is
[40:29] the trade that I want to have on. So,
[40:31] for long short people, for hedge funds
[40:33] that are looking for the basket and
[40:34] looking for access, this is the way I
[40:36] would trade it. Uh I think it's going to
[40:38] have You can see it's a good sharp
[40:40] ratio. They are correlated.
[40:43] It It's going to matter. This is it
[40:45] relative to the S&P, and there's two
[40:46] things I want you to realize. Yes, it's
[40:48] had a breakout,
[40:50] but it was unchanged 6 years ago or 5
[40:53] and 1/2 years ago as of last week. So,
[40:56] this is the beginning of a trend. My
[40:58] portfolio is not something that is late
[41:00] stage. This is the beginning of a trend.
[41:06] Uh thanks to the people who reached out
[41:08] on the technical side. There were a
[41:09] couple RSIs that were wrong. I ran
[41:11] everything. I will put this up every
[41:13] single week. And again, for people
[41:15] calling, this is not what I think people
[41:17] should do. This is not some scoring that
[41:20] is done by other people. This is purely
[41:23] something I put together. There is an
[41:25] explanation that's on the subscriber
[41:27] site. This is giving you the ability of
[41:30] getting all the technicals for them in
[41:31] one place.
[41:33] What I did was say I did a video this
[41:35] week that you guys got
[41:37] how to use Claude to turn your portfolio
[41:39] list into trade ideas. You can do
[41:40] whatever you want. If you like to find
[41:42] names that aren't working on my list,
[41:44] where their RSIs just went from 30 to
[41:46] 35, but they're down 20% for the year,
[41:49] go for it. I show you in this video how
[41:53] you can put different prompts and how
[41:54] you can do it. It's a short video. Watch
[41:56] it. It's 5 minutes, but it'll give you
[41:58] some ideas on how to do this. In terms
[42:00] of the Nvidia
[42:02] 800-V DC
[42:05] not announcement, but report in South
[42:07] Korean media that I talked about. Um
[42:11] again, as a reminder, this was it and
[42:12] the demand for power firms. What I did
[42:15] was I want to remind you this edge side.
[42:17] So, as a reminder in that first part of
[42:19] this video, I talked about the
[42:21] connection to the edge. This is up on
[42:23] the website. It went up last week.
[42:26] Power semis are one of the verticals
[42:29] that are in there, and there are six
[42:30] companies. That was the chart I showed
[42:32] you. That is this chart. Equal weight of
[42:36] those names. That's what it's up now.
[42:37] These are timely things when I put them
[42:40] out.
[42:41] Now,
[42:42] I did a dual one,
[42:45] and I connected the Nvidia side with the
[42:47] edge investment side using the Claude
[42:50] skill, which I gave to you guys last
[42:52] week as well. For those of you who tried
[42:54] to use it and it didn't work, I can't
[42:56] say this loud enough.
[42:58] It works. All you have to do, if it
[43:00] doesn't work for you, is put it into
[43:02] Claude and say, "How do I do this?" You
[43:04] guys have to get used to using AI, and
[43:07] the more that you use it, the better you
[43:09] will get at it. That is the reason why
[43:11] I'm doing the videos. I can't
[43:12] individually go through every single
[43:14] person and figure out what you did that
[43:16] was wrong. So, please take the prompt
[43:19] that I gave you, upload it, try to run
[43:21] it. If it doesn't run, ask it what I'm
[43:23] doing wrong, and then Claude will take
[43:25] you through and fix it. If you're not
[43:27] paying $20 for it, I don't know if it'll
[43:29] work, because this is something that I
[43:31] know that my son, who's a college
[43:32] student, I gave him the entire thing. He
[43:34] ran all of them. I know that a lot of
[43:35] people have reached out and showed me
[43:37] what they're able to do. This is the
[43:38] kind of stuff you can do is take that
[43:40] news report, upload it with my edge
[43:42] investment universe, and say, "Hey, out
[43:45] of these names, give me some other
[43:46] names. Give me small cap names. Give me
[43:48] mid cap names." That's the way you use
[43:49] Claude connected to the work that I'm
[43:51] doing.
[43:52] This is what I'm working on now,
[43:54] molecular moat 2.0, chemistry at the
[43:57] interface. We're getting new news where
[43:59] advanced packaging, photonics, and
[44:01] liquid cooling coverage converge.
[44:03] If you guys are interested more in
[44:05] chemicals, I will get some more, but
[44:06] because of the Middle East conflict and
[44:08] what has happened with the the regional
[44:11] oil energy prices being different, uh I
[44:14] think there's an opportunity here, so I
[44:15] would go spend time on the chemicals. I
[44:18] will bring you something in the coming
[44:20] weeks.
[44:21] Bitcoin. Again, yesterday we got a huge
[44:24] move down in the software side.
[44:25] Remember, this overlay is what I've
[44:27] shown before. This is software relative
[44:30] or software overlaid with Bitcoin.
[44:32] I believe Bitcoin has made a bottom. The
[44:35] MACD weeklies are going. If I'm right,
[44:37] they should start moving at some point
[44:39] and get up to this level, which would
[44:40] get it close to 95 to 100,000. I believe
[44:44] if this is going to happen, it needs to
[44:45] happen soon.
[44:47] Uh and I say that because we have broken
[44:49] the correlation.
[44:50] Uh there's inflation that's coming. I
[44:52] gave you guys the quadrants in terms of
[44:55] if we're where we do phenomenally well
[44:57] for Bitcoin, we're entering that
[44:58] quadrant the next
[45:00] week. Uh sorry, the next month. If you
[45:03] think about PMIs, I've talked about PMIs
[45:05] going higher. Now we've got
[45:06] confirmation, the industrial thing, that
[45:08] PMIs are going to be going higher. We
[45:09] will get up towards my 60 level. So,
[45:12] once we get through the oil shock that's
[45:14] happening, I fully expect that we're
[45:16] going to see it go higher. So, between
[45:17] inflation and PMIs, that is normally a
[45:19] good thing. But the other thing is, by
[45:21] the end of this year, this is my main
[45:23] prediction for the people who are
[45:25] interested in Bitcoin, but also for
[45:26] hedge funds and mutual funds and private
[45:27] wealth managers, I've said this before.
[45:30] I'll say it to you guys. If you want
[45:31] Anthony Pompliano and I to come speak
[45:33] for your FAs, give us a call. By the end
[45:36] of this year,
[45:38] this growth asset basket will be a
[45:40] problem, and people are going to be
[45:41] looking for things they can invest in.
[45:43] And the problem is no one likes to buy
[45:45] the end of parabolic charts. The one
[45:48] thing good about Bitcoin right now is
[45:50] that it isn't a parabolic chart. It is
[45:52] unchanged right now. You're getting the
[45:54] chance to buy something unchanged
[45:57] from 2024.
[45:59] That is a very, very powerful situation,
[46:01] and that reminds me more of things that
[46:04] wealth managers should be doing. I think
[46:06] it's a great opportunity. Now, I have a
[46:08] Bitcoin proxy. I haven't shown it to you
[46:10] guys. It's three components, and it
[46:12] involves the three things I just said.
[46:14] So, it involves the Qs. That is the
[46:15] software side. It involves gold. That is
[46:18] the store of value side {slash} when
[46:21] everything is breaking down, where do we
[46:22] go hide our money? And includes copper.
[46:24] That is the PMI side. This is the
[46:26] long-term chart
[46:28] going back to 2017. The white line here
[46:30] is the proxy. We have alligator jaws,
[46:32] guys. I like the catch-up trade that's
[46:34] coming, and I think gold is going
[46:35] higher. I think copper's going higher. I
[46:37] think the Qs are going higher. I don't
[46:38] think the Qs are going higher because of
[46:40] the hyperscalers. I think the Qs are
[46:41] going higher because of
[46:43] biotech,
[46:44] semis,
[46:46] all of that stuff. There will be changes
[46:48] in the Qs. I don't think the growth
[46:50] baskets are going to do the same thing,
[46:51] and I certainly don't think software.
[46:52] So, I think by the end of the year,
[46:54] you're going to be wanting to increase
[46:55] your weighting in Bitcoin as a growth
[46:56] thing.
[46:58] And ETFs are coming for crypto. So, I've
[47:00] been asked this a lot. If you've been
[47:02] looking for stable coins and
[47:03] tokenization, you've got STBQ, Amplify
[47:06] Stablecoin Technology, Amplify
[47:09] Tokenization Technology. I'd go take a
[47:11] look. Uh and finally, for those of you
[47:13] in California, I will and those of you
[47:16] attending this event on the
[47:17] institutional and the public pension
[47:19] side,
[47:20] uh reach out to me. Um you can reach out
[47:23] to me at on Twitter. You can reach out
[47:25] to me on LinkedIn. You can reach out to
[47:27] me on on email if you have my 22V email.
[47:31] Uh you will be able to find me at this
[47:33] event speaking at the institutional
[47:35] investor event. That's it for this week,
[47:37] guys.
[47:40] Good luck trading these parabolic
[47:41] markets, and I'll see you next week.