Signal Over Noise: “The Most Important Moment in Human History”

[0:00] All right. Congrats to all the uh
[0:02] the Nick fans out there.
[0:04] Uh
[0:06] big week for me. All right. So,
[0:09] uh we're getting towards the end of the
[0:11] first half of the year.
[0:14] Not as much happening this week, but
[0:16] between the new Fed chair, between
[0:20] uh
[0:20] some weakness in the hyperscalers, and a
[0:23] whole bunch of little odds and ends.
[0:25] There's a lot that a lot of dot
[0:26] connections this week um
[0:28] more than anything else. So, let me
[0:30] start just uh
[0:32] buzzing through it. And just so you
[0:33] know, this will be my last
[0:36] weekly in uh Brooklyn. I'll be in Maine
[0:39] uh through at least uh August, maybe
[0:42] some of September as well. So, first
[0:44] thing I wanted to do uh just because I
[0:47] had such a huge uh
[0:49] reach out from people and some great
[0:50] emails in terms of people that made it
[0:53] through this process last week, built
[0:55] their own brain, had their kids work on
[0:58] one. A couple people used the suggestion
[1:00] for fantasy football. I highly recommend
[1:03] for people who uh listen to podcast for
[1:07] fantasy football. You have your top
[1:09] three to five, whatever the case as we
[1:11] get into the summer time. Follow these
[1:12] instructions and do the same thing. Uh
[1:15] you're going to have it. But more
[1:16] importantly, for all of you who did do
[1:18] it, that's part of the game plan in this
[1:20] weekly video is to make sure that you're
[1:22] um
[1:23] ahead of 99% of the people on the planet
[1:25] in terms of using AI. And if you can
[1:27] build follow those instructions for the
[1:29] 30 minutes, build the second one, build
[1:31] the third one, whatever the case, uh
[1:33] you're ahead of the game. So, um just on
[1:35] that note because I made this comment of
[1:38] everything that I'm basically doing is
[1:40] Signa Alpha agency on these podcasts.
[1:43] And now that I'm doing a lot more
[1:44] consulting
[1:45] uh for firms and for them to get through
[1:48] it, I just want to make sure that that
[1:49] is the theme for this and that you leave
[1:51] this video understanding uh how
[1:54] important the signal is to avoid the
[1:55] noise. There is an enormous amount of
[1:57] noise between the bubble talk, the
[1:59] tariffs last year.
[2:01] I'm I I'm not sure the oil doomer
[2:04] situation, the way it's played out,
[2:06] wasn't more than the tariff situation
[2:08] last year.
[2:09] The AI bubble trade, which obviously has
[2:11] been a theme. I don't care what it is,
[2:13] guys. Part of what I'm trying to do is
[2:15] actually give you facts on things. And
[2:18] eventually, if we have a bear market,
[2:22] we're going to talk about it here. If
[2:23] you guys have gotten to know me on this,
[2:24] I really don't care. This is an
[2:26] opportunity for alpha, and this theme
[2:29] continues to play out
[2:31] for all you RIAs and financial advisers
[2:34] who have not reached out yet, your
[2:36] portfolio is suffering every single
[2:38] month that you're not overweight towards
[2:41] the CapEx trade. I'm going to go through
[2:43] a lot of the details on this because it
[2:44] has been an intense first half of the
[2:46] year.
[2:47] Depending on how you measure it,
[2:49] you're talking 40 to 50% of alpha lost.
[2:54] Uh
[2:55] based on just keeping your weightings
[2:56] the same way in passive investments or
[2:59] moving into ETFs that theoretically are
[3:01] AI but have way too much weighting
[3:02] towards software.
[3:04] And then the agency side, like I just
[3:05] said with the knowledge brain,
[3:08] you got to help yourself, you got to
[3:09] help your business, you got to help your
[3:10] children, and you definitely have to
[3:11] hedge your clients. And that's what a
[3:13] lot of the stuff that I'm focused on is
[3:15] to get people to use this.
[3:17] For the week,
[3:18] I don't know how this is a bubble. I I
[3:20] mean, I I had to post this chart
[3:23] just as we come into the week just
[3:25] because we had a consolidation here for
[3:28] a while. We went down on the on the Iran
[3:32] fears, and then we ripped out of here,
[3:33] and now we've been consolidating again
[3:35] for the last month. So, basically, we
[3:37] rallied for
[3:39] 6 weeks, then we consolidated for 4
[3:42] weeks. That is
[3:44] basically a bull market in definition
[3:46] where, over the course, we haven't
[3:47] really moved that far. Now, again, one
[3:50] of the ways to measure bubbles and any
[3:52] issue is both evaluation basis and then
[3:55] also the distance between the 200-day to
[3:57] see how far ahead we are.
[3:59] Uh
[4:00] the bubble talk doesn't stop. So, here's
[4:02] the Economist. It is just hilarious how
[4:05] every one of the permabears when the
[4:06] Economist comes out with anything on the
[4:08] dollar, on rates, on the Fed, on the
[4:10] market, on anything, they immediately
[4:12] use it as a contrarian signal, but no
[4:14] one is sending this out this week where
[4:17] basically it's saying that the US market
[4:19] has reached the mania stage. So, you
[4:21] want to go on the opposite side of the
[4:23] Economist all the time. Well, there's
[4:24] one way to do it. Um I just want to make
[4:27] sure as we finish the first half of the
[4:28] year. It's been an extraordinary year
[4:30] for this.
[4:31] The S&P 500 PEG ratio, so the
[4:35] the PE divided by the earnings growth
[4:38] rate.
[4:39] So, first of all, for the S&P 600, the
[4:42] small cap side, look how much it's
[4:43] collapsed and basically got down to the
[4:45] lowest level. It's not just the level of
[4:47] it, it's how far down it came. The only
[4:49] way that happens
[4:51] is if the PE
[4:54] is expanding by less than the growth
[4:56] rate or go the opposite way. In this
[4:57] case, obviously the PE is growing
[5:00] rapidly. In terms of the blue line here
[5:02] or the red line here, which is the S&P
[5:04] 500,
[5:06] we're finishing this year at basically
[5:08] the lowest PEG ratio
[5:10] in the last 22 years on this.
[5:14] Um and obviously it was much higher as
[5:16] we went into the dot bomb the dot-com
[5:18] bubble.
[5:19] So, from evaluation basis,
[5:21] we just haven't done anything yet. On a
[5:24] relative to the 200-day moving average,
[5:26] this is the S&P rolling
[5:28] uh over time percentage above the
[5:30] 200-day moving average. We're currently
[5:31] 8.66 above the 200-day moving average.
[5:35] I mean, we're nowhere in bubble
[5:37] territory. This is the way a bull market
[5:38] is supposed to be. You hang out in this
[5:40] area for a long period of time. Uh
[5:42] basically in 2016 to 2019, we kind of
[5:46] sat in there the whole time. The thing I
[5:47] would say is
[5:48] we've had a lot of flush outs, a lot of
[5:51] breaking of the 200-day moving average.
[5:54] We just cleanse the system constantly.
[5:57] Um here are the queues again, incredibly
[5:59] strong chart, and now we're getting
[6:01] close to making new highs again. IWMs
[6:04] did make new all-time highs. The fact
[6:06] that small caps are making new all-time
[6:08] highs, the fact that their PEG ratio is
[6:10] collapsing. So, this is being driven by
[6:12] earnings. In fact, the price going up is
[6:14] going slower than the earnings growth is
[6:16] going. That's the only way you have the
[6:17] PEG ratio go down.
[6:19] The momentum trade consolidated, and as
[6:22] I mentioned last week, a lot of
[6:24] volatility in here. In fact, this is the
[6:25] TMT side for the tech side. It is the
[6:28] highest 30-day realized vol. You're just
[6:30] getting And this is a way that you shake
[6:32] people out as well. So, there's a
[6:33] variety of ways to get people to not be
[6:36] comfortable in a trade. For long short
[6:38] people, just increase the volatility.
[6:40] They see the P&L moving a ton, and it's
[6:42] very hard to stay in. You had the
[6:45] uh
[6:45] this is the industrials uh
[6:48] momentum, which came down and has
[6:50] rallied back. So, we're flushing out.
[6:52] It's just not happening the way that
[6:54] people want. Now, for the month, the
[6:57] S&P's down 1%. Again, a consolidation
[6:59] month, and in a consolidation, we're
[7:01] getting rotation.
[7:03] So, you've got the tech side led by the
[7:05] hyperscalers on the downside. You have
[7:07] the AI trade on the upside.
[7:10] And here is the hyperscalers relative to
[7:12] the S&P. This is a big part of my
[7:15] overall thesis. This is the CapEx
[7:17] arbitrage uh benchmark arbitrage for all
[7:20] you
[7:21] RIAs and FAs. Again, if you want a
[7:23] presentation on this, it's very simple.
[7:26] Everyone's overweight the spenders um by
[7:28] definition because of the market cap.
[7:30] So, if you're discretionary, I'm sure
[7:32] you're not as long on the hyperscalers
[7:34] as you've been. The Mag 7 fit into the
[7:36] same thing. We've seen massive multiple
[7:37] compression for the space. Uh Uh, but
[7:40] this month is one of the worst months
[7:42] for the hyperscalers in the last decade
[7:43] relative to the S&P with a 9% or eight
[7:47] and change percent underperformance.
[7:51] Uh, and here are the names
[7:52] month-to-date. Uh,
[7:54] all of them are down. Microsoft and
[7:56] Amazon and Meta, I'm going to get into
[7:58] some of these, but all of them have been
[7:59] hit. And here is my thematic portfolio
[8:03] uh, relative to the hyperscalers.
[8:05] So, now we're unchanged since the launch
[8:08] of ChatGPT
[8:10] on a relative basis. This is just
[8:12] beginning, guys.
[8:15] Still in the early innings.
[8:17] Kevin Warsh comes out. Let's continue
[8:18] with the noise theory. This is the
[8:20] headline. Stance fuels US interest rate
[8:22] hike bets.
[8:24] The insanity that just keeps moving from
[8:26] bearish thing to bearish thing. I don't
[8:28] care what anyone says about was this
[8:30] hawkish, was it not. We're just We treat
[8:33] Everyone's keep trying to look for ways
[8:35] to do it. I'm just going to reiterate
[8:37] two-year rates. If you've been focused
[8:39] on two-year rates since 2022, you're in
[8:41] misery.
[8:42] Why bother?
[8:45] There's nothing there. We're in a
[8:47] long-term range while AI's been
[8:48] exploding. Here's 10-year rates, same
[8:50] theme.
[8:51] If you want to get into what moves them,
[8:53] it's real simple. If you think Warsh is
[8:55] going to raise rates aggressively, not
[8:58] move 25 basis points before the end of
[9:00] the year and maybe 25 more, but move it
[9:02] aggressively, here is the chart of the
[9:05] Fed funds rate versus two-year rates and
[9:07] 10-year rates. If you take out the
[9:09] movement by the Fed,
[9:11] there's been no movement. It's all
[9:13] related to Fed fund rate hikes. So,
[9:15] unless you believe inflation is going to
[9:17] go higher or the job creation is going
[9:19] to accelerate, of which
[9:22] the only one that I thought there was
[9:23] any chance of was the
[9:26] inflation side, and I'm sorry. So, back
[9:30] before the Fed aggressively led, what
[9:32] led? This purple line led. These were
[9:34] inflation expectations for 1 year, which
[9:36] were rising well in front of the Fed,
[9:38] and they bailed on transitory almost
[9:40] near the peak of it. So, once they
[9:41] started to fight inflation, this went
[9:42] down. I just want to show you
[9:45] So,
[9:46] there is the beginning of Iran and the
[9:49] US, and we took this up.
[9:53] The inflation swaps have just collapsed.
[9:57] So, I don't see anything on the
[9:59] inflation front at this point for people
[10:01] to be talking about. So, we had the Iran
[10:04] peace deal. It's Saturday morning.
[10:06] There's the rumor about I
[10:08] I'm sick of even talking about it at
[10:10] this point. But, what I am going to talk
[10:11] about is
[10:13] for someone who really believed So,
[10:15] everyone gets sucked into noise, but
[10:17] people like at my age that have been in
[10:19] the markets for 30 years,
[10:21] we actually do listen. I listen to all
[10:24] of the people like Jeff Curry. All If
[10:26] you've watched my video all this year,
[10:27] I've highlighted all of the things
[10:29] they've talked about. Now,
[10:31] I keep one eye open looking for the
[10:33] charts, cuz at the end of the day, if
[10:35] what they're saying is right, this
[10:36] should not be happening. And I just
[10:38] remember here when everyone said, "Don't
[10:40] pay attention to the paper side. Pay
[10:42] attention to the physical side." And
[10:44] this is the dated Brent price. We got up
[10:47] to over 140,
[10:49] and we had the paper side all the way
[10:51] down here. Well, now they're both in
[10:53] line. The futures
[10:55] on Brent and the dated Brent, they're at
[10:57] the same point. How can you refute If
[11:00] there's going to be some mega spike up
[11:02] to 300, why are we back to where we
[11:04] were? At some point, the prices of the
[11:07] movements happen. You can sit there and
[11:08] speculate all you want, but this is a
[11:11] live gambling market. This is where the
[11:13] odds are. This is the tote board telling
[11:15] you that you're missing something. Now,
[11:17] you still could be right betting it's
[11:19] going to go up. Trust me, I'm long
[11:20] Exxon. I'm long Chevron as part of my
[11:22] portfolio. All I've done is cost me
[11:24] money. Same thing as silver. Same thing
[11:25] as Bitcoin. So, I'll get through that.
[11:28] On Fri- On Thursday alone, Uh, I've 20
[11:31] names in the portfolio that I have
[11:33] personally in my PA.
[11:35] 18 of them were down on the day, two of
[11:37] them were up. Now, luckily, I was still
[11:39] up a good amount, but that's only
[11:40] because of Integrys and because of
[11:42] Marvell. The rest of them were all down.
[11:45] So, you're definitely getting a scenario
[11:47] that if you're not in the the AI theme
[11:49] trade and you're listening to the noise,
[11:51] you're paying a huge price on all of
[11:53] these. You don't hear any of the
[11:55] permabears talking about gold right now.
[11:57] There's nothing really to talk about now
[11:59] that oil prices haven't gone. This is
[12:00] gas at the pump. This was gas at the
[12:02] pump on Thursday. We're below four now.
[12:05] This implies about 375.
[12:07] So, you've got gas moving lower.
[12:10] I I just read
[12:12] I just saw
[12:14] a video which [clears throat] came out
[12:16] over the weekend uh, with Chris Whalen
[12:19] still calling for double-digit inflation
[12:21] and using fertilizer as one of the
[12:23] things for the second half of the year.
[12:25] And again, when I listen to this stuff,
[12:28] here's fertilizer price. I mean, this is
[12:30] urea.
[12:32] It It went up and it has completely
[12:35] collapsed the spot price in New Orleans.
[12:37] So, the reality is for everything going
[12:40] on, crude prices have come down. 10-year
[12:43] rates are still sitting up here.
[12:44] Everyone's focused that the Fed is being
[12:46] hawkish.
[12:48] Here's 10-year rates versus the
[12:50] inflation swaps. So, maybe everyone is
[12:53] right, but right now we had the largest
[12:56] drop in 1-year inflation swaps
[13:00] since basically the lows in stock in
[13:02] 2022. The peak in inflation. That's when
[13:06] I mean, literally, you had the actual
[13:08] peak in inflation swaps here and then
[13:10] from that point it went down. I I've
[13:12] shown many times the 1970s, you don't
[13:14] want to be fading against inflation when
[13:16] it's a PTSD from people there. So,
[13:18] everyone who's macro side is looking for
[13:21] rate hikes, hawkish Fed, inflation's
[13:24] going higher. Here's the reality. This
[13:26] is the CPI. If you don't believe the
[13:28] inflation swaps matter,
[13:31] they matter. Here's the inflation swaps
[13:33] going down, not just 1 year, but 2 year
[13:34] as well, and they're back down into the
[13:36] 2 and 1/2 area. So, CPI
[13:40] it it looks like it's peaked, and it's
[13:42] headed lower, which I showed last week.
[13:44] Um I got to give a shout out to
[13:45] Trueflation. So, I I gave them a lot of
[13:49] hell because I it was obvious to me that
[13:51] we were going to get above 4%, which we
[13:53] did on CPI, but how you have the swaps
[13:56] going down, and you have Trueflation
[13:58] going back down as well. And the reason
[14:01] I care about this
[14:02] is because the forecast right now,
[14:05] midway through June, this will come out
[14:07] in July,
[14:09] for CPI for the month of June is now at
[14:11] zero by the Cleveland Fed, which will
[14:13] take the year-over-year down to 4%.
[14:16] Um
[14:18] don't forget this war story. This is the
[14:20] thing I took out of the press conference
[14:22] is his focus on task forces, but
[14:24] remember the measures I prefer looking
[14:26] at things that are called trim averages.
[14:29] So, if he wants to get buy-in from the
[14:31] group that they shouldn't be focused on
[14:33] the things they're focused on, and they
[14:34] should be focused on trim, which I don't
[14:36] have a problem on, but more importantly,
[14:39] he said there's a task force to
[14:41] reevaluate inflation, productivity, and
[14:43] jobs. This is everything related to AI.
[14:47] It really is. So, again, if AI is a
[14:52] deflationary force, I want you to focus
[14:54] on how little volatility the trimmed
[14:56] has. So, this is the Cleveland Fed trim
[14:59] trimmed median. The main thing about it
[15:01] is, except for this one move here,
[15:03] you're pretty much dealing with less
[15:05] volatility, significantly less than the
[15:07] CPI.
[15:09] Here's the Atlanta Fed core sticky white
[15:12] line here. This is core PCE right now,
[15:16] and then you've got the Dallas trimmed
[15:19] median,
[15:20] or trim mean.
[15:22] This is the San Francisco Fed one. This
[15:24] is a great page to go to just because of
[15:27] one thing in particular. So, it'll give
[15:29] you the their
[15:31] different measures in terms of this. The
[15:33] thing I care the most about in here is
[15:35] how narrow this blue is around this blue
[15:38] for again the median. And you're sitting
[15:40] not too far off where we've were before.
[15:42] Rates are significantly higher than
[15:44] where they were before. So, the median
[15:48] is at 2.99 and the 10-year average is at
[15:51] 2.91.
[15:53] That includes both the disinflation
[15:56] period of the prior decade and obviously
[15:59] the COVID period. So, you're getting a
[16:00] little of both. Um you know, I called
[16:03] out Peter Boockvar as in last week. This
[16:05] is another noise thing. How many times
[16:07] and again, I want to show this because
[16:09] how many times have we basically said
[16:10] the US consumer is dead?
[16:13] How many times? It just continues to
[16:15] happen. Delinquencies are going up.
[16:17] Consumer confidence has collapsed, which
[16:19] again, this is the lowest consumer
[16:21] confidence. I talked about this with
[16:23] Pomp this weekend.
[16:25] Here's the facts, guys. Again, how noisy
[16:27] was that? This is all you got to do. The
[16:29] Johnson Red Book, which comes out every
[16:31] week, year-over-year.
[16:33] 9% highest. I mean, this goes back to
[16:36] 1997. This is 30 years of data except
[16:40] for when we literally gave money out and
[16:42] handed out. Spending is happening. There
[16:44] is no recession. There is no weak
[16:46] consumer.
[16:48] Revisions, another real-time gauge of
[16:50] what's happening. They continue to sit
[16:52] up at high levels. They are not moving
[16:54] down at all and I've shown this chart
[16:56] before. This is the S&P 500 overlaid.
[17:00] Revisions are going higher. Before we
[17:02] get a significant fall in stocks,
[17:05] there's got to be something on the
[17:06] revision side to show a problem. We just
[17:08] every week are pumping out big numbers
[17:11] in estimate revisions. It helps that PEG
[17:13] ratio continue.
[17:15] When we don't go up fast, the S&P's
[17:17] earnings are up 28% year over year. The
[17:20] S&P's not up 28% year over year. There's
[17:22] multiple compression happening. On the
[17:24] credit side, again, I've talked about
[17:27] this. This ended up being noise. At some
[17:29] point, you have to admit, I highlighted
[17:30] repeatedly everybody who was going
[17:33] through this, including myself, worried
[17:35] that this could be something cuz we had
[17:37] this happening. This is credit spreads,
[17:39] junk spreads, inverted. This is private
[17:42] equity relative stocks relative to the
[17:44] S&P, and this is the BDC index. Well,
[17:49] credit spreads are back to all-time
[17:50] tights, and we haven't seen a change
[17:52] there. So, at some point,
[17:54] there's got to be another story. And if
[17:56] you want to go hear the other story, I
[17:58] thought
[17:59] uh I think his name is Dan Ivascyn. Uh
[18:01] he was on Compound and Friends.
[18:04] Uh
[18:05] he's the uh head of PIMCO now.
[18:08] And he basically went through, and I'm
[18:11] just going to read a couple of these
[18:13] things because I think what he
[18:15] Everything he said, despite global
[18:16] uncertainty, Ivascyn remains
[18:18] constructive on the US economy because
[18:19] the consumer balance sheet is solid. We
[18:21] see that through the spending, and we
[18:22] see it through the junk spreads. Labor
[18:24] markets have held up. There's no way to
[18:26] refute that. We're not creating many
[18:28] jobs, but we're holding up. And the
[18:30] AI-related spending is massive.
[18:32] Absolutely. Sees enormous investment
[18:34] needs across AI. Yes, that's not going
[18:37] away anytime soon, even though there
[18:39] could be air pockets, which I'll talk
[18:40] about.
[18:41] Stresses that lenders must negotiate
[18:43] strong terms because many AI-related
[18:45] borrowers still have uncertain long-term
[18:47] economics. Absolutely true. AI
[18:50] infrastructure can offer two-box safer
[18:51] ones and riskier ones that would have to
[18:53] provide equity-like returns.
[18:56] The capital loss cycle is not expected
[18:58] to be a dramatic wave of defaults, but
[18:59] rather a steady stream of higher
[19:02] realized losses in risky credit because
[19:04] old businesses, even in the macro
[19:08] economy, are going to be hurt by the AI
[19:10] disruption. I couldn't put it any
[19:12] better. Um I believe that to be the
[19:14] case. If you want to take it further, I
[19:17] think the thing that's going to The era
[19:19] of income without losses is ending. That
[19:21] is the big thing. The constant bailouts.
[19:23] We are not having bailouts. That's the
[19:25] biggest thing that I will take from
[19:26] Kevin Warsh is that the Fed's not
[19:29] cutting rates. So, if inflation, based
[19:31] on a lot of metrics, is slightly above
[19:33] where it was in 2019 before COVID, and
[19:37] yet the Fed funds rate is significantly
[19:39] above where it was, then we actually in
[19:42] monetary terms, why do we need to be
[19:44] doing any kind of big tightening unless
[19:46] we see the jobs market start creating a
[19:48] significant amount, which I don't see
[19:50] happening with AI. And at the same
[19:52] point, the inflation stuff
[19:54] is just not going anywhere right now,
[19:56] and the economy doesn't look great. The
[19:58] new loss cycle is not one giant default
[20:01] wave. It is a steady grind from
[20:02] borrowers whose old modes no longer
[20:04] protect them. A recession is about
[20:06] getting rid of the weak and letting the
[20:08] strong grow.
[20:10] Guys, that's exactly what's happening.
[20:12] It's just not happening in the way that
[20:13] I've said before. This is not about
[20:16] companies that borrowed money in the
[20:18] last period. This is about companies
[20:20] that have been borrowing money for way
[20:22] too long, that were bailed out. The
[20:24] zombies are going away.
[20:26] So, that is the signal versus the noise.
[20:28] Ignore the noise, and every time it
[20:30] comes up, you're losing money by
[20:32] listening to them. I cost myself money
[20:35] by listening to this in the energy side
[20:38] and believing a lot of what I was
[20:40] hearing, not to get bearish, but to put
[20:42] on hedges. And every time hedges have
[20:44] gone on, they've been wrong. Uh I got
[20:46] out of Micron in the 600s. It's above a
[20:48] thousand. There's plenty of areas where
[20:51] I've just been
[20:52] focused too much that this can't
[20:54] continue because this chart looks scary.
[20:56] It is a cliff chart. But, you know when
[20:58] people are bearish on this? In here. In
[21:01] here. In here. If I go show you every
[21:03] single person in X, I'd be done. And
[21:06] this is literally 100% wrong.
[21:11] Here are the movements, and again,
[21:13] this can't last. It's lasted, guys. You
[21:16] want to be long semi-short software. If
[21:19] you don't believe me, here's Adobe. How
[21:21] many people said this was stupid? Okay?
[21:24] It's either getting more stupid, or
[21:25] maybe
[21:27] myself using AI saying that these things
[21:30] are done. There's no reason to use Adobe
[21:32] when you can get free amazing visuals
[21:35] like the chalkboards I'm showing you
[21:36] that take 2 seconds to do as long as
[21:39] you're using AI. If you're not using it,
[21:41] I get it. You think that we still live
[21:42] in a world like this. Here's CRM,
[21:45] Salesforce,
[21:46] continuing to puke out. So, I want to
[21:48] get back to this.
[21:50] My thematic portfolio relative to the
[21:52] hyperscalers. So, I showed this last
[21:55] week. My thematic portfolio, about 20
[21:58] trillion in market cap. The
[22:00] hyperscalers, four of them.
[22:02] You're dealing with a market cap that's
[22:04] somewhere around 14 trillion.
[22:07] When you add in Nvidia, and you add in
[22:09] the other ones, you're getting up to 22.
[22:12] Regardless of what you want to do,
[22:14] I've given you multiple ways. So, this
[22:16] is the thematic portfolio, then there's
[22:18] the concentrated one with uh 10 names,
[22:21] and then there's the less concentrated
[22:23] but still concentrated 25 name, and then
[22:26] Morgan Stanley for all the RIAs, call up
[22:28] Lou Mandia, you know him, talk to him.
[22:31] We started doing this. We've done some
[22:33] swaps on it. We're going to do more on
[22:35] this, but they have my index, they have
[22:38] it there. We're starting to do more on
[22:40] it, and here's the alpha that's been
[22:41] created. Now, this is the one that they
[22:43] created, which equal weight this month.
[22:46] Um obviously, that reduces the weight of
[22:48] the massive winners, which have
[22:50] contributed more in terms of the way
[22:52] that I show it on here in Bloomberg cuz
[22:53] I don't reweight it. But, regardless,
[22:56] you're going to have an index that will
[22:57] be reweighted every quarter. Um this is
[23:01] it relative to the hyperscalers. So,
[23:03] again, at the beginning of the year,
[23:04] just look at the beauty of the trend in
[23:07] terms of the 50-day moving average just
[23:09] moving and here are the weekly moves.
[23:11] And the main thing I want to take again,
[23:13] when you look at Sortino ratios or you
[23:15] look at anything, look how many up
[23:17] weeks, these are weekly moves, are above
[23:19] 5%. You have one below. You now have
[23:22] five above. If you take it to 3 and
[23:25] 1/2%, you're dealing with about 10 to
[23:27] two. It doesn't really matter. The ratio
[23:29] comes out the same. And that is the
[23:30] alpha. That is what your capex is my
[23:32] opportunity, the benchmark arbitrage.
[23:35] And again, I'm going to say it again and
[23:36] again, I don't think this is changing at
[23:38] all. I am not I'm not only negative on
[23:41] the hyperscalers from a multiple
[23:43] compression basis.
[23:44] I think we're seeing some pretty
[23:46] gruesome signs now, which I'm going to
[23:47] go through. So, alpha, your a capex is
[23:50] my opportunity, that is the focal point.
[23:53] The AI spending boom is creating a
[23:54] depreciation time bomb. Google, Meta,
[23:56] Amazon, Microsoft have all splurged to
[23:58] secure podium spot in the race to build
[24:00] out the infrastructure that will run the
[24:02] artificial intelligence. The reason I
[24:03] wanted to show this, this is in the oil
[24:05] price side. If anyone knows what
[24:07] spending on stuff where price is going
[24:10] down and there's hypercompetition,
[24:12] it would be the fracking side. So, just
[24:15] keep it in the back of your mind, the
[24:16] issue inside. So, if we're looking at
[24:18] the hyperscalers and we're trying to
[24:20] figure out what the negative is, we know
[24:21] that somewhere in the future they hope
[24:24] to make money.
[24:25] Okay, great. Here's the problem.
[24:29] The issue is just growing rapidly. So,
[24:32] they have to spend an enormous amount of
[24:34] money to try and get
[24:36] the ROI they're getting
[24:39] inside Azure, inside all of
[24:42] the cloud. Um, the cash flow. So, I've
[24:45] given Peter some uh negative stuff. Here
[24:48] Here's one thing he put in there this
[24:50] week. Again, this is what the
[24:52] expectations are to come back. This may
[24:55] happen.
[24:56] But again, there's a lot of ifs and a
[24:58] lot of question marks and by 2030, I'm
[25:01] telling you you cannot predict what the
[25:03] world is going to look like. This
[25:05] spending here
[25:07] needs to happen for them to have a
[25:08] chance at this.
[25:10] It's very hard for them to not do this
[25:12] and have a chance at this because they
[25:14] are a big established company.
[25:16] Think about how fast Anthropic has
[25:18] grown. Think about how fast OpenAI has
[25:20] and I'm not going to say this
[25:22] too loud. Gemini has fallen way behind.
[25:26] So for all of you Google fans out there
[25:28] thinking that they're going to dominate
[25:30] this,
[25:31] I don't know what's going on,
[25:33] but the amount we've had 22 model
[25:35] upgrades since the Iran war
[25:38] uh Iran-US war started. 22 upgrades.
[25:41] We've had Opus 4.7, 4.8. We've had
[25:44] Mythos. We've had Fable 5. We've pulled
[25:46] it back. Gemini's had garbage since
[25:49] then. I don't even use it Nana Banana
[25:51] anymore because OpenAI has surpassed
[25:53] that. So I don't know what's going on in
[25:55] Google Gemini, but you start falling
[25:56] behind too fast and you don't you don't
[25:58] have models coming out when we're
[25:59] already looking at 5.6.
[26:02] I
[26:03] there's something going on and as I go
[26:04] through these, just remember the
[26:05] hyperscalers have to spend a lot of
[26:07] money and they have to make sure the
[26:08] money is going to certain places. So
[26:11] there's no way to look at this chart of
[26:12] free cash flow and not realize that this
[26:14] is a massive bet. It's all I'm going to
[26:17] say and that's why I want to be short
[26:18] these. They might end up winning,
[26:20] surviving. I don't think they're
[26:22] winning.
[26:24] Here's the Microsoft chart. This is a
[26:26] gruesome chart. This is not good. They
[26:29] are unchanged and basically since
[26:31] ChatGPT came out, they just keep going
[26:33] lower and lower. This is the 200-week
[26:34] moving average. I showed it when they
[26:35] broke it here. They just cannot get away
[26:37] from it. So
[26:39] like it is with Bitcoin, when you run up
[26:40] to technical levels and you fail again
[26:42] and you go right back down, that is not
[26:44] a good sign. Uh Microsoft fits in the
[26:46] same boat. This one,
[26:48] I can't believe the chart looks this
[26:50] good. I don't even know how Meta is
[26:52] alive in this point at with this. If you
[26:54] go back to 2024 and you go look at who
[26:57] led the open source model side, it was
[27:00] a name you never hear anymore, Llama.
[27:04] Meta hired all these people they paid
[27:06] billions of dollars a year ago. What do
[27:08] they have to show for it?
[27:09] I'll show you what they have to show for
[27:11] it.
[27:13] Um
[27:15] Alexander Wang, this is who they hired.
[27:17] Okay.
[27:20] It was a $15 billion hire.
[27:22] Um
[27:24] Meta's new AI unit is a total mess.
[27:26] Executives and employees alike are
[27:28] struggling with the chaotic AI strategy.
[27:31] This week in TechCrunch,
[27:33] uh Meta's month-old AI unit is a
[27:35] soul-crushing gulag.
[27:37] Now a new report, I mean, it's starting
[27:40] to spread throughout Silicon Valley. I'm
[27:41] a prisoner in the unit Meta calls the
[27:43] gulag.
[27:44] I build labels now. Before the queue, I
[27:47] was an engineer.
[27:50] CTO says morale is near the worst it's
[27:52] ever been during increasing snack
[27:54] budgets.
[27:55] Top Meta AI leader leaves just months
[27:58] after restructuring overhaul.
[28:01] Again, Meta, which company will have uh
[28:03] the number one AI model by December
[28:05] 31st? Like I said, Google is still up
[28:08] here.
[28:09] I don't know what that's all about.
[28:11] Uh Google and Meta may have the worst
[28:14] the most to lose in the AI agentic era.
[28:16] So, this is the thing that I want you
[28:18] guys to think about and just keep in the
[28:20] back of your mind um
[28:22] because I happen to agree with
[28:24] the way that people need to start
[28:26] looking at these things. The companies
[28:28] with the most to lose in an AI agent
[28:30] future could be some of the today's
[28:32] biggest technology platforms. Rather
[28:33] than asking which company develops the
[28:35] smartest AI model,
[28:37] he says the more important question is
[28:39] what happens when AI agents begin
[28:40] performing tasks currently handled by
[28:42] humans. This is why I focus so much on
[28:45] crypto
[28:46] and less on everything that has been in
[28:48] the past about what has worked. So, I
[28:51] would just keep it in the back of your
[28:53] mind. And this is a big change that's
[28:55] happening. So, in the long thematic
[28:57] portfolio short the hyperscalers, Micron
[29:01] could overtake Meta in stunning AI power
[29:03] shift. I have no doubt that is
[29:06] going to happen either from them or from
[29:09] more than them in terms of the hardware
[29:11] names.
[29:12] Fable five, if you guys haven't focused
[29:14] on it, I'm going to do a a lot of things
[29:16] on here and start connecting some dots
[29:18] for you on the other issue of the
[29:19] hyperscalers and why this token index
[29:21] thing is so big. So, Fable five, um I
[29:24] didn't get a chance to talk much about
[29:25] it last week because everything happened
[29:27] Friday night. But, there was a lot of
[29:30] great podcasts which I'm going to highly
[29:32] recommend you understand the importance
[29:34] of shutting down Fable five. First thing
[29:36] I want you to do,
[29:37] uh Leopold Aschenbrenner.
[29:40] So, he's gotten a lot of attention
[29:42] for his situational awareness hedge
[29:45] fund. But, just remember, he wrote a
[29:47] paper, situational awareness. So, his
[29:50] genius is in understanding what is
[29:52] coming.
[29:54] Inside the situational awareness,
[29:56] when you go to number four, the project,
[30:00] as the race to AGI intensifies, the
[30:01] national security state will get
[30:03] involved. Government stops Fable five.
[30:07] The United States government will wake
[30:08] from its slumber and by 27, 2027, 2028,
[30:12] we'll get some form of government AGI
[30:13] project. No startup can handle
[30:15] superintelligence.
[30:18] That is what we're getting to right now.
[30:21] So, he forecasted that in this.
[30:26] I'm writing a paper on it.
[30:28] His original essay no longer reads like
[30:30] science fiction or abstract futurism. It
[30:32] reads more like an early map of the
[30:34] world we're entering. And I want to get
[30:35] into that because there's a lot of other
[30:37] issues
[30:38] that show up which I'll get into by the
[30:40] end that he didn't predict yet. So, all
[30:43] in, very good podcast on this. I didn't
[30:46] care about the rest of it, but the
[30:49] backlash on Fable and what this all
[30:51] means.
[30:53] The panel argued that enterprise AI now
[30:55] face governance and single provider
[30:57] risk.
[30:58] Anthropic's approach is a warning that
[31:00] companies need model diversity and
[31:01] control. So, enterprises who were
[31:03] thinking of just using Anthropic, and
[31:06] especially the big the best model,
[31:10] if you're overseas, how can you do this?
[31:12] If anything was going to direct
[31:15] people to open source, this could push
[31:17] companies towards open source models,
[31:19] including Chinese models, because you
[31:21] just can't turn off AI. Basically,
[31:24] Freeberg said, "You can't regulate this.
[31:25] It's like the internet." And I
[31:27] completely agree, because everyone will
[31:29] just move, as I have already done, and I
[31:32] can't get the hardware necessary. I'll
[31:34] get into GLM 5.2,
[31:37] which came out. This is all a problem.
[31:40] So, if you're one of the model companies
[31:41] trying to get companies to pay for it,
[31:43] your prices are going through the roof.
[31:44] The last thing you want to do is have
[31:46] anything that scares off enterprises to
[31:49] say, "We need to move to Chinese models
[31:50] or some open source model to have it on
[31:53] our thing, to have control over it, not
[31:54] have to deal with that risk."
[31:56] It's an issue. At the same time, the
[31:57] cost of building a gigawatt, Chamath
[32:00] said it's now up to 100 billion. He has
[32:03] money at stake in this. He talked about
[32:05] this on there.
[32:07] And then Freeberg went through and
[32:08] rejected the AI destroys jobs narrative.
[32:12] Uh I'm completely in that camp, as I
[32:14] have been, but I am not in the hiring
[32:18] camp. I am in the we're just stuck with
[32:22] this whole scenario. He talks about how
[32:24] they need people, which is true.
[32:26] Again, what I said before with Dan
[32:28] Iveson at PIMCO, there is a recession
[32:31] happening. A recession from my
[32:32] definition is a rotation from the weak
[32:36] into the strong. AI native users are the
[32:39] strength. If you're a financial advisor
[32:41] working for an RIA that doesn't allow
[32:43] you to use AI,
[32:45] you're getting weaker by the day. If
[32:47] you're an RIA and you want to use AI,
[32:50] that is the reason why I'm spending time
[32:52] with places. You need to invest in it.
[32:54] You need to have an AI native mentality.
[32:56] You need to have this going on in your
[32:58] firm, and it can't just be handing
[32:59] people Claude code. That's not going to
[33:02] do it. Great article to read again on
[33:04] Fable, The Window Has Closed
[33:07] by Andrew Curran. It was I
[33:09] Everything about it to me just felt
[33:11] good, and there were two things. One is
[33:13] how incredible it is in terms of Fable
[33:17] 5. What I care about this, which again,
[33:22] we've reached a point that the models
[33:23] have gotten too good. That's the main
[33:26] thing you have to take. It becomes an
[33:28] issue. For the last 6 months, the money
[33:30] being made, which was not expected by
[33:33] any investor that I talked to, to this
[33:35] degree, is because they all joined into
[33:38] the agentic world. When you hit the
[33:40] agentic world, the next thing is
[33:42] recursive self-improvement. Recursive
[33:43] self-improvement, the models improve
[33:46] insanely [snorts] fast. That's how you
[33:47] can go from Opus 4.7, 4.8, Mythos. You
[33:51] start getting models faster. The places
[33:52] that are further away, they still have
[33:55] their models updating every 6 months to
[33:57] 9 months. So, think
[33:58] XAI at this point or Grok. Think Gemini
[34:01] at this point. The Chinese models are
[34:04] actually moving as fast. As open-source
[34:06] models approach frontier capabilities,
[34:08] the author expects a regulatory
[34:09] crackdown because governments will not
[34:10] tolerate that level of capability being
[34:12] widely widely available. A long AI is no
[34:15] longer just about coding benchmark
[34:17] copyright or productivity equals. It is
[34:19] now the most important technology in
[34:20] human history, and one that will change
[34:23] transform civilization itself.
[34:26] Absolutely the case. Okay.
[34:30] Now,
[34:32] back to the open source. So, again,
[34:35] there are tons of risk with this
[34:37] happening, and this is what Leopold did
[34:39] not predict. So, if you go back to his
[34:41] paper,
[34:42] yes, he predicted that the government
[34:44] would get involved as we get code closer
[34:46] to RSI. Once you get to RSI, you're
[34:49] right before AGI, but he did not think
[34:51] that open-source models could stay
[34:53] there. In fact, in his paper, he talked
[34:55] about the need for the government to get
[34:57] involved, so the Chinese specifically
[35:00] can't catch up.
[35:02] They already have caught up, and the
[35:04] competition means they can offer similar
[35:06] models
[35:08] at
[35:09] 10% of the price. That is the issue for
[35:12] the hyperscalers and all of the model
[35:14] companies is the ability for people to
[35:16] use this
[35:18] open-source. Uh they want control.
[35:21] Uh this is the Moonshots podcast from
[35:23] this week. Absolutely listen to this one
[35:26] as well. They talk about SpaceX. I'm
[35:28] sure most of you don't care. The
[35:30] Anthropic Fable 5 shutdown is treated as
[35:32] a major turning point. Enterprise trust
[35:34] in Frontier Labs is questioned. Altman's
[35:36] possible IPO delay. Codex setting its
[35:40] own goals. I'm going to go through why
[35:42] these are all important. First of all,
[35:43] on the Codex side,
[35:45] it's a shift from tax execution to into
[35:48] intent execution. Basically, the coding
[35:51] agents are doing the designing. So, when
[35:54] you put in where Codex is and that
[35:56] OpenAI is talking possibly about
[35:59] delaying their model, I'll get into the
[36:02] potential reasons why.
[36:04] But, Dave said this is a major moment in
[36:07] human history. Fable 5 and the
[36:09] government getting involved, it will
[36:10] only get worse from here.
[36:12] Um and this will become the new normal,
[36:14] as he said. The deeper issue is AI is no
[36:17] longer being treated like software. It's
[36:19] being treated like a strategic national
[36:21] security asset. I want you again to
[36:23] remember that every time you see AI as a
[36:25] bubble.
[36:27] Everyone who this, and most of them are
[36:29] older than me,
[36:31] I get it, or they're at the same age as
[36:33] me.
[36:34] The problem is they don't use it, guys.
[36:35] So, they're saying, "I've seen this
[36:37] innovation thing before."
[36:41] In terms of the IPO point, this is what
[36:43] Alex said, which I happen to think is a
[36:46] big risk for
[36:48] the CapEx trade.
[36:50] The deeper interpretation of OpenAI
[36:52] possibly um delaying their IPO is
[36:56] because recursive self-improvement is
[36:57] starting to work. Then technology may be
[37:00] beginning to substitute for capital.
[37:04] If it's improving, and again, I said
[37:06] this on Pomp, I'll say it to you guys
[37:08] now.
[37:09] These models are improving without the
[37:12] benefits of all the money being spent on
[37:13] CapEx right now.
[37:15] So, you have to understand that
[37:16] Blackwell and Vera Rubin, I mean,
[37:18] Blackwell's being run out. We just don't
[37:20] have many data centers that are finished
[37:21] yet. So, most of the improvements are
[37:24] actually coming from algorithmic
[37:27] efficiencies and from the models
[37:29] improving themselves. Now we're getting
[37:31] to the point where they're actually
[37:32] improving the models themselves as
[37:34] opposed to people figuring out reasoning
[37:36] and test-time compute and reinforcement
[37:39] learning and reinforcement learning with
[37:40] human feedback. All of those techniques
[37:44] and the Chinese using them and the US
[37:46] copying the Chinese ones and the Chinese
[37:48] doing new things and us get using more
[37:50] data.
[37:51] The major implications if the models are
[37:53] improving fast enough through algorithms
[37:55] and through agents and RSI, the market
[37:57] may eventually wonder whether the CapEx
[37:59] curve has been overestimated. I think
[38:01] about this every day, and the question
[38:03] is when will this become a narrative? I
[38:05] happen to agree with the fact that at
[38:07] some point there is a risk. I showed you
[38:09] the charts on Microsoft. I showed you
[38:11] the charts on Meta. If anyone starts
[38:13] believing, and I'll get into a little
[38:14] bit more Microsoft soon, you could end
[38:16] up with a a CapEx air pocket risk
[38:19] where the intelligence is moving so much
[38:22] faster than how they in infrastructure
[38:24] can be built, it may decide that
[38:26] building all that CapEx is not a good
[38:29] decision. So, the question is for the
[38:31] first time, I think there's a risk that
[38:33] some of these companies might change
[38:35] their CapEx, lower their CapEx. How will
[38:36] the market respond to that? I think
[38:38] we're getting closer to that point, even
[38:40] though in the long run, I think the
[38:41] numbers are going to get to 90 trillion,
[38:43] not the numbers we're talking about,
[38:44] because enterprises are going to have to
[38:46] build their own hardware. But,
[38:48] it's an issue. Fable five, that means
[38:50] other countries have to build their own
[38:52] data centers, and they have to have an a
[38:54] model in there, because they can't
[38:56] depend on the US allowing them to have
[38:57] the best model.
[39:00] Uh the singularity has gone open source,
[39:01] and it speaks Mandarin. Z A eyes GLM
[39:05] 5.2. This is from Alex Wizner Gross from
[39:08] Moonshots.
[39:10] This is where I translate into why the
[39:12] Fable five shutdown matters. So, I went
[39:15] through all of these different
[39:16] components in here.
[39:18] I highly recommend thinking about why it
[39:20] matters and why this is going on. But,
[39:22] one of the main reasons is it may
[39:23] accelerate sovereign AI and open open
[39:26] source uh
[39:27] adoption. And part of the reason is
[39:30] this model was released this week, beats
[39:33] GPT 5.5 on multiple long horizon coding
[39:36] benchmarks for 1/6 the cost.
[39:39] Every week there seems to be another
[39:40] one. The battlefield is different now.
[39:41] If open source is as good as frontier
[39:43] and people have cheaper alternatives,
[39:44] governments can't be as quick to
[39:45] regulate. It will destroy the frontier
[39:47] AI labs.
[39:50] Alex Finn, who uses all of these models,
[39:53] basically said it. They caught up. GLM
[39:54] 5.2 is as good as Opus 4.8. That is an
[39:57] insane statement, and if
[39:59] every metric and every benchmark is
[40:01] saying that, so again,
[40:06] open source 5.2 beats Anthropic I I'm
[40:09] just telling you, we've reached a point
[40:10] which is unbelievable. Now, that is a
[40:12] mixture of experts. So, think of it as a
[40:15] judge with a whole bunch of analysts
[40:18] going through things and then one judge
[40:19] making the final decision on things but
[40:22] using a mixture of experts. If you guys
[40:24] have done the prompt that I put up on
[40:26] the paywall about how to do investment
[40:28] ideas, I just did a new one on 800 volt
[40:32] DC.
[40:33] The names will go out this week.
[40:35] The main thing about it is I've always
[40:38] used a mixture of all of the models to
[40:40] come up and then one judge to go through
[40:42] it. That is similar to what GLM 5.2 that
[40:45] analogy is but so is Fusion API, the
[40:49] smartest compound model on the market.
[40:51] This is more in line with what I was
[40:53] doing which is using all of them. So now
[40:56] you've got Fusion which is taking not a
[40:58] mixture of expert, it's taking all of
[41:00] the best models through open router,
[41:02] taking the best answers and then putting
[41:04] them all together. So it's a compound
[41:05] thing and it's surpassing frontier
[41:08] performance. You're getting all of these
[41:09] ways to do things which in the end will
[41:12] be cheaper.
[41:13] Microsoft weighs DeepSeek for co-pilot
[41:16] work. Think about that.
[41:18] A Chinese company.
[41:22] Satya Nadella wrote this piece. I
[41:24] completely agree. You need to read this
[41:26] if you're running a business, especially
[41:29] understanding
[41:30] that for AI to work in your company, it
[41:32] needs to be an ecosystem and I
[41:34] completely agree and he's saying
[41:36] individual models will not be the
[41:37] solution and everyone's jumping in and
[41:40] leaping for Claude code and and
[41:42] Anthropic.
[41:44] Again, a year ago was OpenAI and chat
[41:46] enterprise. Everyone's just leaping and
[41:48] unless they figure a way to become AI
[41:50] native which is what the ecosystem
[41:52] basically describes,
[41:54] you're really going to be screwed on
[41:55] this. Give me a four sentence summary of
[41:57] this.
[41:58] So
[41:59] AI represents a unique shift.
[42:02] Everything in this and again, the human
[42:04] capital side think about as
[42:07] agency. The agency side is the human
[42:10] using the AI combining the power of the
[42:12] human being with the AI. This is why you
[42:15] have to have agency in doing this stuff
[42:17] yourself if you want to have a job, if
[42:19] you want your kids to have a job, but
[42:21] most importantly, if you want clients to
[42:23] help you out, there's no better way than
[42:24] to help them with the tools that I'm
[42:26] helping you with. Your clients will feel
[42:28] more powerful. This is about being
[42:30] empowered. The reason that the consumer
[42:32] confidence numbers are so low
[42:34] is not because of spending. It's not
[42:37] because of the ability to get a job, but
[42:38] it is the fear related to the job. It is
[42:41] the fear of politics. It is the fear of
[42:43] everything, and all of that, in my
[42:45] opinion, behind it is AI.
[42:48] Um does this fit his decision to
[42:50] possibly use Deep Seek?
[42:53] So, this is back to Satya. Yes, it fits
[42:55] extremely well.
[42:56] This approach avoids commoditizing
[42:58] knowledge or hollowing out value to a
[42:59] few providers. Instead, it distributes
[43:01] capability across the ecosystem, keeps
[43:03] Microsoft in control of the
[43:04] orchestration and data loops, and aligns
[43:06] with Nadella's earlier praise for Deep
[43:08] Seek. So, again, you're starting to get
[43:10] even the biggest, most sophisticated AI
[43:13] companies complaining about the token
[43:15] costs, and then thinking about using
[43:17] open source models for a variety of
[43:18] different reasons, but in particular,
[43:21] it involves owning it.
[43:24] Um here's the token index. It's still
[43:26] not going higher, and again,
[43:28] this has a lot of things going on it,
[43:30] but one of the things is more usage from
[43:33] Chinese models relative to US models.
[43:35] So, cheaper models. That's the way you
[43:36] get things to go down is that you're
[43:38] having higher growth in the cheap stuff
[43:40] versus the expensive stuff. So, we've
[43:42] seen a turn. The question is, will this
[43:44] translate? I think the earnings period
[43:46] for Q2
[43:47] could be a bigger risk than what
[43:48] happened in Q1, and I would be very,
[43:50] very wary again of the hyperscalers at
[43:52] this point. Microsoft also walked away
[43:55] from a $3 billion deal to lease Oracle
[43:57] cloud capacity over security concerns. I
[43:59] think all of these stories are just very
[44:01] critical into where we are.
[44:04] Um we're not at the same place we were
[44:05] in January 1, guys. so the the thematic
[44:08] portfolio is supposed to be diversified.
[44:10] It's supposed to be diversified because
[44:12] if the agency side picks up, but the
[44:14] data center side slows down, what does
[44:17] that mean? It is not good for the
[44:18] hyperscalers, which is the short side.
[44:20] If you guys want to do
[44:22] relative value trades, I I still believe
[44:25] the thematic portfolio relative to this.
[44:27] The agency side getting for the final
[44:29] part here,
[44:31] again,
[44:33] this is why the paywall was set up. It
[44:35] was for all three of those things. It
[44:37] was to keep you away from the noise
[44:40] by doing this every week and at least
[44:41] giving you the other side with a lot of
[44:43] facts. The second was the alpha, giving
[44:46] you the names where you can make money,
[44:47] whether you're a trader or whether
[44:48] you're a PM.
[44:50] At the same time, the signal was to help
[44:52] you specifically be able to use AI and
[44:55] figure out your own ways to make your
[44:57] business better. If you haven't done
[44:58] this yet, again, I put this out on
[45:00] YouTube. It's completely open to
[45:02] everyone. If you haven't started yet,
[45:05] how do I start? Start with that.
[45:07] Then go to the paywall, get the
[45:09] knowledge brain, build this.
[45:12] It will take 30 to 45 minutes with any
[45:17] computer knowledge. If you just upload
[45:19] the sheet into an LLM and say guide me
[45:21] to be able to do this, it will do it.
[45:23] And then if you run into a stumbling
[45:25] block,
[45:26] take a snapshot of the picture, put it
[45:28] into the LLM. Remember, it's visual.
[45:30] That whole process of being able to get
[45:32] this running, I'm telling you, you will
[45:33] learn more doing that than any class
[45:36] will do it.
[45:37] One thing someone did, and I'm sharing
[45:40] it with you guys cuz this is a
[45:41] crowdsourcing thing, they took all of
[45:43] these weekly video podcasts. So, these
[45:45] go on the paywall every single week. It
[45:47] is every podcast. So, this week I showed
[45:50] three podcasts on here that I thought
[45:51] were worthy with information for you
[45:53] guys to listen to. If you just go have
[45:56] the brain go out and say, "I want you to
[45:58] create going out and getting all the
[46:00] transcripts from everyone that Jordy has
[46:02] put. Just upload all of those files into
[46:05] one folder. Have it go out. You can
[46:07] create a brain on just those and get the
[46:08] information from it. Trust me, that kind
[46:11] of stuff will work. Um I did do
[46:13] something here that I uploaded. I
[46:16] created a
[46:17] so you guys can see it. A weekly focus
[46:19] lit uh list prompt for a pullback
[46:23] hunter. So, for you traders or any of
[46:26] you PMs looking for entry points on
[46:28] things during these consolidations or
[46:30] during these pullbacks, I put together a
[46:32] prompt, which is here, so that you can
[46:35] run it on
[46:37] the files that I give you every week.
[46:38] So, the exhaustion file, the hedge fund
[46:41] file, the technical sheet. You can go
[46:43] through here and just go through it and
[46:46] come up with the 20 names. So,
[46:49] this is the way that it comes up with a
[46:51] response and it will give you the
[46:53] details and give you a name by name. So,
[46:55] all you got to do is go take that prompt
[46:57] that I uploaded that I just showed you
[46:59] guys. Go have it run. It will give you
[47:01] an output like that in Claude co-work.
[47:05] Um and here's the prompt. So, basically
[47:07] you can see it's taking the technical
[47:09] sheet, the fundamental sheet, the
[47:11] scores. These are all the workbooks. The
[47:13] workbooks just mean the folder that
[47:15] you've put down into there. You can put
[47:18] all three of them into one folder. And
[47:20] then just go do it.
[47:22] Doesn't have to be everyone in history.
[47:23] Just has to be the last three.
[47:26] There's also this week a consolidation
[47:28] playbook going on. I'm going to do this
[47:30] once a month. This is the way I used to
[47:31] look within some of my portfolio to find
[47:33] names that I wanted to either increase,
[47:36] decrease
[47:37] after a consolidation. So, this is meant
[47:39] to show you names that have
[47:40] consolidated, but technically they've
[47:42] stayed structurally good. So, it has all
[47:45] 100 names. You can go through it, but
[47:47] basically it's going to go through the
[47:49] percentage relative to the 50-day, the
[47:51] 60
[47:52] day percentile. So, what percentile of
[47:55] this range over the course of the last
[47:58] 60 days is that in. There's a whole
[47:59] bunch of things in there. You will see
[48:01] it. Um, I put this out this week. It
[48:04] went up, I believe, on Thursday uh at
[48:06] the end of the day. The Vera Rubin 800
[48:08] volt, this has been a big theme. I've
[48:10] been asked by a lot of people. This is
[48:11] the introductory paper. This paper goes
[48:14] through so you can read about it and
[48:15] understand the importance of Vera Rubin
[48:17] and what this means. There are other
[48:19] names. Uh,
[48:21] I had this my system run for 40 names.
[48:25] Of the 40, 26 are not
[48:28] part of my current portfolio. I'm going
[48:30] to whittle that down, as well, and I'm
[48:31] going to try to find the ones that have
[48:33] the highest percent, and then I'll
[48:36] release it this week.
[48:38] Uh, again, this is the thematic
[48:40] portfolio. This is the end result.
[48:42] That's a bull market. This, for everyone
[48:45] waiting, and I'm waiting, too, uh this
[48:48] is a bear market.
[48:50] So, let's go back.
[48:52] The thematic portfolio in AI right now
[48:54] is a bull market.
[48:56] And really, when did it start in a big
[48:58] way? It was after this
[49:00] pullback here. It just took off, but
[49:02] this here,
[49:03] and on the flip side now, you've got
[49:05] Bitcoin, which cannot get any single
[49:08] movement whatsoever.
[49:10] This is basically the agentic world
[49:12] taking off here. There is no enthusiasm.
[49:14] I do not care about putting more money
[49:17] into something that is below the
[49:19] 200-day, that fails at every step. We
[49:21] failed again at this thing. Until we get
[49:23] positive momentum, I don't care. Since
[49:26] I've said it before, I think this
[49:29] this the next move is the big one.
[49:32] I could be wrong, but I'm not getting in
[49:34] until we break above the 200-day moving
[49:36] average. Uh, that's it for this week,
[49:38] guys. Uh, I will see you uh next week.
[49:42] Uh,
[49:43] I might be in New York to do the last
[49:45] one, but I think I'll be in Maine, so
[49:47] I'll see you from Maine.