While You’re Watching Oil Prices, AI Is Accelerating And Rewriting The Economy — Jordi Visser (5 abril 2026)

[0:00] Another uh
[0:03] another crazy week. Uh
[0:05] market changed uh as I talked about.
[0:08] Last week I thought we'd get a uh a
[0:11] rally and a bounce uh either Monday or
[0:14] Tuesday, and we did get a violent bounce
[0:16] on Tuesday.
[0:17] Uh oil prices continued higher, and I
[0:20] think that is a uh a clear signal that
[0:23] after 5 weeks of the market paying
[0:26] attention to every nuance
[0:28] on Iran, regardless of you know, what
[0:31] you're hearing in terms of doom and
[0:33] gloom and oil being this catastrophic
[0:36] thing,
[0:37] the market has uh now seen it. It's
[0:40] adjusted to it, and especially on
[0:42] Friday, we've reached a point where at a
[0:44] minimum, I think you can get rid of the
[0:47] narrative from people that every day oil
[0:49] goes higher, stocks will be down, and
[0:51] now the market is starting to
[0:53] uh break apart from the story, and I
[0:56] think that is going to be a reflection
[0:58] of uh what I talked about last week. So,
[1:00] as a reminder, last week the entire
[1:01] thing was separating the mindset of a
[1:04] bear market and more the mindset of a
[1:06] regime shift. There will be plenty of
[1:08] bull market places to invest in because
[1:11] AI is durable, because the consumer is
[1:15] uh
[1:15] completely different than 2007,
[1:18] and I use 2007 because that was the last
[1:21] credit cycle. Uh we are in a different
[1:23] time period where credit cycle is
[1:25] weakening, and I think that'll continue.
[1:27] AI is a massively important theme, the
[1:30] most important investment theme, and
[1:31] compute is uh
[1:33] we don't have enough compute. It's
[1:35] insatiable, which means every time that
[1:37] we get one of these corrections, of
[1:39] which I'm sure there will be be many,
[1:41] because on the negative side, we are in
[1:43] stagflation. Um and you don't hear me
[1:45] use that word too much because I think
[1:47] it's just a narrative,
[1:48] but the market cares about that
[1:50] narrative, and I think going forward um
[1:52] even though Iran and ships are starting
[1:54] to flow through, which means we're
[1:56] through the worst of stage one. The
[1:58] impact in terms of how long the delays
[2:00] are going to be has not been priced in
[2:02] yet completely,
[2:03] but you need to see the impact that will
[2:05] show up. The other thing about the
[2:07] consumer that is just different is the
[2:08] insane amount of net worth that's been
[2:10] created, the insane amount of transfer
[2:13] receipts that have grown rapidly since
[2:14] before the great financial crisis, the
[2:16] demographic shift that's happened,
[2:18] and the fact that it's just very
[2:20] difficult to have a significant rise in
[2:23] unemployment. We do have a labor
[2:25] shortage and this is coming from someone
[2:27] who believes that labor will continually
[2:28] be disrupted at the knowledge worker
[2:31] level, but not the daycare level, not
[2:34] the nurse level, not the teacher level,
[2:36] not the blah blah blah, go through it.
[2:39] It may not be a job people want, but you
[2:40] have to separate the two and we live in
[2:43] a world where the extreme opinions of
[2:45] people will guide your emotions towards
[2:47] making big mistakes. So, that that said,
[2:51] thank you to all of the financial
[2:53] advisors and RIAs that have reached out,
[2:55] particularly on the back of the model
[2:56] portfolio I released a spotlight
[2:59] spotlight piece on one of the names
[3:02] within sight the move to Blackwell and
[3:04] Vera Rubin
[3:06] as an investment theme that
[3:08] for subscribers, you guys saw it, the
[3:10] stock did very well this week
[3:12] and had a big announcement. I think
[3:13] you're going to see more of this going
[3:15] forward and I have a lot of things
[3:17] happening on the edge side. This is all
[3:18] the bull market side that you want to
[3:20] invest in. Unfortunately for most
[3:22] financial advisors, they are positioned
[3:24] in things like private credit and things
[3:26] like the Mag 7 and the hyperscalers
[3:29] which have been underperforming. So, I
[3:30] think the portfolio construction
[3:32] changes, the model portfolio, the things
[3:34] I've talked about are going to be more
[3:35] important. I look forward to spending
[3:37] more time in visiting some of the
[3:38] financial advisors. So, if you haven't
[3:39] yet reached out, go sign up. I promise
[3:43] you this is a time where you need to be
[3:45] spending time on how AI can help, but
[3:47] also the themes that are going to be
[3:49] very difficult for people to get
[3:50] invested in because they're just not
[3:52] positioned that way. So, up week,
[3:54] transports continue to send an important
[3:56] signal, stagflation.
[3:59] I wrote a paper on playing around with
[4:01] doge, private credit, the memory panic
[4:03] that went on, all of this stuff. Coming
[4:05] up, S&P up 3.4%
[4:08] for the week, the biggest up week of the
[4:11] year, and it breaks the five-week
[4:14] down move in seven of the last eight.
[4:16] So, we were due for a bounce,
[4:18] particularly coming out of the first
[4:20] quarter, and particularly with everyone
[4:22] being hedged and gross exposure
[4:24] shockingly remaining high.
[4:26] Uh
[4:27] so, you can see here just from the
[4:28] movements,
[4:30] you know, oil traded to the highs, and
[4:32] yet the stock market was not only moving
[4:34] higher, but as we go into a new month,
[4:36] especially one where pension funds had
[4:38] to rebalance in a very big way because
[4:40] of the underperformance of stocks, but
[4:43] also because earnings continue to grow,
[4:44] and the economy right now is fine. More
[4:47] money went into the market, and you saw
[4:49] it each day, you know, a low at the
[4:50] open.
[4:51] This panic morning after the Trump
[4:53] speech
[4:55] came back to 200-day moving average,
[4:56] we're still below it. You can see we're
[4:58] still way below the 50-day moving
[5:00] average. I'll talk more about that, but
[5:01] overall,
[5:02] it still, you know, looks like a a
[5:06] downtrend, and I think it still will be,
[5:08] and I would not be looking for a repeat
[5:10] of last year, which I'll go through. Uh
[5:12] one of my favorite ways to look at
[5:14] markets for mean reversion and just for
[5:16] scenarios, particularly in in either a
[5:19] violent bear market where people start
[5:21] to drastically worry about a recession,
[5:23] and all earnings revisions go lower. We
[5:25] saw that last year. Uh
[5:28] other years, it's not as bad. So, this
[5:31] is the percent below or above the 50-day
[5:34] moving average. Uh the red line here is
[5:37] the bottom decile, meaning we went
[5:38] through a level that you know, you only
[5:41] go through 10% of the time. Here's the
[5:43] upside. That level is about 4.3%
[5:47] on the downside. We've now bounced back
[5:49] up uh
[5:50] It normally takes a while to get back to
[5:53] the 50-day in these types of scenarios
[5:54] unless you're in a situation like last
[5:56] year. So, I'm going to show you two
[5:58] different situations. This is the one I
[5:59] think we're more in.
[6:01] Uh and it's not to say that we will have
[6:03] uh a year that looks like this, but I
[6:05] think there's going to be a lot of these
[6:08] ebbs and flows during bad news.
[6:10] Remember, this is 2022.
[6:12] This is when inflation peaked. Um it
[6:14] peaked around here.
[6:17] Uh year-over-year CPI, and I've said
[6:19] this is when I think the peak will come
[6:21] the bottom will come in most likely, at
[6:22] least for this year, will be when we
[6:24] peak year-over-year CPI, which I think
[6:26] at this point is likely to happen May,
[6:29] June, maybe even later. We'll see, but
[6:32] as long as we haven't had a peak in CPI,
[6:34] I think that's going to be an issue. Uh
[6:36] but this is the way you look. You have
[6:38] these
[6:39] falls, rally, falls, rally. You're
[6:40] bouncing around the 200-day moving
[6:42] average. 2022,
[6:44] we never saw earnings go negative. Yes,
[6:46] we saw revisions move lower, and yes,
[6:48] people started to panic. It was a bad
[6:50] year for tech. Inflation fears were
[6:52] running high. The Fed was raising rates.
[6:54] This is not the same year, but you have
[6:56] that kind of year where every time you
[6:57] rally, you start to run into trouble.
[6:59] And then you have last year where people
[7:01] started calling for a recession.
[7:02] Everyone bailed out of the market
[7:04] because if there's a recession, you
[7:05] don't want to be in stocks cuz earnings
[7:06] are moving lower. Quant strategies go
[7:08] the same way.
[7:10] And then he rolled back the tariffs, and
[7:12] we got this immediate rise.
[7:14] And that's because the analysis was
[7:16] completely wrong by the street saying
[7:18] that we'd have high inflation and that
[7:19] there'd be a recession. They couldn't
[7:21] have been more wrong. They were using
[7:23] the information of 100 years ago to make
[7:25] their forecast and do some analysis,
[7:27] which again was 100% wrong.
[7:31] And now this year, they're not worried.
[7:33] So again, pick your poison. Um here's
[7:35] the reality.
[7:37] Last year, this is what 1-year TIPS
[7:39] breakevens did. They peaked before
[7:44] the tariffs and liberation day. And then
[7:46] they went down.
[7:48] This time when your inflation
[7:49] expectations
[7:51] are up high.
[7:53] Not a little high either. We're up at 5
[7:55] and change percent where last year we
[7:57] went down the entire year. This is the
[7:59] difference between this year and last
[8:00] year's even when tariffs were going in,
[8:02] people were worried about a recession.
[8:04] This year, they're not worried about a
[8:06] recession. They don't think earnings are
[8:07] going to change and that's where the
[8:08] risk is is that the expectations haven't
[8:10] changed. Here's the correlation break
[8:12] between oil and S&P. You had every
[8:14] single person and especially every
[8:16] strategist who's bullish basically
[8:19] saying, well, it's all about oil and I
[8:21] I'm I don't have anything to do with
[8:22] this and blah blah blah. When the
[8:24] reality was again, the S&P was breaking
[8:26] down way before the war. Financials had
[8:28] already been hit. Software had been hit.
[8:30] Everyone was trying to excuse the AI
[8:31] trade. The AI trade is what's driving
[8:33] everything and and strategists and
[8:35] economists don't know how to deal with
[8:36] it and I think that's going to remain
[8:38] and that is the reason why you guys
[8:39] watch this. That is the reason why I
[8:41] spend all my time on AI. So hopefully I
[8:43] can bring you the news and not bring you
[8:45] something that's not based on people
[8:47] using it because the people who are
[8:48] telling you what to do that have a lot
[8:50] of experience, they have no idea what
[8:51] they're talking about when it comes to
[8:52] artificial intelligence because they
[8:54] don't use it. Um NDX, similar type
[8:57] story. Again, below the 200-day moving
[8:59] average but bounces. Uh these look like
[9:02] beginning of month flows. Uh but you've
[9:04] broken the correlation which means I
[9:05] think people are going to try and buy
[9:07] the dip uh unless we get some kind of
[9:09] new news. But oil price is going higher
[9:11] in the short term is clearly not new
[9:12] news at this point. So, we had a huge
[9:15] move of oil on Friday. Let's assume we
[9:16] come back on
[9:18] Tuesday morning and oil peaks and starts
[9:20] to trade even slightly lower. You know
[9:22] what stocks are going to do just because
[9:23] they didn't react to stocks being
[9:24] higher. Once you break that negative
[9:26] feedback loop uh from a bad news is good
[9:29] news or bad news is no longer bad news,
[9:32] usually there are dip buyers and I would
[9:34] expect that we're going to have this.
[9:35] So, uh in terms of levels for the S&P
[9:38] I'm definitely looking uh you know
[9:39] anything above 6600 which is where we
[9:43] are
[9:44] up to 6670 I think it's going to be a
[9:46] very challenging area to go the 50 days
[9:48] all the way up to 6750 so use those as
[9:51] barometers the Russell 2000 has traded
[9:53] much better than the rest of the market
[9:55] it did make one daily close below the
[9:56] 200 day
[9:58] but overall
[9:59] the PMI trades are still working now I
[10:01] want to take you through a little Dow
[10:02] theory um
[10:04] so I'm not going to go through the whole
[10:05] Dow theory but the Dow theory was
[10:07] basically to use the Dow industrial
[10:09] utilities the industrials utilities and
[10:13] the transports to get a sense as to how
[10:14] the economy was doing that was obviously
[10:16] during a time
[10:17] where the industrial revolution and
[10:19] everything was going on so it doesn't
[10:22] matter as much the Dow industrials now
[10:24] have tech names in it as well but here's
[10:26] the thing the Dow again
[10:29] is sitting right near the 200 day this
[10:32] has been a you know a a fall but it has
[10:34] not looked as scary as the tech world
[10:38] utilities um
[10:40] despite rates moving higher utilities
[10:42] have acted great
[10:43] this is uh data centers this is usage of
[10:46] AI this is electricity demand is going
[10:49] higher most importantly transports look
[10:52] great
[10:54] as long as transports are good nominal
[10:55] GDP is good end of story um transports
[10:59] are good and you know someone
[11:01] highlighted to me in my trip to Boston
[11:03] this week to look specifically at the
[11:05] transports and make sure I did some work
[11:07] on them and I've I've used them a lot
[11:09] it's one of the sectors that I think is
[11:10] going to surprise here is year-over-year
[11:13] transports with PMI PMI to me will
[11:16] remain high despite the oil situation uh
[11:20] and we will still head up towards 60
[11:22] over the course of the year in my
[11:23] opinion in terms of the PMI here's a
[11:25] year-over-year transports related to PMI
[11:28] more importantly uh
[11:30] the flatbed market was what I was
[11:32] pointed out.
[11:34] Including, you know, stocks like
[11:35] Landstar and things like that.
[11:38] Uh, in terms of just the flatbed, so
[11:41] big, big, big transport of materials and
[11:44] equipment related again to not only the
[11:47] data center build-out, but the mining
[11:48] side, the energy side, anything that's
[11:49] going on. We still have a nominal GDP
[11:51] story, and I just want you to look at
[11:54] So, this is uh, the the blue line here
[11:56] is during the tran- is during last year,
[11:58] and you could see that as inflation
[12:00] expectations went down, we also saw the
[12:02] PMIs go down until we got in towards the
[12:05] final quarter, and then it started to go
[12:06] higher. This red line here is 2026. We
[12:09] are way above this. This five This is
[12:11] the five-year
[12:13] average. So, you can see that flatbed
[12:15] rates are incredibly strong.
[12:18] And then you had Craig Fuller, who la-
[12:21] over the course of the last few years
[12:22] has called for, you know, a a transport
[12:24] recession or depression in cases, and he
[12:27] was on CNBC this week. Flatbed which are
[12:30] almost always industrial or raw
[12:32] materials are seeing some of the
[12:33] strongest volumes and str- and rate
[12:35] strength
[12:36] in history.
[12:38] So, how can you be bearish for a
[12:41] recession when you have that going on?
[12:43] These are not short-term duration
[12:45] things. These are not based on the
[12:46] consumer. This is based on nominal GDP.
[12:48] This is the AI trade, 100% related to
[12:51] it. I would not get negative on every
[12:54] part of the market. I think we're going
[12:56] to have some sectors that have trouble,
[12:58] and this is one of them. Financials,
[13:01] again,
[13:02] you've had a bounce, but not a big
[13:04] bounce, and you're way below a 200-day,
[13:05] which is pointed downward. No need to go
[13:07] in and pick that up. In terms of the S&P
[13:09] bounce, everyone talked about how
[13:11] historic it was and big, and they were
[13:13] It just shows how much
[13:15] memory changes. So, in 2022, during a
[13:17] bear market, look how many times we had
[13:20] jumps as big as the one we had on
[13:22] Tuesday. I mean, these were continuous.
[13:25] Uh, liberation day, obviously. Now, this
[13:27] was a low
[13:29] This was an important low in the market.
[13:32] Some of these in here in '09 were lows.
[13:35] The only reason I want you to be wary of
[13:37] that is when those lows have occurred
[13:40] we get big upside to downside volume
[13:43] days, and we did not get that on
[13:45] Tuesday. Upside to downside volume was
[13:47] shockingly low.
[13:48] These This is the S&P. You can see that
[13:50] the low was made on this line here. Big
[13:52] upside down. These are thrust days. Have
[13:54] you guys ever hear of thrust days where
[13:57] you get big volume expansion? That
[13:59] usually means that that is a time to
[14:01] basically go, and you can see how it's
[14:03] called every major bottom it did in
[14:05] 2022.
[14:06] So, when we make a bottom this year, I
[14:08] fully expect it will be an upside to
[14:10] downside volume blowout. You guys can
[14:12] look at that on your screen. What we
[14:13] have in a bear market is factor
[14:15] rotation. It has been destruction, and I
[14:17] believe this is the beginning of a
[14:19] massive long-term deleveraging phase.
[14:22] I've said it before. I said my surprise
[14:24] risk from last year would be that we
[14:25] would lose a very important hedge fund
[14:27] this year because of the leveraging. I
[14:29] do think we are going to be in that
[14:31] mode. If it does happen, you're going to
[14:32] see factor volatility stay high.
[14:35] The covariance model the turbulence
[14:37] model that I built was meant to show me
[14:39] when the P&L is high. That still remains
[14:41] as having these blistering days, uh,
[14:44] which means you never get to take a
[14:45] break. It's tiring. It is exhausting,
[14:48] but gross leverage on every PB sheet I
[14:50] saw was shockingly still at the highs.
[14:52] Net leverage had come down rapidly. We
[14:54] had hedges on, which is why bounces are
[14:56] going to happen, but I have not seen
[14:58] gross leverage come down at the entire,
[15:01] uh, index level, especially for
[15:02] long-short hedge funds. I think that
[15:04] needs to come down because I think we're
[15:06] in a different regime.
[15:08] The pressure from AI will remain. The
[15:10] inflationary pressures will be there the
[15:11] entire year, although they will peak.
[15:14] And we're going to be in a point where
[15:15] the Fed's going to have to do something
[15:17] because the credit cycle is real. I will
[15:19] say this over and over again to you
[15:20] guys.
[15:21] You're not paying enough attention to
[15:23] the credit cycle. We haven't had a
[15:25] credit cycle honestly since 2007.
[15:28] You could say we had one in Europe, uh
[15:31] but in the US
[15:33] I don't consider 2015-16
[15:36] a uh
[15:37] a credit cycle. It's very isolated to
[15:39] energy.
[15:41] High yield, this one is different and
[15:43] this is 17-year build-up and we have
[15:46] people trapped trying to get out. This
[15:49] is a credit cycle and when a credit
[15:51] cycle is unwinding it will gradually
[15:53] have more and more news that will pop
[15:54] out. So, don't get caught uh thinking
[15:57] that these problems, AI's disruption,
[15:59] structural. Inflation, especially from
[16:01] commodities, structural.
[16:03] Nominal GDP, structural from the dollars
[16:06] of AI and the credit cycle unwind,
[16:08] structural. That is moving forces and
[16:10] it's going to put the Fed in a very
[16:12] challenging position uh as the year goes
[16:15] on. I didn't didn't describe this chart,
[16:17] but just so you see it. Even with the
[16:19] moves last week, S&P realized 60-day vol
[16:22] is still incredibly low. Until we get
[16:25] this moving higher, this is why the
[16:26] upside the downside volume is not high
[16:28] because it's a rotation. Here's the
[16:30] factor volatility. This is value, this
[16:32] is momentum, and this is growth.
[16:35] As as an equal weight, you're sitting up
[16:37] here. You haven't had not the move. This
[16:38] is about screwing long short and market
[16:41] neutral uh
[16:42] and it's going to continue because of
[16:44] the rotation that is fueling this.
[16:49] This is just highlighting something that
[16:51] I think means we're in now the back and
[16:53] forth. It's going to be very
[16:54] challenging. I don't think you can play
[16:56] and buy puts and just go through this. I
[16:58] think it is a very challenging time. Um
[17:00] this highlights that the bull-bear
[17:01] ratio. This is the one I care the most
[17:03] about. Uh this is the newsletter one.
[17:06] It has now gone to levels which indicate
[17:08] and Ed Yardeni does a great job with it.
[17:11] So, he's reducing his recession call
[17:12] because now everyone's getting bearish.
[17:14] I don't think there's a recession, so
[17:16] I'm not moving things around. But, I do
[17:18] think with this being down here,
[17:21] so much pessimism usually instigates a
[17:23] surprising bullish government policy
[17:25] response, an unexpected rebound in
[17:27] stocks.
[17:28] Again, I'm not saying the war stuff is
[17:32] that, but I do think Ed Yardeni's point
[17:34] is valid. When that ratio gets down
[17:35] there, it usually gets harder to play
[17:37] for the downside. It doesn't mean the
[17:39] the S&P can't be down 20% at some point
[17:41] this year, but it just means that to get
[17:43] to that point, you're probably going to
[17:44] be down 12, then only down eight, then
[17:47] down 13, then only down seven. It's a
[17:49] trader's market. You need to turn over
[17:51] your portfolio, have more cash on, and
[17:54] just move the portfolio around. The
[17:55] opposite of the way the position seems
[17:57] to be street with gross near highs. Uh
[17:59] Goldman pointed out the long-short ratio
[18:02] was down here. Uh Charlie McElligott,
[18:05] you know, said that they were hedged
[18:07] based on skew. Either way,
[18:10] I think everyone buying puts and trying
[18:12] to play this from the downside is going
[18:13] to have a hard time. Uh I think, and
[18:15] I've said this before, you can have vol
[18:18] convexity on.
[18:20] Uh that's the way I'm doing it for my
[18:21] own PA. I took As I said last week, I
[18:23] took off
[18:24] uh a quarter to a third of my VIX uh on
[18:28] the Friday before. I didn't do anything
[18:30] this week with it, except for buy more
[18:32] uh more stocks of the ones that I like
[18:34] cuz especially memory and the name that
[18:37] I referenced in the in the spotlight
[18:39] piece this week. Uh these are areas
[18:41] where you got in a good chance uh to buy
[18:44] stuff. Uh
[18:45] you know, the name that I talked about
[18:47] uh for the subscriber list was was down
[18:50] uh big on Monday and then finished the
[18:52] week up very big from there, over 15%.
[18:55] So, you do have opportunities to stay
[18:58] into things, and I think you're going to
[18:59] have that all year. Now, the stagflation
[19:01] signal. And just so you guys see this, I
[19:03] did this on my phone with Claude. Again,
[19:06] for those of you not using this and not
[19:07] building stuff, I didn't ask it to put
[19:09] it in this form. All I said was go back
[19:12] uh
[19:13] and give me all of the data of the ISM
[19:15] back to the PMIs back to
[19:18] uh 19
[19:19] 70.
[19:21] Go through it and show me how many times
[19:24] we've had a
[19:26] prices paid component above 75
[19:29] with an employment number below 50.
[19:32] Your classic stagflation
[19:34] signal. And I said exclude the COVID
[19:36] period uh because the signals were so
[19:39] messed up. And basically where we're at
[19:41] is uh there's been 32 matches, three
[19:45] errors. Okay? The oil shock, basically
[19:48] they were all in the 70s except for 2018
[19:50] when tariffs trade war went in.
[19:52] And it just says as you go in in that
[19:55] one the S&P fell 48% during this one it
[19:58] fell 30% more than 30% and it even in
[20:01] 2018 we've got down 20%.
[20:04] So, you guys can read it, but the
[20:06] problem is in stagflation you're worried
[20:09] about both the recession and you're
[20:11] worried about inflation. It makes it
[20:12] very challenging and you have a lot of
[20:15] rotation and a lot of winners and
[20:16] losers. Um
[20:18] it's an issue and I think it's going to
[20:19] be there.
[20:20] Uh Goldman cut their global growth
[20:21] forecast. As you're thinking about the
[20:23] economy in the US and I'm showing you
[20:25] the transports are great and nominal GDP
[20:27] is great. Remember a lot of the revenues
[20:28] that come in the S&P 500 come from
[20:29] overseas.
[20:31] Asia has a lot more problems from energy
[20:34] than we do and they've got shortages.
[20:36] They're having to already conserve
[20:38] things. It's a mini version of COVID and
[20:40] I think you're going to see revisions
[20:42] come down from companies that are
[20:44] global. 40% of the revenues in the S&P
[20:46] 500 come from overseas. So, even if we
[20:48] only see a mild reaction here in nominal
[20:51] GDP, we will definitely see a revenue
[20:54] hit from overseas.
[20:55] Uh JP Morgan and PIMCO both say the bond
[20:57] market is misjudging the slowdown risk
[21:00] saying that higher rates and the fear of
[21:03] the Fed reacting to inflation is
[21:05] probably going to be overdone. This is
[21:06] the confusion that we'll go in.
[21:08] Australia released their manufacturing
[21:09] data this week. Big reversal, posted its
[21:12] first contraction in 5 months as demand
[21:14] weakened and cost pressures surged. So,
[21:16] another PMI side. German inflation
[21:18] headed for the highest in more than a
[21:19] year.
[21:21] The global earnings revision ratio
[21:22] finally started to turn down. Important
[21:25] stuff for everyone. Um DOSE 2.0, debt,
[21:28] oil, growth, employment, and why Bitcoin
[21:30] was created. Uh for those of you who
[21:31] don't care about Bitcoin, I would still
[21:33] read the paper. Um and the reason is
[21:35] it's main point is that this is a
[21:36] completely unique situation for the Fed.
[21:39] Uh they're going to be in a very
[21:41] difficult position this year,
[21:42] particularly if the agentic world starts
[21:44] to increase the job losses, which I
[21:46] think it will.
[21:47] If we're having months with negative
[21:50] negative job creation,
[21:52] if we're seeing pressure from gas at the
[21:53] pump hitting consumers, which we're
[21:55] obviously going to see in the K-shaped
[21:57] economy,
[21:58] what are they going to do? Are they
[21:59] going to focus on the temporary side of
[22:01] inflation, or are they still going to
[22:02] cut rates? And then you add in that a
[22:03] new Fed chair is coming in, and you've
[22:05] got pressure from the White House. So,
[22:08] here's the reality.
[22:10] Every one of these red periods is a
[22:12] recession.
[22:13] Here's where we are in job creation.
[22:15] It's going back and forth. There's been
[22:17] many months of negative jobs.
[22:21] It's a very difficult situation cuz this
[22:23] is not a recession. Um
[22:26] but it is pressure. Uh
[22:28] and for everyone who's saying, "Well,
[22:30] this is immigration, blah blah blah." It
[22:31] is not just immigration. The jobs market
[22:34] is weak. I'm
[22:36] like sick to death of listening to
[22:38] economists look at Look, it's a little
[22:40] bit better than it was last time. So,
[22:42] this is the survey of jobs, plentiful
[22:45] versus hard to get. It's still in a
[22:47] downtrend. You've got temp employment,
[22:49] which is this line, making new cycle
[22:51] lows. The white line here is the quits
[22:54] rate. It cannot bounce. People are
[22:56] scared to quit their jobs for fear of
[22:58] not being able to get a job. The New
[23:00] York Fed, time it takes to get a job. I
[23:03] don't care what it is. This is the
[23:05] unemployment rate. Every pressure in the
[23:07] employment market is weak.
[23:09] But the problem is, remember,
[23:13] the Fed is in a unique situa- Here's
[23:14] stagflation.
[23:17] This is debt to GDP. The Fed can't raise
[23:20] rates this time, guys. It's just not
[23:22] going to be able to do it because of
[23:23] where we are. When they did it in 2022,
[23:26] we were creating 4 million jobs this
[23:27] year that year. We had a housing market
[23:30] that was running amok.
[23:32] Housing was going off the rails in terms
[23:35] of people buying.
[23:36] We have completely the opposite
[23:37] scenario, but we still have debt to GDP
[23:40] at 122% right now. So, when you're
[23:43] comparing this period, just remember, in
[23:45] the '70s when they were fighting
[23:46] inflation, when Paul Volcker made the
[23:47] decision, debt to GDP was all the way
[23:49] down here. 2022 does not count. They
[23:52] aggressively tightened, and then they
[23:54] started to aggressively ease the other
[23:56] direction. The reason they tightened,
[23:57] and remember, they stopped earlier than
[23:59] where you would have expected, is
[24:01] because they had the room to do it, but
[24:03] that was a quick thing. It's not the
[24:05] same situation. We still have a budget
[24:08] deficit sitting at the same level it was
[24:10] when they were doing this before. These
[24:12] are all issues that are very, very
[24:14] different than the past. And most
[24:16] importantly, the reason they can't raise
[24:18] rates if they raise rates, they increase
[24:20] the risk of a recession
[24:22] because it's going to hit the stock
[24:24] market. And we learned our lesson last
[24:26] year when this was happening in 2025,
[24:28] and I was on every single week saying,
[24:30] "They cannot
[24:32] do what he's saying. They cannot throw
[24:34] this into a recession because then the
[24:36] debt will get worse because they can't
[24:38] go in with this combination of leverage.
[24:40] Debt to GDP 122%. The stock market,
[24:43] which has allowed them to do what
[24:45] they're doing to continue to have this
[24:48] economy go forward, it has been about
[24:50] since the Great Financial Crisis when we
[24:52] were at 50%. Yes, we are now at 220% in
[24:56] terms of the market cap relative to GDP.
[24:59] The ownership of stocks, guys,
[25:02] here's household assets as a percentage
[25:04] of their owner like we're at 50 This is
[25:07] This is a dangerous level to be in. This
[25:09] is why
[25:11] I'm telling you the overall market is
[25:13] going to have headwinds. We are in this
[25:15] trap now
[25:16] and this is not a new trap. I said this
[25:18] would be a year of a physical upgrade
[25:19] and that we would have shortages. This
[25:21] is a bull market for commodities and as
[25:23] Jeff Currie says so eloquently, you
[25:26] can't print molecules.
[25:29] So, we're in a unique situation and the
[25:31] oil energy situation, regardless of your
[25:33] view, it has a new floor, guys. That is
[25:36] the reality of this is there's a new
[25:37] floor. We need energy independence
[25:39] regardless of how you think this will
[25:41] end. It is very It is impossible for the
[25:44] Strait of Hormuz to go back to what it
[25:45] was. It is impossible for every place
[25:47] that has been blown up and there were
[25:49] still bombs going off from Iran and into
[25:52] Iran at energy side. There is a delay.
[25:56] If we had a hurricane that came in and
[25:57] destroyed every single energy situation
[26:01] in Louisiana and Mississippi, we'd be
[26:03] talking about it. This is going on right
[26:05] now and it just means we're going to
[26:07] have a higher energy. Doesn't mean a
[26:08] recession, but it does mean we're going
[26:11] to have pressure on fertilizer and all
[26:13] kinds of things, helium as you've heard.
[26:16] It's going to have impacts in terms of
[26:17] price. So,
[26:19] this one, which is the physical world.
[26:22] So, forget the paper world, which is the
[26:24] futures. This is what's starting to
[26:26] happen. We're up to 141.
[26:28] In March,
[26:30] we were at 65. This is more than a 100%
[26:33] increase. So, yes, the futures closed at
[26:36] only 111, but if you notice the spread
[26:38] between Brent and WTI collapsed to zero.
[26:41] At the same time, this keeps going
[26:43] higher. This is the physical world.
[26:46] People are demanding and needing oil at
[26:49] any price. The futures market and that
[26:51] they need it now.
[26:53] So, the spot market is telling you that
[26:55] we're going to see inflation higher than
[26:56] what people think. The global wave of
[26:58] energy rationing, that means growth is
[27:00] going to suffer.
[27:02] South Korea weighs first driving curb,
[27:04] driving cutting off, these are
[27:05] COVID-like things.
[27:08] Here is diesel prices in the US.
[27:11] All the way back, look at this spike.
[27:13] This is bigger than 2020 than the COVID
[27:16] situation of the shortage and remember
[27:17] that a lot of this
[27:19] is the back of Russia. Here's gas at the
[27:22] pump which made it to 408, but diesel is
[27:25] up much much higher. So, again, just
[27:27] keep this in the back of your mind as
[27:30] you're thinking about things.
[27:32] We're going to have
[27:34] more inflation. Now,
[27:37] I completely agree with this, so you're
[27:38] not going to hear me talk about
[27:39] recessions cuz I don't believe it
[27:41] matters. The fact that oil is still
[27:44] lower than it was in before the great
[27:47] financial crisis in May of that year
[27:49] when the S&P was much much lower and the
[27:51] economy was much smaller.
[27:53] Oil is lower than it was then. This is
[27:56] not going to take down the economy.
[27:58] Don't listen to what you're reading on
[28:00] TV and if you go out to a macro dinner
[28:01] and someone says, "This is just like
[28:03] COVID." This is not just like COVID, but
[28:06] it is inflationary and it will lead to a
[28:08] slowdown. It's not a stopping, it's a
[28:10] slowdown.
[28:13] Blue Owl, if you thought it was over,
[28:16] it's not over. Um
[28:18] So, this was the big news on Friday.
[28:21] They're It's hard to see here, but uh
[28:24] they Investors asked to pull 22% out of
[28:27] one fund and 40% out of another.
[28:32] Here are the numbers, guys.
[28:34] You don't need to be a math wizard to
[28:36] see what's happening. This all started
[28:38] in terms of people wanting their money
[28:40] back from these private credit funds.
[28:42] Here are the redemption notices. Now,
[28:44] remember, they have a 5% limit. So, the
[28:46] reason this became such a big story is
[28:48] because we've had to halt redemptions
[28:50] and actually they've had to step in and
[28:53] provide their own capital. So, that's
[28:55] what was going on. Blue Owl then put in
[28:57] more capital, but what they did is they
[28:59] sold off 1.4 billion dollars
[29:02] to an insurance company they're
[29:03] connected to and two pension funds who
[29:05] already have invested in some of their
[29:06] funds as the Wall Street Journal pointed
[29:08] out. Here are the numbers. They're not
[29:10] getting better and I don't see this is a
[29:12] run that's happening. These numbers are
[29:14] staggering. It will lead to them now
[29:17] needing to sell bonds. So, in the same
[29:19] way that in the energy market you've
[29:20] heard what was eight what kept the
[29:22] futures from breaking out was we had a
[29:24] lot of oil on the water. We went through
[29:26] all of the excess capacity of stuff that
[29:28] was there, but now we're at the point
[29:30] where we're starting to go through
[29:31] inventories. Starting to release things.
[29:33] That's why you're starting to see
[29:35] curbing you're starting to curb the
[29:36] demand side. Well, in this case
[29:39] they're going to have to sell some of
[29:41] these bonds off. When you sell the bonds
[29:42] off you're taking marks on them. When
[29:44] those marks happen, other funds take
[29:46] marks. Pension funds take marks.
[29:48] Insurance companies take marks or
[29:49] they're supposed to at least and you
[29:51] start to get something that reminds me a
[29:53] lot of every credit cycle. So, don't
[29:55] minimize the credit situation. The Wall
[29:58] Street Journal is another financial
[29:59] crisis lurking private credit. It's by
[30:01] Greg Ipp. You should read all of these
[30:03] ones that I'm pointing out.
[30:05] This is starting to become a bigger
[30:07] story. I've talked about it here.
[30:08] Everyone I talk to on the credit side
[30:10] who's smart knows about it.
[30:13] Many, many, many of the names that I
[30:14] just showed you on the private equity
[30:16] side completed purchases of insurance
[30:18] companies over the course of the last
[30:19] five years. They have captive insurers.
[30:21] Part of their distribution is through
[30:23] private wealth channels and part of
[30:25] their distribution is to insurance who
[30:26] stick these on their balance sheet.
[30:28] Well, the IMF warned
[30:31] in the fall on inflated credit ratings
[30:34] on life insurers private credit holdings
[30:36] which could result in default far
[30:37] exceeding predictions in a downturn.
[30:39] The Treasury started to meet with
[30:41] insurance regulators. You have to
[30:43] connect these dots and just realize that
[30:45] this is going on. Treasury smart,
[30:48] Fed is smart. I'm sure some of the Fed
[30:49] people are watching this.
[30:51] This is a mess. Um
[30:53] it's not systemic, but it is a mess and
[30:55] it is a credit cycle and it comes at a
[30:57] time when inflation's going higher.
[31:00] Uh the banks, we start have earnings
[31:01] coming out. We'll see what gets
[31:03] released. We'll see if they can make it
[31:05] through without a problem, but these are
[31:07] some of the chain reaction risks as as
[31:10] uh DZ Bank in Germany talked about this
[31:13] week. The European Central Bank will
[31:14] begin checks on banks it supervises as
[31:16] concerns intensify over loan quality in
[31:18] the private credit sector.
[31:21] Remember the private credit sector has a
[31:23] lot of loans out to the K-shaped
[31:24] economy. It has a lot of loans out to
[31:27] SAS companies. It is ridiculous to me
[31:29] that people can say to my face with
[31:31] total conviction, this is blown out of
[31:33] proportion. A credit cycle is a credit
[31:35] cycle. They always go further than what
[31:38] their logic is because we live in a
[31:40] levered Ponzi scheme fractional reserve
[31:43] banking system where we are seven to 10
[31:45] times the size of the money that exists.
[31:48] If people are forced to sell assets in
[31:50] any asset class, but in particular in
[31:52] credit, the credit may be mispriced
[31:55] relative to the current fundamentals,
[31:56] but we just learned with the software
[31:58] side that that is happening more because
[32:01] AI is a headwind. AI is a disruptive
[32:04] force which will make terminal value out
[32:07] three years impossible to judge
[32:09] including getting paid from companies.
[32:13] KPMG faces allegations of blown audit in
[32:15] private credit collapse.
[32:18] Exposure to ailing software industry is
[32:20] bigger than advertised. Private credit
[32:22] wobbles could prove perilous for Trump.
[32:26] Great podcast. Um never met Kieran
[32:28] Goodwin. I've read some stuff in X. This
[32:30] was a great podcast. Um and it's great
[32:33] for two reasons. One is he gives a very
[32:35] clear picture of everything, but most
[32:37] importantly, he does not do it with any
[32:39] drama. He talks about it from a credit
[32:42] credit cycle perspective. He is part of
[32:44] the Boaz Weinstein team, meaning they
[32:46] are the ones that have made a bid down,
[32:48] whatever it is, 30 to 40% on on
[32:52] some of these situations. And this is
[32:54] important.
[32:56] He's talking his book, meaning what he
[32:58] wants to be able to do, but I think the
[33:00] thing that's important in a visual, and
[33:02] you guys I'll move my face out of the
[33:04] way so you guys can take a snapshot. You
[33:06] can go look at it. You can I would, you
[33:08] know, upload it into uh chat GPT or
[33:11] whatever your favorite LLM is, but the
[33:13] point is, when you guys look at this,
[33:15] he covers this very, very well. So, this
[33:18] is a schematic and infographic uh
[33:22] look at it. This is all happening, and
[33:25] there are so many issues, but they go
[33:27] from asset-liability mismatch to a
[33:28] liquidity crunch. We're seeing that. It
[33:31] turns to a credit crunch. It's going to
[33:33] have impacts on things that need money,
[33:35] which obviously gets somewhat into AI. I
[33:38] don't think there's a risk there, but I
[33:39] do think it pushes meta type deals with
[33:42] Blue Owl for $27 billion, the largest
[33:45] bond issuance, I think, ever. Uh or at
[33:48] least one of uh for that type of product
[33:51] done through an SPV. It kind of
[33:52] restricts that from going on. So,
[33:55] there's going to be financing, and meta
[33:57] was doing that to avoid it on their
[33:58] balance sheet.
[33:59] You put it on the balance sheet, bad for
[34:01] the multiples of the hyperscalers. So,
[34:02] just remember, credit cycle matters when
[34:07] people are borrowing money regardless of
[34:09] who is borrowing money.
[34:11] It's still showing up um
[34:13] it's very hard to see this, but this is
[34:15] the triple C versus the lowest part of
[34:17] high yield. And again, this is using the
[34:19] OAS, the options adjusted spread.
[34:22] We continue to widen out. Um it's
[34:24] happening.
[34:25] With the rally, the massive rally we
[34:27] saw, you didn't even get a bump in this
[34:29] stuff. That is usually the sign of a
[34:30] bigger problem.
[34:32] So, this is the BDCs. This is the
[34:34] private equity names relative to the
[34:37] S&P. So, the BDCs are down, but so are
[34:40] the private equity names relative to the
[34:42] S&P.
[34:43] I don't think that's going away. And
[34:45] here is Blue Owl making and I didn't
[34:47] include Friday. Friday it made new lows.
[34:49] It rallied back.
[34:51] And I don't know anyone was buying it,
[34:52] but it seemed to me a little bit like
[34:54] maybe there was deleveraging on Friday.
[34:57] That would be my gut telling me that
[34:59] that's someone was taking down stuff,
[35:01] but we'll see as time goes on. Here is
[35:03] private equity relative to software. You
[35:05] cannot separate the two guys. The white
[35:07] line is private equity, software led
[35:09] deflation.
[35:11] Software hasn't bounced. IGV. Remember
[35:13] everyone telling you step in? Remember
[35:15] all the things I highlighted 7-8 weeks
[35:17] ago of people trying to go in and go for
[35:19] dead assets?
[35:21] We could get a bounce to 90 in IGV. I
[35:24] doubt it, but we could. The pressure
[35:26] from AI is getting bigger by the day. I
[35:29] will go through this.
[35:31] And this is the main point. The Mythos
[35:33] signal. Again, for the subscribers, read
[35:35] this. If you don't know what recursive
[35:37] self-improvement is,
[35:38] computers are starting to teach
[35:40] themselves in Mythos. If you go through
[35:42] the details,
[35:43] is scary. Anthropic confirmed that it's
[35:46] testing a new model Claude Mythos
[35:47] describing it as a step change but
[35:50] beyond even Opus and calling it the most
[35:52] capable model. Opus is what caused the
[35:54] software collapse as it was. We've had
[35:56] so many releases since that time.
[36:00] Along with Open Claude, along with
[36:01] Cohort, along with computer. I can't
[36:03] even keep up with it.
[36:05] You have to start listening to some of
[36:06] the podcasts I'm showing. This one, AI
[36:09] Daily Brief, should be on your regular
[36:10] listening. Usually it's 15 to 30
[36:12] minutes. They went through the Mythos
[36:14] side.
[36:15] The Artificial Intelligence Show, the
[36:17] same thing.
[36:18] Lex Fridman with Peter Steinberger of
[36:21] Open Claude fame.
[36:24] Andrej Karpathy.
[36:26] The reason I brought that up. Oh, one
[36:28] more, Dara from Uber, the CEO.
[36:31] AI will replace 9.4 million jobs
[36:34] at Uber.
[36:37] The reason I took I showed you all of
[36:38] those I uploaded all of those
[36:40] transcripts because all of them are
[36:42] getting into the impact that we are at
[36:44] now. The agentic world is rising fast.
[36:48] The reason all of you need to spend time
[36:51] getting the information at this point
[36:53] from me is the agentic world is going
[36:55] fast and people are still talking about
[36:58] the GPU world. They're still talking
[37:00] about the brain getting up to 140 IQ.
[37:02] They're still talking about the data
[37:03] center build-out. We've moved past that
[37:06] at the end of last year and we've moved
[37:07] into the things that I've been
[37:08] highlighting. All of my model portfolio
[37:11] is geared in some way towards this next
[37:13] phase of the agentic world. And what I
[37:15] went through with those five, this
[37:17] continues to happen every week. All of
[37:18] those people mentioned there have been
[37:20] involved in describing what is happening
[37:22] with the agentic world. So, what I did
[37:23] is I just put them up went through it.
[37:26] Everything is there now for adoption to
[37:29] replace labor because once agents are
[37:32] working, once agents are learning on
[37:35] their own, once recursive
[37:36] self-improvement is here, and the model
[37:39] mythos is a model that isn't released
[37:41] yet because it is too scary.
[37:44] We have reached a level that you have to
[37:46] adapt to a world where we will be hiring
[37:49] millions of digital employees and even
[37:52] if the humans don't get fired initially,
[37:55] it's only a matter of time before the
[37:56] pressure grows, but we're at a point
[37:58] where an agent can learn the workflows
[38:00] on their own.
[38:01] That's where we are.
[38:04] I don't think people understand what
[38:05] this actually means. Every application
[38:07] on Earth can now build an agent that
[38:08] teaches itself how to use the
[38:10] application through the UI. You define
[38:11] what success looks like in eval. You
[38:14] point Claude at your application via
[38:15] computer usage. Claude tries to complete
[38:17] the task and it keeps going until it
[38:21] reaches it.
[38:23] Even if it takes hundreds of times.
[38:26] Again, think about when you get a piece
[38:28] of software. Think about your own
[38:29] experience with using some of the tools
[38:32] in Claude code.
[38:34] It takes a human a long time because you
[38:36] have to unlearn what you already know,
[38:38] which is I just want to press a button.
[38:40] How do I just press a button?
[38:42] Agents don't work that way. They learn
[38:44] everything quickly.
[38:45] This is why This is why I have my open
[38:47] claws to make sure that I'm ready for
[38:49] the agentic world. This is happening.
[38:51] Claude computer usage was out this week.
[38:53] Alex Finn, the velocity in which
[38:55] Anthropic ships is like any unlike
[38:57] anything I've ever seen. We have to
[38:59] assume they have access to an AGI like
[39:01] AGI like model that nobody else in the
[39:03] world has, correct?
[39:05] That is the whole point of Mythos. That
[39:07] is the whole point of what's going on is
[39:08] everyone assumes that this is going.
[39:11] This story, and this is
[39:13] an Anthropic person, is becoming more
[39:15] common face, which means
[39:17] you're getting more and more agentic
[39:19] side. People are hitting limits there. I
[39:21] hit my first limit usage this week. As
[39:24] someone who pays the max, this was only
[39:27] for people who weren't paying the max. I
[39:29] hit it this week. Not on Claude,
[39:32] on Perplexity computer.
[39:35] Still a great GPU shortage. So, rental
[39:37] capacity is going up, and that's because
[39:39] now you're getting millions of new
[39:40] employees. So, you have to think of it
[39:42] as
[39:43] we are now hiring hundreds of thousands
[39:46] of digital employees
[39:50] a month.
[39:51] So, if you went back to the payroll and
[39:52] they would show you how many agents are
[39:53] there and you got released, that is not
[39:55] a positive GDP story from a consumption
[39:58] perspective. It is from a velocity of
[40:01] money perspective.
[40:03] That will continue, and this is why
[40:04] nominal GDP is going to be fine, guys.
[40:06] We need more GPUs, so the rental
[40:08] capacity keeps going higher. We have a
[40:10] shortage,
[40:11] a complete shortage. Mustafa Suleyman
[40:14] from Microsoft recently moved down, even
[40:17] though they won't call it a demotion,
[40:18] but sure seems to be that way to me,
[40:20] formerly of Google.
[40:22] Um
[40:24] For the next couple years, the entire AI
[40:26] industry is going to be defined by this
[40:27] fact. Demand is going to wildly outstrip
[40:29] supply.
[40:32] Which is what matters. Companies'
[40:34] products have to pay margin for tokens.
[40:36] Again,
[40:37] compute power, silver, commodities,
[40:42] GPUs, CPUs,
[40:44] uh optical fiber, all of that stuff is a
[40:47] necessity to try and get enough compute.
[40:51] This guy, Zinsner, Zinsner, who I've
[40:54] referenced before, continues to talk
[40:56] about the shortages in
[40:59] server CPUs.
[41:01] Remember the whole rack, guys? One of my
[41:03] thematic pieces. Just continue to stay
[41:05] on it, particularly for those
[41:07] interested in the edge. Uh
[41:10] Anthropic, this is just again going
[41:11] through that they released their most
[41:12] powerful
[41:14] model.
[41:17] I wanted to give you guys an example of
[41:19] what is starting to happen
[41:21] because this was on that podcast. And
[41:23] one of the things they talked about is
[41:25] how the frontier labs are under extreme
[41:27] pressure from the ability of
[41:31] using some of these efficiency gains,
[41:33] but also using things like auto
[41:34] research, and basically take cheap
[41:36] models and make them more powerful. This
[41:38] is the way that Cursor is getting around
[41:40] the massive amounts of dollars. So,
[41:41] Cursor is a coding tool, and they
[41:43] basically, think of them as a wrapper
[41:45] that is using the models cuz they don't
[41:47] own the models.
[41:49] But if they keep raising their price,
[41:50] and other people have models, then their
[41:52] business is under pressure, which I've
[41:53] talked about. So, these guys are trying
[41:55] to find ways to get cheaper models, but
[41:57] get performance that is almost as good.
[42:00] So, the one thing in there is arguing
[42:02] the frontier labs won't be able to
[42:03] charge the top dollar forever in
[42:04] domain-specific workflows because
[42:06] specialized post-trained models
[42:08] can increasingly undercut them on price
[42:11] price and performance. This is a
[42:12] commoditized thing and this again is bad
[42:14] for the frontier model companies in
[42:16] terms of how are they going to monetize
[42:17] this? I do not know how any of them are
[42:19] going to monetize it. Um, the AI cost
[42:22] war is over. The best models are
[42:23] becoming free.
[42:25] Gemma 4 was released. This has huge
[42:27] implications. I will do a paper on this
[42:29] for subscribers, but Gemma 4 was came
[42:32] basically came to the market on Friday.
[42:35] Agentic skills to the edge with Gemma 4.
[42:38] Remember Gavin Baker holding up a phone
[42:40] saying the most dangerous thing for the
[42:42] capex buildout is when we start moving
[42:44] to the edge. It is getting easier every
[42:46] day. Turbo Quant last week and now Gemma
[42:49] 4. Gemma 4 basically an open source
[42:51] model that you can download and use
[42:53] locally on your Mac mini. Uh, don't need
[42:55] to use Chinese open source models.
[42:58] Also leads to ability to get it on your
[42:59] phone. Uh,
[43:01] the data centers. So, this would scare
[43:03] everyone. The data center's not get
[43:04] Maybe maybe they're not going to get
[43:06] built out. Well,
[43:07] that's not the case. Uh, even though the
[43:10] news is so sensationalized, almost half
[43:12] of the US data centers planned for this
[43:13] year are expected to be delayed or
[43:15] canceled.
[43:17] Okay. So, you read a story like that. I
[43:19] spend
[43:21] I
[43:21] If you guys want to call up 22V and and
[43:26] get the data on the data centers, you
[43:28] can make it you know, strike someone. I
[43:30] do this type of stuff to keep track of
[43:32] what's going on in the data centers.
[43:34] Um, I run it every single month through
[43:37] every LLM that I have to get all of the
[43:39] most recent data. Basically, the data
[43:42] center side is actually gone higher in
[43:44] terms of the expected numbers and this
[43:45] includes everything that they reference
[43:47] in there. So, my model puts out the
[43:51] terawatts at risk.
[43:53] Every single one of these, whether it's
[43:55] the grid, whether it's the processing,
[43:56] whether there's labor and construction,
[43:58] all of these things fit in with what's
[44:00] happening. We've got, you know, gas
[44:02] turbine issues that are going to be here
[44:04] for the next 3 years. All of this stuff
[44:06] is there. So, in the same week though,
[44:10] or a week ago, Meta ordered 10 gas-fired
[44:13] to expand its Hyperion one. So, my
[44:15] numbers went up because they're building
[44:16] bigger. Oracle was able to
[44:20] get 16 billion financing after Blue Owl
[44:23] bailed out of some stuff. They were
[44:24] still able to get it. The question is
[44:27] what was the cost?
[44:28] In terms of token factories, you got to
[44:31] spend some time with Eli Lilly. This is
[44:32] on my model portfolio. I've talked about
[44:34] Eli Lilly before. Um
[44:37] you have to go see what they're going
[44:39] through, but they um one of the reasons
[44:40] that I've been very bullish on it was
[44:42] the relationship with Insilico, which is
[44:43] also connected to Google DeepMind.
[44:46] And also Eli Lilly is connected to
[44:48] Nvidia. You can't have a better
[44:51] connection, DeepMind, Nvidia, Insilico.
[44:53] And if you go through
[44:54] the whole mapping of this, it is
[44:56] unbelievable that they now have their
[44:58] own token factory. If you want to
[44:59] understand what is going on, you need to
[45:02] go read more about the supercomputer
[45:06] could change the future of medicine. We
[45:07] are in that stage right now where the
[45:09] token factories are going, guys. It is
[45:11] happening.
[45:12] While everyone's seeing that, they are
[45:13] getting worried about memory chips, and
[45:15] now the headlines are coming out. I love
[45:17] that cuz I love all this stuff. Um this
[45:19] was a good article if you want to go go
[45:21] through as to you know, what caused
[45:24] Micron to go from 470 to 330, which
[45:28] again is part of a correction that
[45:30] should be expected, especially something
[45:31] that's up four times where it was.
[45:34] DDR prices face a complete collapse. I
[45:37] got this from a few people.
[45:40] Here's a reality. Go read something like
[45:42] this. Um pricing is soft, demand isn't.
[45:45] Here's what buyers should understand.
[45:46] Nice, calm, all Turbo Quant. Let's go
[45:49] through it. I completely agree with
[45:51] this, and as someone who still has a
[45:53] very large position in Micron, my belief
[45:56] has always been that sometime towards
[45:57] the middle of next year, and I say
[45:59] believe, I'm looking that the memory
[46:02] stuff will start to build in a peak in
[46:06] some sort of the demand side a year to a
[46:09] year and a half beforehand,
[46:11] and the news will still be good at that
[46:12] point. Well, I still believe that we're
[46:15] going to make it into 2027 with things
[46:17] still being incredibly tight. Uh that
[46:20] being said, it's no longer my biggest
[46:21] position after this week because of the
[46:23] name that I highlighted uh because I
[46:25] think the agentic side is now more
[46:27] important and even though memory is a
[46:28] part of it, memory has already been a
[46:29] big part and everyone is focused on
[46:31] bringing supply to memory and focused on
[46:33] efficiency. Here is why this move in
[46:37] memory should not be a surprise.
[46:39] Magically, here's Micron. Here is the
[46:41] price of DRAM. And again, it peaked.
[46:44] It's been declining. That's where it
[46:47] was. Guess what? The SMH, same thing.
[46:50] All you got to do, guys, is pay
[46:51] attention to what's happening cuz
[46:53] remember, when this started to go
[46:54] higher,
[46:55] well, guess what? Micron, that's when
[46:57] people finally started to bail in. It
[46:58] was still 100 You still could have
[47:00] bought this name at 175 when memory
[47:02] started to go higher. So, just follow
[47:04] the DRAM prices and you'll find out
[47:06] whether it should be going up as fast or
[47:08] not.
[47:09] This is again for Gemma 4.
[47:12] I just want to highlight this is where
[47:13] my brain is moving to. It's for the edge
[47:16] across the hardware stack, okay?
[47:21] Memory to non-memory semiconductor
[47:23] prices rising across the board. We are
[47:25] at the agentic phase. This is why
[47:27] all of my baskets
[47:29] have far more in them than memory. They
[47:33] are all about the packaging, the
[47:35] advanced packaging, the move to analog,
[47:37] the move to cars, the move to phones,
[47:39] the move to computers, the move to the
[47:40] agentic side.
[47:42] Go get the model portfolio names.
[47:44] They've all They're all on the website.
[47:46] The model portfolio itself is not there
[47:48] yet, guys, but all of the names and the
[47:50] themes are there. There's 100 There is
[47:54] I believe 95 in total, plus I added five
[47:56] in there. The five I added role names I
[47:58] referenced here on the videos. They are
[48:01] all part of it including Brazil, silver,
[48:04] some other things.
[48:05] Uh
[48:06] This is from SemiAnalysis, the best of
[48:08] the best. And again, I had to do work
[48:11] and someone I remember someone asked me
[48:12] and said, "What do you mean memory is
[48:14] now
[48:15] on a $50 thing? It's 15 to 20 billion of
[48:18] it. That's not right. It's
[48:20] People are not keeping up with how
[48:21] quickly things change." So,
[48:23] 7 to 8 billion on a $50 thing of memory
[48:27] on a on a data center. That's what it
[48:29] was. Now for the hyperscalers, they
[48:30] spend 30% of their spend
[48:33] going up to 36% is on memory. We have
[48:36] insatiable demand.
[48:39] This story is growing. OpenAI is falling
[48:41] out of favor with secondary buyers. They
[48:42] completed their $122 billion Say that
[48:45] number again. $122 billion raise.
[48:48] And that's before the IPO where they're
[48:50] going to raise a hell of a lot more
[48:52] money by the end of the year. Think
[48:53] about how much money they need to raise
[48:55] and just realize they need to start
[48:56] getting some revenues in the door
[48:58] quickly or their equity multiple is
[49:00] going to start to take it. I'm sure once
[49:02] they go public that will become clear to
[49:03] people. Uh this is the story I keep
[49:06] talking about. I think it's a major
[49:07] macro story. Artificial intelligence is
[49:09] being misapplied by executives who treat
[49:11] it as a technology rollout rather than a
[49:13] fundamental rethink of business. I could
[49:14] not agree more. And this is coming from
[49:16] someone who's spending time with these
[49:18] people.
[49:19] Absolutely the problem. This is why
[49:21] adoption is moving slow. The more people
[49:23] I talk to at institutions, they have no
[49:25] idea how to deal with this. I'm sure
[49:28] that 5% or 10% of the companies in the
[49:30] S&P 500 are doing it this way, but
[49:33] viewing it as a technology rollout means
[49:36] you will by definition miss. How many
[49:38] firms have ChatGPT Enterprise?
[49:41] And Claude has taken over. I will tell
[49:44] you that if I talk to 10 people, I would
[49:46] say at least five of them at this point
[49:48] if not six are stuck in ChatGPT
[49:51] Enterprise.
[49:53] Infra rushes to limit leak of code
[49:55] Claude code. I want to just a few more
[49:57] slides here, but I want this I'm going
[49:59] to keep talking about the move to the
[50:01] edge in the enterprise. The reason you
[50:02] want to focus on the move to the edge is
[50:04] not just the cost side, it's the
[50:06] security side. And we're getting more
[50:07] and more of these stories. Now, this was
[50:09] a a leak.
[50:11] But Claude code source code was leaked.
[50:14] You have to go read the stories. I
[50:16] choose the VentureBeat one. When you
[50:17] read through it, this is like if
[50:19] Kentucky Fried Chicken released the
[50:21] recipe by mistake.
[50:23] The rush to
[50:25] to basically do what they did because
[50:28] they released their source code. This
[50:29] was a major news story from a
[50:31] competitive basis. It will accelerate
[50:33] the ability, I'm sure, of other places
[50:35] to do similar things.
[50:37] This is just unbelievable at this stage.
[50:42] But a looming cyber nightmare. So, not
[50:44] only is Mythos being held back because
[50:47] they wanted to give it to the government
[50:48] and cyber companies because of their
[50:50] fear on cyber. This is becoming bigger.
[50:53] So, the AI government officials tell
[50:55] Axios that Anthropic, OpenAI, and other
[50:58] tech giants will soon release the new
[50:59] models that are scary good at hacking
[51:02] systems at scale.
[51:05] Systems mean
[51:07] your bank account. It means everything,
[51:09] guys. So, again, as someone who follows
[51:11] cryptography for Bitcoin, and every time
[51:13] a quantum thing comes out, I get just
[51:15] raced with people. It's a long way away
[51:17] from quantum to cryptography. It is days
[51:20] away from Claude code and the things
[51:23] that a kid can use on their computer,
[51:25] having the ability for $200 a month to
[51:27] break into almost anything using AI
[51:30] agents.
[51:32] Five actions enterprise security leaders
[51:34] should take now based on the source code
[51:37] leak.
[51:40] First, the Claude code leak, and now
[51:42] this. In the same week, AI safety by way
[51:44] of will lock it up is just totally dead
[51:46] kaput, pushing up daisies.
[51:49] I mean, I again,
[51:51] everything was released to China. I
[51:54] Guys, this is Marc Andreessen saying
[51:55] this.
[51:56] Mercur was hacked. I've mentioned Mercur
[51:58] many times on here. They are
[52:01] a software place doing specialized
[52:04] models. They are doing
[52:06] human feedback reinforcement learning
[52:08] through have hiring people. So, if you
[52:10] go on LinkedIn and say Mercur, you will
[52:12] find job offers for bankers, for
[52:15] lawyers. They are training models based
[52:16] on experts to get through the nuances of
[52:19] particular professions. They are clients
[52:21] of all the major LLMs, which are trying
[52:23] to sell these specialized models at a
[52:25] very high price to get revenues in the
[52:26] door.
[52:28] They were hacked.
[52:30] Hackings are going to accelerate. So, to
[52:33] finish up the last two slides,
[52:35] Morgan Stanley's head of head of digital
[52:36] asset strategy dropped a bomb. On the
[52:38] institutional security side, we are
[52:40] turning our dark pools to support
[52:42] tokenized equities by the end of 2026.
[52:45] Again, I know crypto is in the doldrums.
[52:48] I know it's hanging out. Read this in
[52:50] Substack.
[52:52] Read it for the 22V people. I sent it
[52:54] out this week. Bitcoin will be the best
[52:56] performing asset once we get through the
[52:59] point where the government has to get
[53:01] involved. The point may be when the Fed
[53:03] has to cut rates despite inflation being
[53:05] higher or at least say that they're
[53:06] keeping a dovish bias. Whatever it is,
[53:09] we have a new Fed chair coming in in
[53:10] May. We will eventually make a bottom in
[53:12] here. The credit cycle will force the
[53:13] government to do what they always have
[53:15] to do, which is protect the voters. The
[53:17] voters are going to be angry as we go
[53:18] into the midterms cuz they're trapped
[53:19] inside all these assets. You can see
[53:21] where all this is going, but in
[53:22] particular, if it gets to the insurance
[53:24] industries and it hurts the annuities.
[53:26] Thanks again for those of you who liked
[53:27] this. Please subscribe. It helps me. It
[53:30] helps my business. It helps me keep
[53:32] getting this content out. I do need to
[53:34] make a little bit of money off some of
[53:36] this, and I'm not charging you for it.
[53:38] So, if you can help me out, you're part
[53:39] of the community, great. I will keep
[53:41] trying to help you guys feel empowered
[53:42] by making money and keep you on top of
[53:44] things and give you some AI names. Go
[53:46] subscribe at the paywall, especially if
[53:48] you're a financial advisor or a RIA. I'm
[53:50] working with them trying to build things
[53:52] and I will have things coming out
[53:54] regularly.
[53:55] Uh and overall for your kids
[53:59] have them watch this. Uh I will do a
[54:01] YouTube soon where I'll release some
[54:02] kind of video on a basic training
[54:04] ground. Maybe I'll release one of the
[54:06] videos, the five videos I've put up on
[54:08] the paywall, just so people can see
[54:10] what's going on, but I'll probably make
[54:12] a new one. I'll put it up on YouTube.
[54:14] It'll be a basic thing related to how to
[54:16] use one of the agentic sites. Thanks,
[54:18] guys.
[54:19] Have a great long weekend and I will see
[54:21] you next week. And by the way, I did
[54:23] this uh before the payroll number came
[54:25] out on Friday, so whatever's happened, I
[54:27] don't know. See you.