Supersonic Tsunami Hits SaaS: My Turbulence Model Is Flashing Risk — Jordi Visser (15 febrero 2026)

[0:00] All right, let's jump right into it. Um,
[0:03] first of all, uh, it was great to see
[0:05] everyone the last three weeks. This
[0:06] week, uh, at the Bitcoin investor event,
[0:10] um,
[0:12] it's just a good time right now, guys,
[0:14] to uh, spend talking about the markets,
[0:16] going through it, particularly when it
[0:18] comes to AI, crypto together, uh,
[0:21] despite the fact that Bitcoin has
[0:23] fallen. I'll talk more about that
[0:24] towards the end. Uh, this week, the payw
[0:27] wall was launched. So for those of you
[0:29] who have not yet signed up, hopefully uh
[0:31] you will uh this week um the whole theme
[0:34] of of this and really going forward is
[0:37] that we have reached the point that Elon
[0:39] Musk called the supersonic tsunami. Uh I
[0:42] think it's really important particularly
[0:44] for uh global investors, mutual funds,
[0:46] anyone who is making choices between the
[0:49] US uh and the rest of the world. anyone
[0:52] who's trading who's been fixated on
[0:54] picking the bottom in SAS. Hopefully
[0:56] this week was the week that everyone
[0:58] learned their lesson. If not, I'm going
[1:01] to go through, but the main message I
[1:02] want to get through and I'm going to
[1:04] show it a bunch of times is you really
[1:06] have to think about supersonic tsunami
[1:08] because we are in the midst of it now.
[1:10] Uh this is the website. Again, I
[1:13] highlighted exactly where you can go see
[1:16] it. You can go to the 22vresearch.com.
[1:19] But again, I'll keep adding stuff there.
[1:21] A lot of the things you see in the
[1:22] weekly videos uh are in there in depth
[1:25] and plus all the research and stuff. I
[1:27] did a lot of things this week for it.
[1:29] But just so you guys hear this, when I
[1:30] started Viscer Labs and I started my
[1:32] consulting work, the the main point was
[1:34] I have a lot of domain experience and
[1:36] regardless of what people over the years
[1:39] have whether they've known me, not known
[1:41] me, uh I do approach markets from a deep
[1:44] macro background and I've been using AI,
[1:48] I don't even know how it's impossible to
[1:50] measure, but uh how many hours a day,
[1:52] but this has been now 18 months of
[1:54] constantly using it for everything in my
[1:56] life. Everything that you see in this
[1:57] video, AI is involved in. Everything I
[1:59] do all day long, AI is involved in. It
[2:02] has consumed me. I am a curious person
[2:04] that loves this. But also, I'm an
[2:06] investor who's making investments not
[2:08] only for me, my family, people that I
[2:10] know. Uh 22V has an asset management
[2:12] unit. I am looking into doing stuff on
[2:14] that. So, if you have uh interest in
[2:16] that in terms of getting the names and
[2:19] getting a deep dive. Uh there's all
[2:20] kinds of different things that I believe
[2:22] artificial intelligence has allowed me.
[2:24] But I also think we're at the most
[2:25] disruptive point in this journey. Uh
[2:28] that's why I did the piece at the
[2:29] beginning of the year and a and crypto
[2:31] is connected to it. So for those of you
[2:33] who have not done anything in crypto,
[2:35] you're getting at your last opportunity
[2:37] as far as I'm concerned. If you've
[2:38] watched me over the years, I fully
[2:40] expected that eventually what would have
[2:41] to happen is the mag 7, the hyperscalers
[2:43] and software would have to start
[2:45] underperformance. That is the form of
[2:47] abundance and because Bitcoin is built
[2:50] on code and all of the ecosystem is it
[2:52] would have to be disrupted as well to
[2:54] some degree. But as people realize that
[2:56] the capital structure of the future is
[2:58] changing and AI agents will be the
[3:00] consumers not humans. You have to think
[3:02] more about that what that is the utility
[3:04] function of crypto is going to grow
[3:06] rapidly. Before you know it before the
[3:08] end of this year SAS will still be down.
[3:11] Bitcoin will be up towards the high and
[3:12] all of you that have been looking for
[3:14] deep dives and you're it's a value trap
[3:17] and I'm going to go through why. Uh and
[3:19] the final message on this is use AI. So
[3:22] the videos that I'm doing, I will have a
[3:23] five-part video series.
[3:26] For those of you who are stubborn and
[3:27] don't want to get involved in it, that's
[3:29] great. But for your kids, at least get
[3:30] them the videos and let them see. I will
[3:32] make it simplistic and I will make it
[3:34] how they can replicate everything that
[3:35] I'm doing. This is the visual of the
[3:38] supersonic tsunami. There's two
[3:39] components of supersonic. One, speed.
[3:44] Two, structural power. I was on stage at
[3:46] the Bitcoin investor event following
[3:48] Kathy Wood this week and she made a
[3:51] statement that what's happening in SAS
[3:52] is just like co. I've seen this from Dan
[3:54] Ies. I've seen this from Morgan Stanley.
[3:56] I've seen this from JP Morgan. I've seen
[3:57] it all week with people trying to pick
[3:59] the bottom in SAS. And I will continue
[4:01] to say it is a mistake to do this. AI is
[4:05] now in the acceleration phase. It is
[4:07] going way too fast. The dangerous part
[4:09] of it is coming up. But everything gets
[4:12] consumed by artificial intelligence
[4:14] going forward. And if you think that
[4:16] you're not going to be exposed to it,
[4:17] that is whether you're a $5 trillion
[4:19] company or not, you will be exposed by
[4:21] it at some point over the next three
[4:23] years. So be very aware of your
[4:24] investments on this. This is exactly
[4:27] what Elon talked about. You can go back
[4:28] and read the moonshots. Uh go watch the
[4:31] video. Artificial intelligence and
[4:32] robotics are a supersonic tsunami. So
[4:35] I'm not even including the robotics.
[4:36] That will be the thing that disrupts all
[4:38] the physical side, but we're not going
[4:40] to get into that for another four to
[4:41] five years. Right now, where we are in
[4:43] the abundance disruptive size and you
[4:45] want to rotate your money into the
[4:46] physical side to avoid this. That will
[4:48] continue to be the play. $85 trillion of
[4:51] investment will happen. There's no other
[4:53] way around it. We'll be shoving
[4:54] artificial intelligence into everything.
[4:56] The supersonic part eliminates value. It
[4:59] makes everything a value trap where it
[5:01] just destroys things in its path because
[5:02] it's moving so fast. We've never seen an
[5:05] innovation that changes this fast. And
[5:06] I'm going to highlight today what has
[5:08] happened just in the last 10 days that
[5:10] I'm sure most of you have missed. And
[5:12] we're at the singularity point which is
[5:15] really where recursive self-improvement
[5:17] where the machines are teaching the
[5:18] machines. So this week it spread. It was
[5:20] not just in SAS. We had broker stocks on
[5:23] Monday's uh insurance broker stocks that
[5:25] were pummeled on Monday. Then we had the
[5:28] wealth management platforms and as I
[5:31] said on CNBC this week, Altruist, which
[5:33] is a $ 1.5 billion dollar valuation
[5:37] company, I'm sure almost no revenues. At
[5:40] the same time, they took down LPL,
[5:43] Raymond James, Schwab, quarter of a
[5:46] trillion dollars of market cap.
[5:48] Entrepreneur versus the big guys. This
[5:51] is what I've been preaching now. And
[5:53] this is the reason why I'm heavily
[5:54] involved in Bitcoin because eventually
[5:57] the eight billion people on the planet
[5:59] that are at the bottom end of the K have
[6:02] a tool in their hands that allows them
[6:04] to build a business with no people, no
[6:06] cost, and they can get up very rapidly
[6:09] and actually disrupt businesses. Now,
[6:11] are we going to see that in the revenues
[6:12] for those companies anytime soon? No.
[6:14] Will it impact the growth? Absolutely.
[6:16] Unless they can find a way to adopt it
[6:17] quickly and offset some of that. You got
[6:20] into commercial real estate service
[6:22] names this week. Then you got into
[6:26] trucking stocks. So we went everywhere
[6:28] this week. We really disrupted the whole
[6:30] side. And this is where it's starting to
[6:32] become an issue. And as I go through
[6:33] this, um, I have my turbulence model.
[6:36] The market is warning everyone that this
[6:38] rotation is getting very, very bad. The
[6:40] last time we saw a rotation this big and
[6:42] this important was back here. So this is
[6:45] from Michael Bat Batnik on the compound.
[6:48] Um it's a great chart and I think it
[6:51] highlights it uh in terms of just going
[6:53] through when 115 or more stocks fell at
[6:55] least 7% in a single day across a
[6:57] rolling 8-day window within the S&P 500.
[7:00] So the last time we did this when and
[7:03] these dots are where the S&P was
[7:05] relative to all-time highs. This is back
[7:08] in 2000. I do believe that we are in the
[7:11] midst of a rotation that is equivalent
[7:12] to the dot bubble. I think this is a
[7:14] very similar situation. So instead of
[7:17] calling this a bubble, earnings are
[7:18] growing. I don't see this as being a
[7:20] systemic issue. I don't see this as
[7:22] being anything that's going to impact
[7:23] the economy in any meaningful way. But
[7:26] for those of you who went through the
[7:27] dotcom bubble, 2000 was a very unique
[7:29] period because we did see the Dow make
[7:31] new all-time highs while tech was
[7:33] getting destroyed. And it wasn't until
[7:34] after 911 that we actually saw the
[7:36] economic fallout that led to a
[7:38] recession, which wasn't even technically
[7:39] called a recession. Just more highlights
[7:42] on how much dispersion there was. people
[7:44] talked about it, but this is a fragile
[7:46] market and this is why I have a
[7:48] turbulence model and this is the way I'm
[7:49] towards the end I'll start putting it
[7:50] into context as to what to watch out
[7:52] for. Um, a phrase that I've used always
[7:55] I've done three risk calls this week not
[7:57] only with um clients but also with major
[8:01] uh regulatory bodies because I have the
[8:05] ability of talking to them and at least
[8:06] highlighting what I'm worried about.
[8:08] What I'm worried about is the amount of
[8:10] leverage that has happened inside the
[8:12] hedge fund world since the great
[8:14] financial crisis. In my emerging market
[8:16] days and all the risk committees that I
[8:17] was on at Morgan Stanley, I always had a
[8:19] phrase, the seeds of the next crisis
[8:22] were always planted in planted in the
[8:24] prior crisis. The prior crisis was the
[8:25] great financial crisis. That was when
[8:27] the uh the iPhone came out. That was
[8:29] also when quant strategy started to grow
[8:31] rapidly. uh multistrats started having
[8:34] quant strategies and they became experts
[8:36] at reducing risk and being able to lever
[8:39] up more and control the coariance
[8:41] matrices in their businesses. This is a
[8:44] warning shot and in my opinion this will
[8:46] lead to problems as the year goes on. So
[8:48] for anyone watching on the risk side you
[8:50] want to have a conversation feel free
[8:52] otherwise uh just deal with these
[8:54] because I think these are going to be
[8:55] more rapid and more normal going
[8:57] forward. uh great person to follow and
[9:00] just kind of look. Charlie Mcelikott
[9:03] does a great job uh at no more on this
[9:05] and he just highlights some more of the
[9:07] dispersion. I'm not going to read you
[9:08] all the thing but it's happening at a
[9:10] time when the S&P is not moving which is
[9:12] what my turbulence model was built on.
[9:14] Uh that is usually when the risk is
[9:16] growing the most. S&P is flat over the
[9:18] past month but the average stock moved
[9:19] 10.8%.
[9:21] a dispersion spread at the 99th
[9:23] percentile
[9:25] um leading to pod shop de-risisking at a
[9:28] 100th percentile gross leverage. I'm
[9:31] going to go through that. A lot of
[9:32] people were sending things. I put
[9:33] something out on Friday and people said
[9:35] there's no gross leverage on the hedge
[9:36] fund side. You guys got some work to do.
[9:39] Um because if you don't think there's
[9:40] gross leverage out there after what's
[9:42] been going on, I'm sorry. You're missing
[9:44] out on this. There is absolutely gross
[9:46] leverage. And if you want to have a
[9:47] class on leverage, give me a call
[9:49] because for whatever reason, if people
[9:51] are going to argue on something I wrote
[9:52] out about gross leverage, you're in the
[9:54] crosshairs. Uh you need to understand
[9:56] how much leverage is in the system. And
[9:59] the gross leverage is a representation
[10:00] of the lower V and the fact that again
[10:03] we've been dominated by one group which
[10:06] has no earnings cyclicality to it which
[10:09] over time has grown and allowed people
[10:11] to find hedges. You've had been able to
[10:13] use the Russell 2000 as both a beta
[10:14] hedge but also as a consistent hedge
[10:16] against software. That trade why I show
[10:19] IWM vers QQQ is critical to the risk
[10:22] structure of the framework of the
[10:24] market. That is my belief and I will
[10:25] keep going with it. Uh Dan Ies I love
[10:28] listening to him on his enthusiasm with
[10:30] AI but I disagree wholeheartedly in the
[10:33] fact that all of these companies in the
[10:34] selloff are generational opportunity. I
[10:37] disagree completely. There are some that
[10:39] you can go buy. I wrote a piece on
[10:40] Palunteer. I think that'll get a bounce.
[10:42] But for a lot of these companies, no
[10:44] offense to Dan, I I I can't imagine he
[10:47] uses AI all day. He's on the road all
[10:49] day. I don't think he sees what's
[10:51] happening at the at the pace that it's
[10:53] happening, and he doesn't use it. He
[10:55] can't he's on the road too much and
[10:56] selling clothes, too. So, no offense to
[10:59] Dan, but I think we've reached a point
[11:00] where the disruption and the tsunami has
[11:01] gone too fast, and you're going to start
[11:03] cannibalizing its own winners. Um JP
[11:07] Morgan tries to buy the bottom. I'm only
[11:09] doing this from the basis of I'm sure
[11:11] we're going to get a bounce in these
[11:13] names at some point, but there's two
[11:15] ways that waterfall charts like that
[11:17] historically end for me, especially when
[11:20] they are the most owned thing on the
[11:22] planet. It either owns in a value trap
[11:24] where they go sideways and the market
[11:27] deals with it in a rotational basis and
[11:30] then the other way is it continues in a
[11:32] waterfall and you hope there's no
[11:34] contagion. That's what happened in the
[11:35] dotcom bubble. Eventually, it did turn
[11:37] into a contagion. and you started having
[11:39] debt issues eventually because the
[11:40] equity uh names went down. We are
[11:43] getting closer to that every day. Even
[11:45] though there's not a lot of debt in
[11:46] those companies, they are overowned to
[11:48] such a huge degree. And growth verse
[11:50] value is such an overowned asset that if
[11:53] everyone tries to get through the door
[11:54] at the same time, it will have
[11:55] disruptive impacts. The leverage for me
[11:57] is not on the debt. It is at the hedge
[11:59] fund level and in particular the
[12:01] multistrat and quant level where they
[12:02] all are sitting on highly lever
[12:05] dependent on the coariance matrix. And I
[12:06] think if there's a rotation of this
[12:08] basis into small things out of tech into
[12:11] basic materials and energy where tech is
[12:13] 10 times the size and small cap it is
[12:17] more than 10 times the size since the
[12:19] Russell 2000 is smaller than the
[12:21] materials and energy combined are about
[12:23] the same. You just don't have the
[12:25] ability of dealing with this. So I think
[12:26] this is going to be an issue. I can
[12:28] emphas I can't emphasize it more and
[12:29] more. Um our job in managing money is to
[12:32] try to find things that have a higher
[12:34] probability of occurring than what the
[12:35] market is discounting. I think the
[12:37] probability of a unwind uh event that
[12:40] turns into something big is even if it's
[12:42] only 25% the market right now in terms
[12:44] of the positioning and the pricing is
[12:46] much closer to 2 to 5% than it is 25. So
[12:50] uh there are ways to hedge this at this
[12:52] point and I'm going to show some things
[12:53] as we go on. Goldman came out David the
[12:56] software itself has been too broad.
[12:57] Everyone wants to pick a bottom in it.
[12:59] And again, it's either a buying
[13:00] opportunity or value trap. I don't think
[13:02] it's a buying opportunity at all. I
[13:04] think it's a value trap. I think you can
[13:05] find some names to go in there and buy
[13:07] them for sure. But I think the
[13:09] disruption that is going to happen is
[13:11] much greater than this. Um, I talked
[13:14] about this before and I'm going to keep
[13:15] saying it. I wrote this on January 20th.
[13:17] Seems like a long time ago. It's three
[13:19] weeks ago.
[13:21] AI bubble has stopped as kind of a story
[13:23] which is ironic because I'm getting more
[13:25] and more negative on the hyperscalers by
[13:26] the day because of the money that
[13:27] they're spending and I'll show that
[13:29] later. Uh but buying cheap software
[13:31] became the new bubble trade. Stepping in
[13:33] to buy software because it looks
[13:34] overdone ignores what is actually
[13:35] happening in AI. Will there be relative
[13:37] bounces? Of course, but this isn't a
[13:39] typical sector rotation or buying
[13:40] opportunity and beaten down software
[13:42] names. We are watching the opening act
[13:44] of a demand destruction cycle driven by
[13:46] the steady deflationary pressure of
[13:47] exponential AI progress where coding is
[13:49] increasingly ubiquitous and effectively
[13:51] free. They're confusing the suppressant
[13:53] for the bubble and the bloat for the
[13:55] value. It's going to be a big story. And
[13:58] look how cheap they are now. They're
[13:59] only the fifth most or yeah fifth most
[14:02] expensive uh sector
[14:05] uh from where they were the most
[14:06] expensive sector. The problem is again
[14:09] three years from now I know that
[14:11] semiconductors are going to be needed. I
[14:13] have no idea which software companies
[14:15] are going to be needed. I know capital
[14:16] goods are. Look how cheap energy is.
[14:18] Look how cheap all of these are. I will
[14:20] talk about these later because these are
[14:22] a warning sign also on what's going on
[14:23] in the market, but I'll cover that
[14:25] later. Again, emphasis how fast things
[14:28] are moving and how structural the force
[14:31] is. So, we are going faster on something
[14:33] that people have faded. They watched it
[14:35] from a distance and now it's right on
[14:37] top of everything and you're starting to
[14:38] see the disruption. So on January 7th, I
[14:41] wrote this one on Opus 4.5. I had
[14:44] written another one on December 8th.
[14:46] This one was just how much had happened
[14:49] over the course of the month and a half
[14:52] or not even the five weeks from the time
[14:54] it was I built the turbulence model
[14:56] sitting on my couch on Christmas. I
[14:58] started using it more and more. So, Opus
[15:00] 4.5 was a major event and I want to
[15:03] remind you when chat GPT finally
[15:06] released 5.1 chat GPT 5 over the summer
[15:10] and it got panned that was a long time
[15:13] between 4.5 and 5. I just want to remind
[15:16] you that as we go through this because
[15:18] Opus 4.5 was a major event and then we
[15:21] got Opus 4.6.
[15:24] So, I would go watch and listen to the
[15:27] recent moonshots. I would also go listen
[15:29] to Nate Jones talk about this. The
[15:32] biggest AI jump I've covered. I see
[15:34] people fading Nate Jones. I I again
[15:36] people are just becoming polarizing on
[15:38] fading AI and not listening to people
[15:40] and thinking they know stuff about it
[15:41] when the people that are always fading
[15:43] these things. Do not use it. Opus 4.6
[15:47] shipped with a fivetime expansion in the
[15:50] context window versus Opus 4.5. This is
[15:53] uh notes from a combination of those two
[15:55] videos. That is a there a four-time
[15:58] improvement in coder document retrieval
[16:00] over just a couple of months. The model
[16:02] can hold 50,000 lines of code and know
[16:05] what's on every line at the same time.
[16:07] It now mimics a senior engineer in
[16:10] intuition. This is the difference
[16:12] between a model that sees one file at a
[16:14] time and a model that sees the entire
[16:15] system in its head simultaneously. The
[16:17] only reason I want you guys to spend
[16:19] time reading this,
[16:21] this again is only five weeks. At the
[16:24] same time, 30 minutes after 4.6 was
[16:26] released by Opus, OpenAI released 5.3
[16:31] Codeex, its newest GPT model helped to
[16:34] build itself,
[16:36] also known as recursive
[16:37] self-improvement.
[16:40] Give me the episode and Den titles from
[16:42] a moonshot episode we went through with
[16:44] Eric Schmidt and what happens when we
[16:45] hit recursive self-improvement. So if
[16:47] you guys are not using chat GPT regular
[16:48] and if you're not downloading stuff and
[16:50] going through this and again I use chat
[16:52] GPT for about 30 some odd percent. The
[16:53] best thing it has is it remembers
[16:55] everything that I do. So I can just go
[16:57] in and say hey I downloaded I've
[16:58] downloaded thousands of transcripts but
[17:00] now I can go in and say hey this one
[17:02] transcript it happened. Give me the
[17:03] details. Oh yeah this is when this is
[17:06] from July 17th. The reason I wanted is
[17:08] he brought up and said RSI is imminent.
[17:12] He emphasized that AI's biggest
[17:14] multiplier is its ability to learn and
[17:16] improve. not just scale computer data
[17:17] sets. Once AI systems begin meaningfully
[17:20] contributing to their own improvement
[17:21] process, a hallmark of RSI, progress
[17:25] starts to grow super exponentially.
[17:26] That's what he said back in July. Open
[17:29] AAI just told you that we are there now.
[17:31] Claude has shown that in terms of what's
[17:34] gone on. So this is now the inflection
[17:36] point. This is no longer AI assisting
[17:38] humans. This is AI generating
[17:39] infrastructure that AI will use. Again,
[17:42] five weeks ago is when software started
[17:45] to get hit because of opus 4.5 because
[17:48] they released you've got clawed code,
[17:51] you've got co-work, you've got clawed
[17:53] bot, you've got claude excel, you've got
[17:54] clawed powerpoint. Every single week
[17:56] there's a new thing going on. You have
[17:58] openclaw in terms of people using it at
[18:00] this point. OpenAI stated GPT35 codeex
[18:03] was instrumental in its own development.
[18:05] This is no longer theory. So if you were
[18:07] waiting for when it was going to happen
[18:08] and for those of you who reached out
[18:10] over the year and said Andre Carpathy
[18:11] he's on this now too we have reached a
[18:14] completely different point what Eric
[18:16] Schmidt broke down acceleration beyond
[18:18] human time scales supersonic new
[18:21] economic structures
[18:23] all of this gets disrupted that's the
[18:25] tsunami part energy and physical
[18:27] constraints matter that's where you go
[18:29] invest right now that's the scarcity
[18:31] because the faster this things goes they
[18:34] cannot back off they have to actually
[18:35] accelerate they're spending, which is
[18:37] what we're seeing. Because if they
[18:38] don't, as I'll get into later, they have
[18:41] existential risk. So, they have to keep
[18:43] spending. They're in a trap. Now, all of
[18:45] the frontier models, they're in a race,
[18:47] but they're also in a trap. Risk and
[18:49] reward duality. This is the problem.
[18:52] You're starting to now go from the
[18:54] benefits that come, but also at the same
[18:56] time, cyber security. There will be
[18:57] security issues this year. Be very
[19:00] careful with not owning V during the
[19:02] course of this year. There will be times
[19:03] that the market is going to dis be
[19:05] disrupted and the most likely places are
[19:07] going to be places that are vulnerable
[19:09] to all this coding. The software names
[19:11] that are cyber, those are other places
[19:13] along with Palunteer that I think people
[19:15] need to be um on this. Um Alex from
[19:19] Moonshots just talked about this. This
[19:21] is not productivity gain. This is
[19:22] intelligence cost collapse. 10 engineer
[19:26] uh years equals $2 million. AI cost
[19:29] $20,000. The deflation that is coming or
[19:32] that is here already is just insane. The
[19:35] models are getting better and the cost
[19:37] is staying the same. The models are
[19:38] getting four, five, 10 times better. And
[19:40] as I go through this, this is not just
[19:42] in the US. The scary part about this and
[19:44] what has happened that Dan Ies can't
[19:46] possibly be thinking about is absolutely
[19:49] how fast the Chinese models are getting
[19:50] there as well. And what we saw with them
[19:52] as well, AGI moment, people are getting
[19:55] scared of it because now they're
[19:57] realizing the negative side is going to
[19:58] pop out as well. They're building things
[20:00] humans historically built over years and
[20:03] it's now taking a matter of weeks if not
[20:05] days. It's no longer a tool. It's an
[20:08] independent productive system. This is
[20:10] why the warnings have come out from
[20:12] Demisabas, Sam Alman, uh Dario Modi, as
[20:16] I'll show you later, he had a safety
[20:17] engineer quit. Everybody is starting to
[20:20] get worried because they have the models
[20:22] six months ahead. We only have the
[20:23] models they've released now.
[20:26] Chad GPT's market share has fallen from
[20:28] 70 to 45 while Google, Grock, and
[20:32] Anthropic obviously go. They're
[20:33] leaprogging each other in real time. It
[20:35] used to take a long time for the models
[20:37] to come out. Now they're coming out
[20:38] almost there. We are in a model velocity
[20:40] war, not a product cycle, which means we
[20:43] are in the biggest part of the
[20:45] deflationary spiral. Now, Miniax, this
[20:48] is on Thursday, new open M2.5 and
[20:52] lightning near state-of-the-art while
[20:54] costing 120th of Claude Opus 4.6. It's
[20:57] being benchmarked right around the new
[20:59] one, not the old one, not 4.5.
[21:02] Little meme on this, they just dropped
[21:04] two. It's on par with Opus 4.6 while
[21:06] being 20 times cheaper. As I mentioned
[21:09] last week, I have a Mac Mini and now I'm
[21:11] using open source. It's a Chinese model
[21:13] because I don't want to have to pay lots
[21:14] of money for the tokens. This is what
[21:16] entrepreneurs around the globe are going
[21:17] to do because they're in the K-shaped
[21:19] economy. They want to build something
[21:20] and they care about the Spain.
[21:22] Enterprises won't. Now, you could see
[21:23] where big businesses are having to pay
[21:25] up for something while there's other
[21:26] models that are free. And the reason
[21:27] they don't they can't pay for them is
[21:29] because they have too much stuff in
[21:32] their business and too much to lose.
[21:33] This is the story of Bitcoin as well. We
[21:35] will get into that as time goes on. Bite
[21:38] Dance new video this week.
[21:41] Deep Seat coming out with a new model
[21:42] this week.
[21:44] This is the uh Alex Finn who this one
[21:48] had I forget how many millions I think
[21:50] 15 million. It went viral. I will be
[21:52] running Opus level super intelligence on
[21:54] my desk for free. This is quite
[21:55] literally changing everything. It's
[21:58] faster than sonnet. It's better than 4.6
[22:00] for coding. This is all related to that
[22:02] mini 2.5 that I showed. Open claw ran
[22:06] wildly for two weeks waking up hardware
[22:07] and agent manufacturers. It's turning
[22:10] Mac minis into the physical forms of
[22:12] digital labors. Cloud service providers
[22:15] such as Alibaba, Tencent, and BU. You
[22:18] don't see any American companies here
[22:21] have quickly launched one-click
[22:22] deployment of an open of a open-source
[22:26] AI agent platform. There's no US
[22:28] companies mentioned on this. The 16GB
[22:31] version of the Mac Mini has sold out.
[22:33] Good thing I got mine before and good
[22:35] thing I warned people to go get theirs
[22:36] before. AI agents are here to stay.
[22:39] Businesses say they are now widely
[22:41] deployed in large enterprises.
[22:43] Significant shift. So the profit margins
[22:45] are going to grow with these companies
[22:46] because they are already deploying
[22:47] agents. So again, for everyone who told
[22:49] you there's no adoption, the MIT piece
[22:51] last year, all of this BS that people
[22:54] has thrown out there, it's now going so
[22:56] fast that these people that said uh
[22:59] nobody's using it. Well, now you're
[23:00] going to sit there and go through the
[23:02] damage that has to be done. Whatever
[23:03] your long in this, whatever you're
[23:05] trying to buy, this disruption will be
[23:07] going at a faster pace. There is no way
[23:09] to get around this anymore, it's
[23:11] actually getting out of control. I
[23:13] wanted to bring this up because I wrote
[23:14] this before I started at 22V. This was
[23:16] after the Deep Seek moment. I had so
[23:17] many people calling up and I kind of
[23:19] laughed at it and said, um, it's not a
[23:21] market event in terms of ending things,
[23:24] but it does highlight something. The
[23:26] jaws of disruption, Deepseek and the AI
[23:28] shockwave. The whole thing about this
[23:30] was there are no moes. AI breakthroughs
[23:33] like DeepSeek will continue to emerge,
[23:34] shocking the market and rewriting
[23:36] industry narratives overnight. Just
[23:37] think about what we saw this week. Four
[23:40] separate industries destroyed
[23:42] for the for the week. Names falling 10
[23:46] 15 CH Robinson falling 16 20% in a day.
[23:50] Also, just like Jaws, there will be more
[23:52] sequels than people want to see. This is
[23:53] never ending now. So don't think that
[23:55] this is over. There will be one every
[23:58] week that happens. The only question is
[24:00] will people get used to it and not react
[24:01] to it the same way. Those who recognize
[24:03] the pattern, anticipate the inevitable
[24:05] disruptions and maintain a steady hand
[24:07] in the chaos will be positioned to
[24:08] capitalize on the seismic shifts of AI
[24:11] revolution. And my favorite catchphrase,
[24:13] you're going to need a bigger moat.
[24:16] Guys, coariance matrix screwed. Anytime
[24:21] bombs can happen every day in businesses
[24:23] that have moes that are winning, not
[24:25] companies that are losing. Go through
[24:27] some of these names this week and see
[24:29] where they were trading near their 52-
[24:30] week highs. It is normal for things to
[24:33] gap down when they're at 52- week lows.
[24:35] When they're gapping down at 52- week
[24:37] highs, it is a major problem. And that
[24:40] is the one thing when Michael Batnik did
[24:41] this, how many how many names inside the
[24:43] S&P fell at least 7% over the last eight
[24:46] days? Okay, it was 115. Great. Uh, and
[24:48] we're near all-time highs. Well, if you
[24:50] take that a step further, how many of
[24:51] these names that fell 7% fell close to
[24:54] all-time highs? That's when you start
[24:56] getting into problems because that is
[24:57] the way hedge fund investors and quants
[24:59] invest momentum based what's working
[25:01] starts. If we start getting in the point
[25:04] where momentum where you can shoot
[25:05] something overnight, that is why you end
[25:07] up getting these unwinds so quickly. And
[25:10] if everyone tries to unwind at the same
[25:11] time and they're in the same names
[25:14] because of the liquidity that does not
[25:16] allow them to be in smaller names, this
[25:18] is an idiosyncratic problem. So
[25:20] everyone's been trying to manage a book
[25:22] where they've got idiosyncratic risk as
[25:24] the bigger part of their book. Well, if
[25:26] that idiosyncratic risk is no longer
[25:28] idiosyncratic, if it's a theme which is
[25:30] I'm disrupted by AI and I'm crowded,
[25:33] that's a new factor that doesn't exist.
[25:35] Uh just highlighting again that we're in
[25:38] the final stages of capitalism where the
[25:41] labor profit pools just continue to grow
[25:43] or lab the profit pools grow while the
[25:46] labor compensation goes down. I heard a
[25:48] lot of economists seek happy with the
[25:50] jobs number and as I go through there is
[25:52] no good jobs number and now the bond
[25:54] market is starting to show the risk
[25:56] associated with um for everyone who has
[25:58] been calling for higher rates higher
[26:00] inflation in the midst of the tsunami.
[26:04] Good luck. This is what is happening now
[26:07] is we have tenure rates that have moved
[26:09] lower. And if I ask everyone why when we
[26:11] just had a payroll number that beat I
[26:14] think people should start getting
[26:15] worried. Uh the reason they should get
[26:17] worried is this falls in also with
[26:19] credit widening and it falls in with all
[26:20] the turbulence we're seeing. I worry
[26:22] again about the coariance matrix which
[26:24] is not just about the stock market. It's
[26:26] cross asset. Again for people on the
[26:27] risk side who want to talk about this in
[26:29] more detail. The only reason I feel I'm
[26:32] the epicenter of this from talking is
[26:34] because I very seldom meet anyone on
[26:35] risk or in macro who fully grasp the
[26:39] supersonic tsunami theme and why this is
[26:41] so important to markets because of
[26:43] mispositioning. There has never been a
[26:46] bigger mispositioning for something
[26:48] coming on as this tsunami that I've ever
[26:51] seen because people believe the MAG 7
[26:54] are a safety valve. They are all
[26:56] overweighted. When I say they, I mean
[26:58] all of you are overweighted. if you're
[26:59] in the S&P 500, if you're in MSCI world.
[27:02] So, if you're ignoring this, why are
[27:04] 10ear rates going down at this point?
[27:07] Give me the reasons why this should be
[27:09] happening when we have a time where
[27:11] nominal GDP is growing at 7%. What is
[27:14] the rationale when we just had a payroll
[27:15] number? Think about it and then just
[27:17] realize there's something bigger going
[27:19] on here. What is happening? I'm going to
[27:21] go through it. So on the payroll numbers
[27:23] for everyone who said everything was
[27:24] fine. I'm going to continue to repeat
[27:26] the disruption from AI is happening at
[27:29] the level where humanoids cannot
[27:30] replace. So let's just start with there
[27:33] were 137,000 jobs in education and
[27:36] health services 130 overall which means
[27:38] there was 7,000 jobs. Now again we've
[27:41] got the government jobs which continue
[27:43] to be negative and going through but the
[27:45] reality is this is not a strong payroll
[27:47] number and absolutely AI is still having
[27:50] an impact on these things. the wage
[27:51] pressure. So, I highlighted last week
[27:55] the Atlanta Fed and I said employment
[27:57] costs will continue to move lower. The
[27:59] employment cost uh index came out at
[28:02] even lower annualized than this. We are
[28:04] tracking lower on this. Right now, we're
[28:06] 3.4 on this. I continue to believe that
[28:09] the wage pressure will remain on the
[28:10] weaker end because of everything
[28:12] happening on this true inflation which
[28:14] gets ignored by people. It is absolutely
[28:17] correlated over time to core inflation.
[28:20] So true fl true core inflation came out
[28:23] at 1.17 while we got the 2.6 here. I
[28:27] continue to believe the deflationary
[28:28] pressures from AI will show up in
[28:30] services especially with the wage
[28:32] pressure and all this stuff. The S&P not
[28:35] falling. It's just sitting around here.
[28:36] It is below the 50-day and it is kind of
[28:39] a chart that I'd be a little bit worried
[28:40] about in terms of going through it.
[28:42] We'll see what ends up happening. I
[28:43] don't want to be bearish because I do
[28:45] believe earnings are going to be great.
[28:46] The economy is going to be great. This
[28:48] is really a disruptive thing that I
[28:50] don't think a waterfall impact from
[28:51] software which is the biggest component
[28:53] and is a reason why the S&P has not
[28:55] broken out. You want to be long the
[28:57] stuff you can be long. So it would not
[28:59] be surprising for me to see us go down
[29:01] here and then to have a dead cat bounce
[29:04] again back up to the highs. And I say
[29:06] dead cat because I don't think software
[29:08] and the mag 7 are going to be able to be
[29:09] the things driving it. And if it's
[29:11] energy and materials they're too small
[29:14] relative to tech. So this is an
[29:15] abundance scarcity thing. be long
[29:17] scarcity be short abundance and that
[29:20] means that the black widow
[29:22] IWM versus QQQ no one I remember someone
[29:25] reached out in here when I started
[29:27] really going through this because I
[29:28] wrote my PMI piece right here and in
[29:31] there I said when PMIs go higher on a
[29:33] surprising basis you should see small
[29:35] caps outperform and the reason I don't
[29:37] think Q's are going to go where they
[29:38] normally do okay during this is because
[29:41] they will be disrupted by AI because the
[29:43] only way that PMIs can go higher is if
[29:46] the capex spending continues to increase
[29:48] and the buildout for AI is real. And if
[29:50] you believe that's going to happen, the
[29:52] PMIs will eventually go higher because
[29:54] AI will be put into everything. Well,
[29:56] that's where we are now. So, I expect
[29:57] this to fully go. If you want to hedge
[29:58] your portfolio, just make sure you don't
[30:01] have a lot of shorts on IWM.
[30:04] The amount of people that have tried to
[30:05] pick the bottom of this trade, and this
[30:08] is just a gift that keeps on giving from
[30:10] my perspective, um IGV over SMH. These
[30:14] are the weekly moves. This was a great
[30:16] week for it. It was only down 1%. I
[30:18] remember the first week of the year,
[30:20] everyone kind of calling saying, "I
[30:21] think this thing has peaked. I want to
[30:23] be long IGV, short SMH." Look at these
[30:26] numbers and just what has happened. I
[30:27] mean, we've had three numbers of 8% or
[30:30] more. Semis continue to move away. I
[30:33] want to be long the analog names and
[30:35] this is at a time without Nvidia doing
[30:37] much. Retail has decided to jump into
[30:40] IGV and put money in. Now, I will give
[30:43] Retail credit. They were the ones that
[30:45] really took us out of the liberation
[30:47] day. So may maybe they're going to
[30:49] figure something out that hasn't worked
[30:50] yet because uh that was before last
[30:53] week. But a lot of money going in there.
[30:55] So we'll see if we can get a bounce.
[30:56] Maybe retail can hold things in for a
[30:58] little while. The earnings on software,
[31:02] they're at alltime highs. So this is
[31:04] obviously one year forward. So it resets
[31:06] when we get to through the first quarter
[31:08] earnings. You're back up here. Um, so
[31:11] again, it's getting cheaper and they're
[31:13] still forecasting earnings. This is not
[31:15] about earnings this year. And I think
[31:16] that's what's caused a lot of problems
[31:18] for people and why they want to step in
[31:19] and buy it. SAS isn't dead, it's worse
[31:22] than that. This is a good, good piece.
[31:24] It's the first one I've seen written on
[31:25] it. We're finally starting to get to
[31:27] some stuff that I've been saying and I
[31:28] believe in. SAS isn't dead, it's worse
[31:30] than that. The comment that is made in
[31:32] here. The real threat to SAS was never
[31:34] demand destruction. It's supply
[31:36] explosion. So the demand side is what
[31:38] shows up in that chart I just showed.
[31:40] These companies are getting
[31:41] Salesforce.com is getting people to sign
[31:44] up. Microsoft has 3.3% people pay of
[31:47] users paying for AI co-pilot. Uh I'm not
[31:51] one of them and I find it to be one of
[31:52] the silliest things in the world to use
[31:53] it. But regardless the cost of building
[31:56] software approaches zero. You don't get
[31:58] fewer software companies, you get 10
[32:00] times more. What people don't understand
[32:02] is I would rather have a turbulence
[32:04] model that's a little bit buggy that I
[32:05] have to go into clawed code and get
[32:07] fixed and change myself than get a set
[32:11] thing that I can't customize. So will
[32:13] that work for an enterprise? Absolutely
[32:15] not. Will it work for startup for
[32:17] startup businesses? Absolutely
[32:19] positively. Will it work for
[32:20] entrepreneurs? 100%. Well, there's 8
[32:22] billion people on the planet. There's
[32:24] not that many enterprises. So as 10
[32:27] times more competition chasing the same
[32:29] customers means margins. I have seven
[32:32] revenue streams now. I could probably
[32:34] have 50 but I don't have the time. Each
[32:36] one of those re revenue streams
[32:38] absolutely competes with an enterprise
[32:40] company 100%.
[32:44] Prices drop, margins compress, customer
[32:47] acquisition gets harder and more
[32:48] expensive and that's what's going to
[32:50] happen. So when you are looking at these
[32:52] software companies and saying software
[32:54] is needed, I I tend to agree, but not to
[32:57] the way that you guys are thinking,
[32:59] you're not thinking about how AI every
[33:01] month gets faster and faster and the
[33:03] capabilities get better and better and
[33:05] the more that people use it, the more
[33:07] they realize it. And the more people
[33:08] that are out of work are going to need
[33:10] to use AI. This becomes cannibalism on
[33:13] the enterprises. It eats at their
[33:15] margins. Not for an energy company, not
[33:18] for a copper company, not for Corning.
[33:21] And that's the reason why I want you
[33:22] guys to be in there because we need to
[33:24] build that stuff out. And they've got a
[33:26] defense. The other side, anything built
[33:28] on code is up for disruption. I don't
[33:30] care what it is. It does not matter.
[33:32] This is a paper I'm writing this week
[33:34] about democratization.
[33:36] Um, falling marginal cost of capability
[33:38] that makes it rational for entrepreneurs
[33:40] and consumers to bypass incumbents
[33:41] entirely. That is a major theme. That is
[33:43] something I do every day. Uh, it's
[33:45] something I believe in. Why would I pay
[33:46] for C-pilot when I'm spending $12,000 a
[33:48] year for five employees, all the big
[33:51] packages for all the big AI fun to do
[33:53] help me with work? Andre Carpathy, a new
[33:56] kind of coding where you fully give it
[33:58] vibes. Embrace exponentials and forget
[34:00] that the code even existed. That wasn't
[34:02] poetry. It was a death certificate for
[34:04] software scarcity premium. The
[34:05] replacement cough of a 20 to$100 million
[34:07] ARR SAS product has fallen below 10,000
[34:09] in compute in one founders weekend. It's
[34:12] the speed and the competition. It it
[34:15] prevents you from ever getting to the
[34:17] point of monetization or remote. It
[34:19] means all of the public companies which
[34:21] I have said by 2035 I expect there to be
[34:23] complete disruption for all public
[34:25] companies the entire list of them. And
[34:28] the reason is because we'll have a rise
[34:30] of smaller businesses, entrepreneurial
[34:32] businesses. That's the ecosystem that's
[34:34] growing and that is run by AI agents
[34:37] which is where crypto and the entire
[34:39] framework goes. So again, you don't have
[34:42] to believe in Bitcoin. You don't have to
[34:43] believe in stable coins. You don't have
[34:44] to believe in tokenization. But in the
[34:46] same way that you have faded the bubble
[34:48] of AI, you will then suffer the pressure
[34:49] of the new capital structure of the
[34:51] future. Palunteer, I wrote this week. I
[34:55] also released it on Substack to try and
[34:57] get people to realize the research that
[34:58] I do on 22V. Every week I'm putting
[35:01] something out. I'll highlight what it
[35:02] is. How to hedge bubble AI edition. I'm
[35:04] only bringing this thing up there
[35:06] because remember that it was not more
[35:08] than three months ago that there was an
[35:10] AI bubble and now we've got capex
[35:12] spending which is going to be 70% higher
[35:14] than what was expected a year ago for
[35:16] this year. This is the negative that
[35:18] comes. So I'm not some uber bull. If I'm
[35:21] starting to sound bearish, it's because
[35:22] we're in the most destructive phase of
[35:24] capitalism where continuous innovation
[35:26] incessantly revol revolutionizes
[35:29] economic structure from within
[35:30] destroying old industries while creating
[35:32] new ones. I've talked about Joseph
[35:33] Shumpeder and his whole theory of
[35:35] creative destruction. The entrepreneurs
[35:38] role in this. Entrepreneurs are the
[35:39] agents of this change. Introducing new
[35:41] goods, methods of production, markets,
[35:43] organizational structures. That is
[35:45] effectively the exact theme I'm talking
[35:47] about with AI and with crypto. He called
[35:50] it back in his 1942 book, Go Spend Time
[35:53] on AI. And there it is. This is again
[35:56] creative destruction to the point where
[35:57] eventually it gets so fast and so
[35:59] powerful that even the most powerful
[36:02] suffer at the hands of the small. That
[36:04] is the fourth turning. That is the
[36:06] revolution that everyone has been
[36:08] worried about for the last 15 years as
[36:10] the distribution of wealth get worse.
[36:12] Now everyone who wants to use AI is very
[36:15] very powerful and they're able to build
[36:16] businesses and get revenue streams out
[36:18] of nowhere if they want. So here's the
[36:19] negative side. So for everyone out there
[36:21] in the mutual fund side, everyone out
[36:23] there on the private wealth management
[36:24] side, everyone who's an FA, all of you
[36:27] have so much money in these names and I
[36:30] know everyone is going on TV saying the
[36:32] Mag 7 is going to win this, they're
[36:34] going to win this. I do not believe
[36:35] that. So at some point over the course
[36:37] of the next two months, if you start to
[36:39] doubt the people that you listen to and
[36:41] you want to hear the story on why this
[36:43] can't happen, why they are not going to
[36:45] win this, this is the beginning of that
[36:47] journey for me to explain it. So this is
[36:49] the MAG seven chart. We're getting close
[36:50] to the 200 day moving average. Very
[36:53] dangerous for us to be breaking below.
[36:54] We're at the lowest close since
[36:56] September. So, something different is
[36:58] happening. It is impossible for the S&P
[37:00] 500 to break out when such a high part,
[37:03] the concentration we've heard about is
[37:06] not going up. If you look at it relative
[37:08] to the S&P, we've already broken the 200
[37:10] day moving average. This is this is not
[37:13] good. Now, if I bring up a longer term
[37:16] chart, this is when chat GPT was
[37:19] released. They won. The MAG7 was going
[37:21] to be the winners. When this happened,
[37:24] this was bad for the market. When this
[37:26] happened, this was bad for the market.
[37:28] If I did an overlay with this versus the
[37:30] S&P, when this is going down, usually
[37:32] the S&P is going down. I'm not calling
[37:34] for that to happen, but there's a risk
[37:36] that's growing in it that if this goes
[37:37] at a fast pace, it will get disruptive.
[37:40] You can have this kind of go on and not
[37:43] have the market go down because the
[37:44] market was actually going up during
[37:45] this. What you can't have is this turn
[37:48] into a waterfall which is what's
[37:49] happened with software.
[37:52] The MAG7 relative to the S&P though is
[37:54] not the issue and this is what I want
[37:56] you guys to start focusing on. When
[37:58] chatpt was launched very quickly in 23
[38:02] when you started getting these gaps
[38:03] higher the winner was going to be
[38:04] Microsoft because Microsoft was in
[38:06] everyone's machine and they were going
[38:07] to issue co-pilot and 10% to 20% to 30%
[38:10] of the people that have Microsoft were
[38:12] going to use it. Uh Microsoft to me has
[38:16] a lot of issues in this. Um that's just
[38:19] me. You guys can do what you want. But
[38:20] this is Microsoft relative to the S&P.
[38:23] Uh we're back to where we were in 2020.
[38:25] This name has lagged. This is a
[38:27] waterfall chart. I don't like waterfall
[38:29] charts and trying to pick the bottom.
[38:30] And I usually think there's something
[38:32] wrong. What is wrong is this is when we
[38:34] started to realize that AI was moving
[38:36] faster and Claude and Anthropic in
[38:38] particular was starting to dominate the
[38:39] enterprise world. That's how fast this
[38:41] has been guys. Anthropic started to
[38:43] dominate and then when you get into here
[38:45] and here, this is when Claude became
[38:47] there. I do believe that people in
[38:49] Silicon Valley that are long mag 7 that
[38:52] are long any of this, they recognize
[38:53] what's happening. You cannot have claude
[38:55] release claud me use it for a week
[38:59] claude powerpoint I didn't use it
[39:01] because I use nano banana the point is
[39:03] why do people need Microsoft and need to
[39:05] pay the money for copilot when they can
[39:07] just use chatpt Gemini and all these and
[39:09] then oh by the way are the Chinese
[39:10] models which are free so this is a
[39:13] problem but we've now taken out the lows
[39:16] of where it was when chatpt was launched
[39:20] it is connected to open AI so here's the
[39:22] scary chart
[39:24] Microsoft, Amazon, Google, yes, and
[39:27] Meta, the hyperscalers, the ones
[39:28] spending the 650 billion, not including
[39:32] the money that Oracle spending, not
[39:33] including OpenAI, not including
[39:35] anthropic, not including XAI. There's a
[39:37] lot of money being spent in a massive
[39:39] race. And then you have the Chinese
[39:40] models, which are free, the hyperscalers
[39:42] relative to the S&P. Now, a lot of this
[39:45] is related to the earnings that came out
[39:46] in the gap down, but this is an equal
[39:49] weight index of the hyperscalers. This
[39:52] is for you guys where I'd be using it as
[39:54] a funding cost. I think you want to be
[39:56] long. Everything's scarce and against
[39:58] it, you want to be underweight. All of
[40:00] the hyperscalers, those four companies
[40:02] that I mentioned, I think those four
[40:04] companies have a big issue. So, this is
[40:06] the hyperscalers relative to the S&P
[40:08] about to hit the lowest level since 23.
[40:11] If they don't change something, it's a
[40:13] problem. And for everyone saying come in
[40:15] and buy them. I when I get to the point
[40:17] where I'm going to show you what has
[40:19] changed over the last six to eight weeks
[40:22] is new news. Are they being basian? Are
[40:25] they taking in new news and even
[40:27] referencing this or are they just saying
[40:29] gobbledegoop of they will be the
[40:31] winners? That's what Kathy Wood said
[40:33] before I went on is that I've studied AI
[40:35] companies for 15 years. I know the ones
[40:37] that were building this stuff. They have
[40:39] the edge. I didn't use AI for that
[40:42] entire time. I had data scientists.
[40:44] Everything that I'm able to do that you
[40:46] guys see on this, every reason I'm
[40:47] trying to help people with my payw wall
[40:49] to both invest but also to learn how to
[40:51] use it that comes from knowledge
[40:52] recently. You don't have a l a legacy
[40:55] advantage by being in there. Elon Musk
[40:58] said software guys are becoming hardware
[41:00] guys. If you didn't listen to this and
[41:01] you want to question your own beliefs in
[41:03] the hyperscalers and not just wake up
[41:05] and say they're going to win because
[41:06] they're big, that is the framework of
[41:08] the last 15 years. That is not the
[41:10] framework of the future. The people who
[41:12] thought that they were just writing code
[41:14] now have to think about transformers,
[41:16] substations, cooling, and energy
[41:18] generation. And you know what? They know
[41:19] nothing about it, but they've hired
[41:22] someone who knows something about it.
[41:24] But that person is like the
[41:26] subcontractor or the contractor on your
[41:28] home where things never get done on time
[41:30] and where they can't possibly know every
[41:32] single supplier and issue. So when I put
[41:34] transformers, substations, cooling,
[41:36] energy generation, which is what Elon
[41:38] put, there are so many things on here.
[41:40] There's memory, high bandwidth memory.
[41:43] There's optical fiber that Meta is
[41:45] trying to hoard. There's copper. There's
[41:46] silver. There's rare earth. Any one of
[41:48] these can delay this coming out. And
[41:52] then you have a real problem because you
[41:53] can't get the revenue without the
[41:55] capacity. The capacity is not there. Go
[41:59] listen to this podcast. I've shown how
[42:01] important this is. Some of you have
[42:03] listened to it and agreed with me. If
[42:05] you haven't listened to it, again,
[42:07] there's a lot of this that is not about
[42:09] copper and not about silver. Jeff Curry
[42:11] does a great job of explaining why the
[42:14] hyperscalers are at risk. His asset
[42:17] heavy point, and these charts are going
[42:18] around, asset light versus asset heavy.
[42:21] You didn't want to be long asset heavy
[42:23] over the last 15 years. You wanted to be
[42:24] long asset light. Well, now the asset
[42:27] light companies are becoming asset heavy
[42:29] in the hyperscaler side. He makes this
[42:31] thing. The digital economy is now
[42:33] colliding with the physical world. in
[42:34] the physical world has constraints. Talk
[42:36] about it every week. You can go through
[42:37] it. Here are the IPO the RPO numbers.
[42:40] This is
[42:42] $1.2 trillion now and growing. This is a
[42:46] big problem because this is dependent on
[42:47] getting the data centers done. So that
[42:49] chart I just showed you to get these
[42:51] numbers we need more capacity. To get
[42:53] more capacity we need the software guys
[42:56] to be able to build hardware. If they
[42:58] can't get it then Dario Modi has said
[43:00] there's a real problem because you're
[43:01] spending tons of money. So you're sp
[43:03] you're you're doing that and now we've
[43:05] raised the numbers on the capex
[43:06] significantly. This were revised higher.
[43:08] This was revised higher. We haven't even
[43:10] gotten the data centers done yet from
[43:12] the numbers on this. You're starting to
[43:14] get a backup. I don't want to say this
[43:16] is a problem because the demand is
[43:18] greater. But the issue is if the delays
[43:20] take too long, it does become a problem
[43:22] and they're obviously borrowing money.
[43:24] So it's not just asset heavy. They're
[43:25] having to come out. So Alphabet goes out
[43:27] and sells 32 billion. We've obviously
[43:28] seen OpenAI. So, here are the things
[43:31] that we've gotten new news on. The
[43:33] RPOS's continue to grow. Did anyone talk
[43:35] about this as a negative? They're
[43:37] getting more orders, but at some point,
[43:39] if they can't bring the capacity online,
[43:41] when does that become an issue? Number
[43:44] two, the Chinese models are improving
[43:47] rapidly. Free models
[43:50] connected to the expensive models, they
[43:53] are losing market share at without a
[43:56] doubt. Now, maybe not to US companies,
[43:57] but certainly to Asian, certainly to all
[43:59] the future enterprises or the future
[44:01] wannabe enterprises, all startup
[44:03] businesses. If they're trying to make
[44:04] money, they care about costs. Why
[44:07] wouldn't they use open source? So,
[44:09] again, I I you you've got Chinese models
[44:11] going. Claude code has accelerated. We
[44:14] didn't think we'd get to recursive
[44:15] self-improvement, but claude code has
[44:17] accelerated dramatically since November.
[44:19] The bottlenecks are growing. We're
[44:20] getting cancellations across the board.
[44:22] If you don't, if you're a multistrat and
[44:23] you've got utility teams, ask them how
[44:25] the capex is starting to show up in
[44:27] certain uh uh states, whether it's
[44:30] Michigan, Iowa, Missouri, how many
[44:32] things are being canceled, how many
[44:33] complaints are going out of Louisiana
[44:35] with the metawan. What's happening with
[44:36] Elon in terms of complaints? The
[44:38] bottlenecks in the data center delays
[44:40] are growing. It's one of the themes or
[44:42] theories behind why Elon Elon Musk is
[44:44] going to actually issue and go public
[44:47] merge. He needs more capital. And one of
[44:49] the things he wants to focus on is
[44:51] getting the data centers in space where
[44:52] he doesn't have to deal with all the
[44:53] issues. The capex's guidance remains
[44:57] going up higher. So they're spending
[44:58] more. We've got bottlenecks growing on
[45:00] the buildout that allows them to get the
[45:02] money on the RPOS. Their competition is
[45:04] growing rapidly and a private company is
[45:06] winning the battle in ter these are all
[45:09] news things in the last six weeks. So if
[45:11] Dan Ies is not saying this, if Morgan
[45:13] Stanley is not saying this, if JP
[45:14] Morgan's not saying this, if Goldman
[45:15] Sachs is not saying this, why are they
[45:17] not saying this? I just don't think they
[45:19] have the domain experience in this to be
[45:21] paying attention to data centers and to
[45:22] the RPOS's connecting them back to the
[45:24] usage. I just don't think people fully
[45:26] grasp and the startup world is going to
[45:28] be using the Chinese models. So Dario
[45:31] Mod did an interview with Dwaresh Patel
[45:34] came out yesterday.
[45:37] He talked again about how we are near
[45:39] the end of the exponential. Overbuilding
[45:41] is exponential. Underbuilding is the
[45:43] strategic loss. This is the trap to roll
[45:45] in. They have to build. If you
[45:47] overestimate demand, you've got
[45:49] bankruptcy risk. If you underestimate
[45:51] demand, you lose competitive position.
[45:53] This asymmetry forces cautious scaling,
[45:55] but most people are not doing this. And
[45:57] he talks openly about how, again, he's
[45:59] not naming them, but he's saying open AI
[46:01] has just spent too much money. He's
[46:03] basically predicting that they're going
[46:04] to be in trouble unless they can get
[46:07] more of the building done. Now, for
[46:09] those of you on the payw wall, one of
[46:11] the things I want to highlight, I'm
[46:12] releasing this each week. So, all of the
[46:14] podcasts that I go in here, you will get
[46:16] a list of them, the ones that I've used
[46:18] on the video. It'll have the link so you
[46:21] can just copy it and paste it and go
[46:22] into it. It'll have the details on it if
[46:24] you don't want to go do it. So, I'm
[46:26] giving everyone the same way whatever
[46:28] I've figured is the signal, I'm giving
[46:30] you guys access to the signal on the
[46:31] payw wall. Now, the reason the
[46:33] hyperscaler is so important, not just
[46:35] from those names of perspective, this is
[46:37] the relationship of growth verse value.
[46:40] So, if the hyperscalers go up, go down,
[46:42] we got an issue. The other new news that
[46:44] came out in the last six weeks that was
[46:45] not news is the size of the IPOs that
[46:47] are coming to the market. So with
[46:49] increase in capex, we have less
[46:51] buybacks. At the same point, you have
[46:52] more supply coming in. This chart's
[46:54] going to get worse because where's the
[46:55] money going to come from to buy SpaceX
[46:57] and XAI and Anthropic and OpenAI. It's
[46:59] going to come from the growth bucket.
[47:00] It's not going to come from the value
[47:02] bucket. So what you're going to have is
[47:03] the demand side. They're going to have
[47:04] to sell something. This is just going to
[47:06] continue to get worse over the course of
[47:07] the year. The only question is where's
[47:09] it going to go? You want to continue to
[47:10] be short growth for it value in my
[47:11] opinion. But the hyperscalers relative
[47:13] to it are critical. Here is the Mag 7
[47:15] relative to it. The reason I keep
[47:17] showing this, I still believe Tesla's
[47:18] going to be the best performer out of
[47:20] the Mag 7. Nvidia, I don't know what to
[47:22] do with it. It's still cheap. I do think
[47:25] that there's going to be an issue with
[47:26] the compute side going forward, but
[47:28] Nvidia to me is still the best place to
[47:30] be with inside the Mag 7, uh, outside of
[47:33] Tesla. And so, you're left with that.
[47:34] And then with Apple, the fact that they
[47:36] haven't spent the money, they haven't
[47:37] gone through, they don't have the same
[47:38] issues, but they definitely have a
[47:39] multiple compression issue. Their
[47:41] biggest issue is going to be did they
[47:42] lock in the supply chain for memory
[47:44] where they're going to be able to
[47:45] release things. They are sold out of Mac
[47:47] Mac minis at this point. I'd rather be
[47:48] long Apple than short it. The turbulence
[47:51] model I released something on Friday.
[47:53] And again, what I want to highlight here
[47:55] is how many green bars I've got. To get
[47:58] a green bar in this, what I need is a
[48:00] shock in the cobar matrix at the same
[48:02] time that the S&P is above the 50-day
[48:05] moving average and at the same time that
[48:06] the VIX is below 25. So what these are
[48:10] meant to do is say that the market's not
[48:11] ready for something and there's
[48:13] something big going on. When there's big
[48:15] going on, the Kovar matrix is really
[48:17] shocking. That's what the turbulence
[48:18] models measures. So again, for people on
[48:21] the payw wall, you get access to this. I
[48:22] wrote a paper on this to describe what
[48:24] it is. I'm not going to go through it,
[48:27] but again, I've I I
[48:30] been involved in risk management since
[48:31] my days at Morgan Stanley. I built a lot
[48:33] of the risk systems at the hedge funds
[48:35] uh I was at or at least constructed
[48:37] them. I always have paid attention to
[48:40] Kovar. That is the risk for the
[48:42] multistrap model. It is dependent that
[48:45] the market acts in a certain way. I
[48:47] believe the act of the market has
[48:50] changed significantly. And so the main
[48:52] point here is we have had approximately
[48:54] 12 to 15 green bars. And I say
[48:57] approximately because it depends on when
[48:59] you uh what you had as the close, what
[49:02] wasn't the close. The main point is you
[49:04] can see how stable it was. This was the
[49:06] only period in early 24 that we had any
[49:09] kind of multiple bars within a period of
[49:11] time. These are happening now
[49:12] relentlessly. So, so far in the last
[49:14] five weeks, we've had 12 to 15 and then
[49:17] we had about 20 to 25 in the prior 28
[49:20] months. So, it's telling me there's risk
[49:23] and it comes at a time when hedge fund
[49:25] gross leverage is near all-time highs.
[49:26] So, for anyone who's doubting this, I
[49:29] don't know what you're talking about. If
[49:30] you don't think gross leverage is high
[49:32] and you haven't measured what's gone on
[49:35] since chat GBT and the fact that we've
[49:37] seen this go on consistently, gross
[49:39] leverage, it's not even a question. Net
[49:41] leverage not the same. In fact, net
[49:44] leverage kind of middle of the range
[49:46] except for this one pukage down in in
[49:48] during liberation day. So, it's not the
[49:50] net leverage. So, for everyone who's a
[49:51] long short hedge fund that's like
[49:53] leverage is not high, that's the one you
[49:55] care about, great. It doesn't really
[49:57] matter. Same thing going on o over here
[49:58] in terms of looking at it. What you have
[50:00] here though is the gross leverage and
[50:02] this is the issue that comes up when
[50:04] gross leverage for long short is at
[50:06] those levels. We know gross leverage is
[50:07] high for the multistrat side because of
[50:09] where V has been and what and the
[50:11] correlations that have gone on. What you
[50:13] normally get in what's happened over the
[50:15] last week especially when we've got
[50:16] stuff going on is people hedge their
[50:17] nets. So the nets go down gross goes up
[50:20] because you add shorts on top of it to
[50:22] protect against the downside move. The
[50:25] worst thing is when you spend a month
[50:26] not falling and you just go sideways.
[50:28] Now, what I look for once the turbulence
[50:30] model is telling me there's risk
[50:31] growing, if I want to see if there's
[50:33] going to be a big unwind across assets
[50:36] and especially in the multistrap world,
[50:38] what I want to look for is for credit to
[50:40] start to weaken. The only time that that
[50:42] can happen in my opinion is when we get
[50:43] a credit event. Now, credit weakening at
[50:46] the same time as the market going down,
[50:48] not going down is a big issue. And
[50:50] that's where we're starting to see at
[50:51] this point because one of the big hedges
[50:53] for credit, especially widening credit,
[50:56] ends up being VIX options and eventually
[50:58] it will catch up. But in the meantime,
[51:00] if your measured mark month-to-month,
[51:02] this can be an issue. Now, I'm showing
[51:04] here the levered loan total return
[51:05] index. This peaked here, we've been
[51:09] weakening. This is at the end of the
[51:10] year. So, January, we've seen this
[51:12] weaken at this point. This is a little
[51:15] bit alarming. The only time we've been
[51:17] below the and this is the 100 day moving
[51:18] average, the green line. The only time
[51:19] we've been below the 100 day moving
[51:21] average since we came out of the rate
[51:22] hike cycle was this period in liberation
[51:24] day and we were leaning over before it.
[51:26] We obviously have seen weakness because
[51:29] we've seen the private credit fears. We
[51:31] saw the the auto lender fears everything
[51:33] happening in September. That's what I
[51:35] would worry about. So HY to IEF is my
[51:38] liquid version of this. I never like it
[51:41] when it's going down. It's a great proxy
[51:44] for anything related to credit fears in
[51:46] terms of getting liquidity. Well, it's
[51:49] trending lower at the same time that I'm
[51:51] getting turbulent shocks, which means
[51:52] we're getting some contagion. Now, here
[51:55] it is overlaid with junk spreads, just
[51:57] to give you an idea of how it leads.
[51:59] This has always been my big proxy HY
[52:01] relative to IEF over the junk spreads.
[52:03] You're getting junk spreads to widen.
[52:05] You've also seen it on uh option
[52:08] adjusted spreads. It doesn't really
[52:09] matter. Everything is kind of leaning
[52:11] higher and weakening. This is the BDC
[52:15] environment. So, it's not like private
[52:16] credit has gotten better. So this is the
[52:17] BDC index. We're still sitting down
[52:20] here. So we started to break down during
[52:22] liberation day. We bounced all the way
[52:23] back up and then in September is when we
[52:25] started to get the weakness. So
[52:26] remember, we've had a weakening credit
[52:28] market. We have delinquencies happening
[52:29] at the consumer side. As long as it
[52:32] doesn't all flush at once, it's not a
[52:34] big deal. I think we can get through it.
[52:36] But the problem is if we start getting
[52:38] hedge fund deleveraging, you're going to
[52:40] end up impacting credit markets. You'll
[52:42] impact the V markets. and then you'll
[52:44] have a scenario that rates are going to
[52:46] move lower and blah blah blah. Well,
[52:47] that's what we're seeing. So again, I'm
[52:50] just saying that there's a growing
[52:52] probability of an event to occur which I
[52:54] think will happen at some point this
[52:55] year. You want to look the contagion is
[52:58] r is increasing. One of the issues and
[53:00] Warren Pies brought this up and I
[53:02] completely agree with him. Leadership is
[53:04] transitioning to an unhealthy place with
[53:07] energy and staples leading. So he has
[53:09] this chart here. When staples and
[53:11] utilities are the best performers, it's
[53:14] not great. Energy also not great. More
[53:16] importantly, you need financials and
[53:19] tech normally to be at the top. Well, we
[53:21] know tech's not going to be there if I'm
[53:23] right about what I'm showing. But
[53:25] financials join the party this week.
[53:27] That is the risk that goes with credit.
[53:29] Financials had a very bad week. And it's
[53:33] been insurance brokers. It's been in
[53:34] wealth management. you haven't seen
[53:36] Goldman and the the upper echelon
[53:38] things, but the BKX underperformed the
[53:40] S&P significantly this week. So, I would
[53:42] watch this stuff. The other place where
[53:44] this will spread, services and software.
[53:47] So, obviously the software and services
[53:48] are falling. The reason I brought up
[53:50] MSCI world for this is because I just
[53:52] want to make sure you realize it is a
[53:54] huge percentage of the market. And more
[53:57] importantly, this is how much of it is
[54:00] the US 90%. So, I will keep saying it.
[54:03] If you guys want to follow my scarcity
[54:05] routine, the other thing I'm worried
[54:07] about if this turns into an event where
[54:10] there's a deleveraging, the Fed will be
[54:13] forced to cut rates, provide liquidity,
[54:15] do what they normally do, if it is
[54:16] happening at the hedge fund level, which
[54:18] I think can be a risk. If that happens,
[54:21] the dollar is going to weaken. If
[54:22] there's repatriation out of the MAG 7
[54:24] and out of the hyperscalers and out of
[54:26] software, that should be weak dollar.
[54:29] That's what I think is going to happen.
[54:30] And so an embedded put against what I'm
[54:32] saying is to be long foreign stocks X
[54:35] the US. Look at this chart. This is MCI
[54:38] World X the US. If you are not looking
[54:40] at this, if this is not a benchmark for
[54:42] you, it was unchanged
[54:44] from the end of 2007 all the way to the
[54:47] end of 2024.
[54:50] This is what you want to be long. And it
[54:51] is accelerating. This is a weekly chart.
[54:54] I I just I literally cannot say it loud
[54:56] enough. I'm going to keep pitching
[54:58] foreign foreign stocks. Now, I'm going
[55:00] to keep showing Brazil. So, Brazil, the
[55:04] sell rate, this orange line. You guys
[55:05] know I lived in Brazil. I love Brazil. I
[55:08] follow Brazil. Uh here's the green line.
[55:11] EWZ, I want to be long this for all the
[55:13] reasons I put. For those of you with the
[55:14] payw wall, go read my research piece
[55:16] from last week. But here's the CRB Ren,
[55:19] the white line. We like this because we
[55:20] like scarcity. We like metals, minerals,
[55:22] all of that stuff should be trending
[55:24] higher. The whole thing. And I believe
[55:26] they're going to be cutting rates. So
[55:28] instead of saying me, just copy this
[55:30] chart right off my screen. Knock my head
[55:33] out of it. Put it into an LLM and say,
[55:35] "Hey, I want you to look at this chart
[55:36] and tell me what I when I should be long
[55:37] EWZ."
[55:39] Uh uh you want to be long EWZ when the
[55:41] ideal the sweet spot is when you have a
[55:44] rising CRB raw industrials index coupled
[55:46] with a falling or destabil or
[55:47] stabilizing at low level select rate. So
[55:49] if they're going to be cutting rates,
[55:50] which is going to start, there's a 90%
[55:52] probability in the market that they're
[55:53] going to be cutting in March. We've got
[55:55] this going higher and I think this is
[55:56] structurally going higher. It says
[55:58] that's when you want to be long EWZ. You
[56:00] don't have to listen to me. I don't need
[56:01] to get on a panel and debate this with
[56:03] people who are bearish on Brazil and
[56:04] talking about politics. I really don't
[56:06] care. This is what is going to happen.
[56:08] You're going to have rates cutting
[56:09] lower. These guys have a thousand basis
[56:12] points of real rates at this point. They
[56:14] have plenty of room to cut and the
[56:16] deflationary pressure. So if you believe
[56:17] in trueflation, if you believe in tenure
[56:20] rates coming down, if you believe the US
[56:22] economy is going to be strong and that
[56:23] the spending is going to happen and it's
[56:25] going to be focused mainly on minerals
[56:26] and commodities, historically this would
[56:29] say be long Brazil. And I'm going to say
[56:31] for many, many reasons outside of that,
[56:32] you want to be long it. You want to be
[56:34] short anything built on code and
[56:37] abundance. That's what this whole thing
[56:38] is. And then scarcity, these ugly things
[56:40] here, utilities, copper, optical fiber,
[56:44] any of it, any of that stuff. Well,
[56:45] there's there's CVX. Whoa. Chevron. I
[56:48] got told by so many people this was not
[56:50] a good place to be because it doesn't
[56:51] participate. We're about to break out
[56:54] again. It's been basically dead money
[56:57] for over a decade.
[57:00] The rotation will happen. They will
[57:02] benefit if we need to spend money and
[57:04] build out these things. The same thing.
[57:05] Eaton another name. It's been sideways.
[57:08] When you can find these patterns, these
[57:10] kind of bases that John Rog likes to be
[57:11] in where it was unchanged from the
[57:13] middle of 24 to early 26th.
[57:16] almost two years and now it's starting
[57:18] to break out. Their name shows up on
[57:20] almost every single screen that I do on
[57:22] benefiting and it's not just one
[57:24] component. The one thing about Eaton is
[57:26] they're diversified and they've got a
[57:28] lot of revenue streams that are showing
[57:29] up. They will also show up on the
[57:30] optical fiber side, everything. But if
[57:33] you don't like them for the reasons I'm
[57:35] saying, just know that PMIs just had
[57:37] their biggest upswing outside of coming
[57:39] out of COVID since back in I don't even
[57:42] know how long what did I say it was? 95,
[57:44] 92, I don't even know. It's a big PMI
[57:47] jump. So again, you've had low PMIs for
[57:50] a long time. Underweight, it's breaking
[57:52] out. The thing I'm gonna write about
[57:54] this week, the next place. So you guys
[57:56] like Corning. I get a lot of fanfare
[57:57] from Corning. Well, chemical time,
[58:00] boring chemicals, Vanderlay Industries,
[58:02] Plastics, Latex, all this stuff. I'm
[58:05] going to write a paper on it and give
[58:07] you guys some names on the payw wall. Um
[58:09] the Corning and Meta deal is what
[58:10] triggered the research and I wrote I'm
[58:14] writing finishing this is part of my
[58:17] initial thing but how the meting deal
[58:19] signals the beginning of a multi-deade
[58:20] broadening cycle in AI infrastructure
[58:22] materials. If you guys are looking for
[58:24] names that nobody could possibly own and
[58:26] ones that are boring and not part of the
[58:28] AI cycle chemicals are going to be a big
[58:30] part of it. And look how many chemical
[58:31] names I will this is the John Rog I'm
[58:33] not going to show you his technical
[58:34] score. This is for the people on the
[58:35] payw wall and the institutional people,
[58:37] but I will be putting something out on
[58:39] this in terms of names. If you guys want
[58:40] the names, just go sign up and spend the
[58:42] time. I will go here's one freebie in
[58:44] terms of things to look at. Selony
[58:46] announces acetic acid blah blah blah
[58:49] price increases. We're looking for price
[58:51] increases on things. We're looking for
[58:54] companies to be talking about it. Again,
[58:57] you can go to the payw wall. You can go
[58:59] there.
[59:00] Bitcoin. Saw a lot of people there. I
[59:03] spoke after Kathy Wood. Um I am
[59:06] disappointed that Bitcoin has gone
[59:08] lower. But if you would have told me
[59:09] that software would end up falling as
[59:11] much as it has, and again this is
[59:13] basically when the software problem
[59:16] started, uh I didn't think we get a
[59:20] waterfall. You can't get away from the
[59:21] fact that software and and Bitcoin
[59:23] historically have been correlated. Uh
[59:25] that's why I don't believe in this
[59:26] liquidity thing. Uh other than the fact
[59:29] that risk on means you're not having a
[59:31] recession. So if you strip out
[59:32] recessions, liquidity to me has always
[59:34] been a lacking argument I don't really
[59:36] care about. Uh I like the correlation,
[59:38] but this correlation makes more sense to
[59:40] me and it's what I believe in. If you
[59:41] guys go back and listen to Michael
[59:43] Sailor, there are two things he talked
[59:44] about. One is the debasement side.
[59:46] Everyone gets fixated on that. That's
[59:48] where the liquidity side comes from. My
[59:50] argument has always been the Joseph
[59:51] Shumpeter side. and living in Brazil.
[59:53] The real theme behind Bitcoin is the
[59:56] fact that there will be very few winners
[59:58] on the enterprise side and the people
[60:00] that have money that are invested in the
[60:01] Kypers scalers in the Mag 7 eventually
[60:04] will see their assets start to become an
[60:06] issue and they will be forced to move
[60:08] into things. Bitcoin is a growth asset.
[60:10] There's no doubt about it. It's built on
[60:12] code. It's innovation. It's the future.
[60:14] The ecosystem is part of it. They are
[60:16] related to VC. They're related to all of
[60:18] it. I think all of the money on the VC
[60:21] world should not be going into AI, it
[60:22] should be going into Bitcoin. It should
[60:24] instead of the NDX, which outperformed
[60:26] private credit, private equity, NVC over
[60:29] the last decade should be going into
[60:30] Bitcoin. We'll see if I'm right on this,
[60:33] but I do believe that the software side
[60:35] has created an opportunity. So far,
[60:37] we've held in well in Bitcoin this week.
[60:39] If we have another waterfall leg, like I
[60:41] said, and this becomes a deleveraging
[60:43] event, Bitcoin will absolutely head down
[60:45] to 40,000 with everything else with it.
[60:47] But then I would expect rate cuts,
[60:49] dollar weakening, and everything to set
[60:51] up where software reaches a level where
[60:52] it's getting cheaper by the day. If we
[60:54] have another 20% fall, you're going to
[60:56] have the earnings for these things down
[60:58] into the teens. And I think that's where
[61:01] they do become value names that everyone
[61:03] will be moving into. Here is the
[61:04] relationship between Bitcoin and the
[61:07] software side. When you do a ratio of
[61:08] them, what's critical on this is look at
[61:11] the beta on the upside. So for everyone
[61:12] who's an endowment, a foundation, a
[61:14] pension fund, anyone who's making
[61:16] investments in this, the software side,
[61:18] Bitcoin outperformed on the way up, it
[61:20] has underperformed since September. But
[61:23] again, when you look over the course of
[61:25] this time period, it's gone sideways. I
[61:27] believe once software finds a bottom,
[61:30] this will go higher. What may happen is
[61:32] software has another leg. I think it'll
[61:34] be one for one. Bitcoin is this year
[61:36] down about as much as software. It's
[61:38] about exactly the same. So it is not
[61:40] having a higher beta on the way down. If
[61:42] you haven't read this, go read it. This
[61:44] again gets into the tsunami side of how
[61:47] fast this is coming, how many jobs are
[61:49] at risk on the enterprise side.
[61:51] Absolutely want to be paying attention.
[61:53] I believe you read this, you go see the
[61:56] impact that is going to have. And I
[61:57] think you need to make sure your kids
[61:59] are learning AI.
[62:02] That's why I do these things. And I'm
[62:03] glad for the students that get heavily
[62:05] involved with me from the college side.
[62:07] Um this is just an unbelievable story.
[62:10] But remember Dario Modi wrote a paper
[62:12] about how risky things are getting. Uh
[62:15] he had written an optimistic piece less
[62:17] than a year ago. Now he wrote a
[62:18] pessimistic piece in January. He had one
[62:21] of his lead researchers on the safety
[62:23] side. An AI safety researcher has quit
[62:26] anthropic
[62:27] that the world is in peril. So this is
[62:29] someone who knows what's going on and he
[62:31] quit to study poetry. That's how scared
[62:33] he is of the impact that is coming. I
[62:35] asked to combine both the Dario Modi
[62:37] piece with the piece what he said coming
[62:40] out. The world is in peril not just from
[62:43] AI or or bioweapons but from a whole
[62:46] series of interconnected crisis
[62:47] unfolding in this very moment. This is
[62:50] the thing for regulators and for
[62:51] everyone out there. The reason I care
[62:53] about the turbulence model is because of
[62:55] how fast the tsunami is moving. At some
[62:58] point here you have to realize there is
[63:00] 0.0 chance that AI won't be getting
[63:03] stronger. that AI agents won't be out
[63:05] there and that leads to a lot of dangers
[63:08] and nobody is prepared for it. And so
[63:10] there's negatives that come and there's
[63:12] positives. We're going to start to see
[63:14] the negatives in this this year because
[63:16] recursive self-improvement and the
[63:17] escalation of cyber risk is real. So
[63:20] again, just remember my job here in
[63:23] launching the payw wall is to help you
[63:25] guys from a perspective of domain
[63:27] experience on the investment side, but
[63:29] also now in the AI native side. For
[63:31] those of you who look at me as some
[63:32] researcher, some whatever, I know more
[63:35] than most of the banks do because I've
[63:37] spent more time than a lot of them in
[63:40] the markets in connecting macro. AI is a
[63:42] macro theme that goes across sectors. It
[63:45] is not siloed. It is not a technology.
[63:47] It's impacting everything. And if you
[63:49] didn't think that was happening, you saw
[63:50] it this week. I understand how crypto is
[63:53] connected. This is where you want to be
[63:54] putting your money once software finds a
[63:57] bottom. Either is going to come from
[63:58] software stabilizing and being a value
[64:01] trap or it's going to come from the Fed
[64:02] having to come in and do something
[64:03] because the market has a disruption from
[64:05] hedge funds that are deleveraging. One
[64:06] of the two, in my opinion, is going to
[64:08] happen this year because of the
[64:09] waterfall that's going, it is not a time
[64:12] to step in and buy software. And for
[64:14] those of you who realize that the risk
[64:16] is growing every day, I will I am
[64:18] putting out videos. There's one already
[64:20] up on there. There will be a complete
[64:21] webinar series to help you learn. Have a
[64:24] great uh long weekend. I will see you
[64:27] guys next week. And for those of you
[64:29] who've signed up, thank you very much. I
[64:30] appreciate again for those of you who
[64:32] reached out to me at the events. I'll be
[64:34] doing more events in New York City and
[64:36] then I'll be traveling to some bigger
[64:37] events over the course of May and April.
[64:40] Thanks. Have a great weekend. And I'll
[64:41] see you guys.