Financials Are Warning Something Is Wrong: Oil, AI and Credit Are Colliding — Jordi Visser (8 marzo 2026)

[0:00] Uh, sorry for the mic confusion last
[0:03] week. Apparently had some issues with
[0:05] the mic. So hopefully we're back to
[0:07] normal this week. Um, things keep uh
[0:11] snowballing right now. So um there's a
[0:14] lot to cover. I think we're uh you know
[0:17] the message of this week before I go
[0:19] through this is that you know I I think
[0:22] uh everyone is thinking way too binary
[0:24] on everything. I think this year is
[0:26] going to be very complicated the way
[0:28] things have started and especially since
[0:31] credit is becoming more of an issue. And
[0:34] if there's one thing that I have been
[0:36] through since the beginning of my
[0:38] career, the '9s for me was all about
[0:40] emerging markets and credit events. Uh,
[0:43] I enjoy credit events because usually it
[0:46] ends up with uh the Fed needing to come
[0:50] in and do something and especially when
[0:52] it involves retail being trapped and
[0:54] that's effectively where we are. So,
[0:57] thanks again to everyone who's been
[0:59] reaching out. Uh, you know, hopefully
[1:01] the webinar from 10 days ago helped you
[1:04] guys prepare uh with the VIX. Uh there's
[1:08] a lot more ideas and I'm going to be
[1:10] doing a lot more um for the site over
[1:13] the course of the next month. Uh the
[1:15] video should be done in terms of
[1:17] training for AI. I'll go through a
[1:18] little bit of that um when we get into
[1:20] some of the idea portion. But I do think
[1:23] again this is more of a trading
[1:24] environment. I think this is going to be
[1:26] a year of deleveraging. Uh it's not a
[1:28] bearish thing. It is a necessary thing
[1:30] that has to happen because of the
[1:33] crowdedness in covariance models for
[1:36] risk. Uh I'm writing a paper this week
[1:38] comparing this time in my opinion to
[1:40] LTCM and the Quantquake uh back in 2007.
[1:45] And for those of you again think this is
[1:46] bearish. Both of those years um you know
[1:49] we went much we we went up higher levels
[1:52] after that. Uh this is going to be a
[1:55] year where I still think AI is going to
[1:56] dominate. Earnings will be good. Uh
[1:59] growth will be fine. Uh it'll probably
[2:01] uh depend somewhat on oil now. But I'll
[2:04] go through all of that and uh we'll keep
[2:07] trying to make money together.
[2:12] So, I just wanted to bring this up
[2:14] because this is the one thing I will say
[2:16] and if you guys saw the viral um Glenn
[2:19] Beck uh interview with George
[2:21] Washington, AI avatar George Washington
[2:25] uh in X, he did mention one thing in
[2:28] there. There was one thing that in my
[2:29] opinion was factual which is important
[2:31] for the rest of this year. To be
[2:33] prepared for war is one of the most
[2:34] effective means of preserving peace. I'm
[2:36] doing this from your portfolio. Uh I
[2:38] think you need to be prepared for more
[2:40] to go on. And I think people that are
[2:42] just looking for a bottom to buy things
[2:43] thinking we're going to go through what
[2:45] we went through from the end of 22 until
[2:47] this is not going to happen. Uh this is
[2:49] not liberation day. These problems
[2:51] cannot go away with just a reversal of
[2:55] something that the administration
[2:56] decides. uh regardless and people are
[3:00] too focused right now on Iran and I
[3:02] agree this is a very important thing for
[3:04] the markets and will have an overhang
[3:06] but it's not the only thing going on.
[3:08] I've been more focused on AI and credit
[3:10] and that's where I'm going to continue
[3:11] to drive this. So I just want to make
[3:13] sure you realize this is a different
[3:15] environment than before and the
[3:17] conditions of the market matter as an
[3:19] old handicapping uh side. You can spend
[3:22] all your time handicapping a race like
[3:24] the Kentucky Derby and then all of a
[3:25] sudden if you get two inches of rain,
[3:27] you can pretty much throw out the
[3:29] playbook that you had in there because
[3:30] the rain changes the conditions of the
[3:32] market, especially financials and credit
[3:36] changes the game plan for this year. So,
[3:37] as traders and as people trying to make
[3:39] money, you can still focus on the names
[3:41] that are going to do well related to AI.
[3:43] Uh there's going to be a lot more
[3:45] choppiness. You're going to have to
[3:46] trade more this year. So again, AI is
[3:50] disrupting everything. And I just put
[3:52] this up there because this week was
[3:53] another week of all of the AI
[3:56] disruptions. If you don't think Iran is
[3:58] related to AI, both through the
[3:59] anthropic fight beforehand and how much
[4:01] artificial intelligence is being used or
[4:04] because again Iran, Venezuela, Cuba,
[4:09] even with Greenland, the first three
[4:11] proxies for China, we have a meeting
[4:13] coming up with shei. Energy is obviously
[4:15] a major part of this. you just get into
[4:18] a situation where regardless of whether
[4:20] it's the disruption for software, the
[4:23] labor, we had a negative job print,
[4:24] we'll go through that, the government
[4:26] trying to get more involved with AI, the
[4:30] speed of everything where I'm making the
[4:32] argument that debt verse equity are
[4:34] converging to one because we're
[4:36] shrinking the time period. You don't
[4:37] know what three years are going to look
[4:38] like. So what's the difference between
[4:40] debt and equity and the capital
[4:41] structure if you can't save with
[4:43] certainty if business is going to
[4:44] survive in three years? And then the
[4:46] commodity side, the need for minerals,
[4:48] the need for controlling this after the
[4:50] rare earth uh card was played last year.
[4:53] I think this is all relevant. Um my
[4:56] turbulence model, this is a new
[4:57] turbulence model for those of you
[4:59] subscribers and people at 22V, you got
[5:01] this. Um I'm just highlighting it again
[5:04] for people, particularly for people on
[5:06] the uh RA and FA world where we've been
[5:08] getting a lot more interest. My job this
[5:11] year is not only to teach you AI and
[5:12] help you um be able to service your
[5:14] clients better by using some of the
[5:16] tools that I use, but also by giving you
[5:18] the types of models that I think are
[5:20] going to be very important in a world of
[5:21] deleveraging. So you understand the best
[5:23] way for me to describe what turbulence
[5:25] model is. It is a way to measure the
[5:28] P&L, not the actual up or down, but the
[5:32] daily P&L change that's going on for
[5:34] anything that is a levered
[5:37] uh say model, quant strategy, hedge
[5:40] fund, that's a multistrat, anything
[5:42] along those lines that has a an element
[5:45] of cross asset in it. But even really,
[5:47] as we're going to see this week, what
[5:49] happened with momentum, it just means
[5:51] the daily V is going higher. And what I
[5:53] want you to look at uh on the bigger
[5:56] piece, this is the environment we've now
[5:58] been in and the market hasn't yet
[6:00] fallen. So, we're starting to get it.
[6:03] You look at the times where we had
[6:04] clusters like this one and I this is a
[6:06] second model that I built this week.
[6:08] This week is meant to be more
[6:10] representative of the current situation.
[6:13] So, the past one that I've shown you
[6:15] guys uh use a 250 day 252-day look back
[6:19] or one year worth of trading days. This
[6:21] one is exponentially weighting it
[6:23] towards the more near-term just to get
[6:25] signals that are more representative of
[6:27] the volatility. And you can see that
[6:28] most of the time when you get these
[6:30] clusters, it's after we've had a down
[6:31] move in the S&P. These are happening
[6:34] before we have it. That says to me this
[6:36] is going to be a deleveraging period for
[6:38] the entire year. And it comes when the
[6:40] prime brokerage numbers for uh for
[6:43] leverage from the OFR
[6:47] up at all-time highs. This is a
[6:49] multistrat um one and you can see the
[6:51] massive move since 23. This is from the
[6:55] Q3. So it doesn't show the current
[6:57] situation but with volatility or coar
[7:00] moving higher with correlations breaking
[7:03] down you will need less leverage to
[7:05] generate the same daily P&L which means
[7:08] leverage is going to come down. It is
[7:09] not a directional thing but it does mean
[7:12] that when you see this happening which
[7:14] historically happens during deleveraging
[7:15] cycles is the VIX goes higher. We closed
[7:18] at 29 almost 30. I mentioned uh on the
[7:21] webinar and I think last week that I
[7:24] think for certain we'll get to 30. Well,
[7:26] now we've gotten there. The question is
[7:28] with this kind of situation. We've got a
[7:29] lot of oversold signals where we bounce
[7:31] back down. But you see this upward
[7:33] trending move which to me is the way
[7:36] that the move index looked in rates V. I
[7:38] think because of the liquidity needs
[7:40] that are going to keep coming through
[7:41] private credit and the spreading of this
[7:44] which is showing up in all the credit
[7:45] markets that the VIX will be the liquid
[7:48] part and I do think there is a high high
[7:50] risk if I'm my scenario is correct of us
[7:52] getting up into these levels before um
[7:55] the stuff that I put in my portfolio I
[7:57] sold a little bit on Friday but my goal
[7:59] is really uh to hedge things into 50
[8:02] these things were put on to hedge the
[8:03] commodity the um semiconductor and
[8:06] things like that it has worked Well, I
[8:08] want to be in a position to add. This is
[8:10] going to be a critical thing for me and
[8:11] I'm going to talk about this in terms of
[8:13] the conditions side. So, think of this
[8:14] chart as representative of [sighs]
[8:18] the sloppy conditions for a racetrack.
[8:20] The two inches of rain. You have to
[8:21] think and throw the playbook out. This
[8:23] is the financial sector. It continues to
[8:26] be weak. It is below the 200 day moving
[8:28] average. And now the 200 day moving
[8:30] average is pointed downward.
[8:32] Never a good sign for the market. Uh
[8:35] people keep focusing too much attention
[8:37] on this. This is software relative to
[8:39] the S&P. Only thing I want you to look
[8:41] at is just how violent this fall is and
[8:44] how this move was so
[8:47] smooth, nothing going on. And that's
[8:49] where software was. My guess is best
[8:52] case scenario is we go sideways for a
[8:54] long time. Uh I don't think we're going
[8:56] to collapse from here because I think
[8:57] we've already done enough damage. But
[8:59] this fall for a sector that is arguably
[9:02] the most important in the world and is
[9:04] heavily dominated by the US it is
[9:06] leading to some of the credit issues as
[9:09] we've seen. Now the big news this week
[9:10] oil gets above 80. There is no way to
[9:14] ignore this. Too many people uh are I
[9:16] think have been conditioned
[9:18] especially on the sell side to believe
[9:20] that these things uh always turn around
[9:23] and that's because of taco. uh they just
[9:25] have this belief that Trump is going to
[9:27] not let the economy get hurt. This one's
[9:30] obviously a little bit more complex uh
[9:32] and it does involve the Straits of
[9:34] Hormuse and unfortunately we've reached
[9:36] a point where as the oil got above 90
[9:39] for WTI uh you've reached a point where
[9:42] the straight of Hormuse is definitely an
[9:44] issue that has become a bigger issue. I
[9:46] you know this may end at some point but
[9:48] the question is how much damage has
[9:50] already been done? Uh we're losing 450
[9:54] million barrels per month
[9:57] with no ships. This is not just oil.
[10:00] This goes into smelting, refined metals,
[10:03] coals, chemicals, fertilizer.
[10:05] They will all go up in price. This is
[10:07] pretty similar to 2022 as I go through
[10:09] some of the actual numbers. Like gas at
[10:11] the pump is now up 33 cents.
[10:15] Here are the weekly moves going back to
[10:18] 2004. This is 22 years. Just remember
[10:21] what happened to inflation and what
[10:23] happened to the market in these
[10:25] conditions. So we've reached a point
[10:26] already where this is not good for the
[10:29] economy. And remember whenever I talk
[10:31] about things, the thing that would make
[10:33] the S&P trade poorly is if something
[10:36] happened to change the certainty on
[10:38] earnings and to change the certainty on
[10:41] the economy. Well, this does both. Um
[10:44] we're going to have inflation prints
[10:46] which are extremely high just based on
[10:48] the movements that we've seen. And
[10:49] normally the way that I look at oil
[10:51] prices or gas prices, I look for the CPI
[10:54] for March uh to end some middle of the
[10:57] month. So I always look from like
[10:58] February midmon to March midmon. So when
[11:02] we get these numbers um in April, you're
[11:05] going to have a very high CPI print on
[11:07] the headline number. Whether it goes
[11:08] through the core,
[11:10] not that important, but here are the
[11:12] CPIs relative to gas at the pump.
[11:16] Um,
[11:18] so you you've got that issue in terms of
[11:20] of all this. I forgot. Did I include
[11:22] something? Oh, yeah. I just wanted to
[11:24] show this. I didn't I didn't speak to
[11:25] it. So, this is using the Arbob futures
[11:28] versus gas. And so, currently, we're
[11:30] actually got another, you know, 40 cents
[11:34] based on where we were just based on
[11:36] this. If this comes through, we may not
[11:38] get there because futures could turn
[11:39] back down. We could get oil to fall back
[11:41] down. You know, something could happen.
[11:43] But the reality is gas at the pump is
[11:46] not done yet. Uh this is a signal that
[11:48] China doesn't expect this to end anytime
[11:50] soon. They halt diesel gasoline exports.
[11:54] And this is where I just want to bring
[11:56] the attention back because this is the
[11:57] most important thing. This is again this
[11:59] week. Blackstone hit with a wave of
[12:02] redemptions. Uh investors pull a net 1.7
[12:05] billion out. There was even more than
[12:08] that that would came out. they had some
[12:10] inflows somehow um as well. So net went
[12:14] out, the partners had to put some money
[12:16] up to keep it so they didn't have to
[12:18] truly shut it the way Blue did. And then
[12:21] later
[12:23] Black Rockck limits redemptions. This
[12:26] came out on Thursday that they had a
[12:30] a loan that had been marked just three
[12:33] months earlier at 100 cents on the
[12:35] dollar and moving it to zero. These are
[12:38] just not good signs in terms of the
[12:40] whole thing and it's not over yet. And
[12:42] in my career, I cannot remember one time
[12:45] that a credit issue of this magnitude
[12:48] where it's been going on now for quite
[12:50] some time. I mean, the first wave of it
[12:52] was in October after Tricolor and First
[12:54] Brands. But if you really go through it
[12:56] and you think about everything that's
[12:58] happened and as I get through this, it's
[13:00] starting to get into the data center
[13:01] side into some of the decisions being
[13:03] made on the funding side. I didn't
[13:05] remember Blue Owl is heavily involved in
[13:07] the data center funding side. Um,
[13:10] interval funds are also another place
[13:12] and this was a story that came around
[13:15] about an interval fund. Uh, and if you
[13:18] haven't read it, I would go read it and
[13:20] just go through uh Cliff Water Corporate
[13:23] Lending CLFX.
[13:25] You can go look it up yourself. It's
[13:27] about a 32.5 billion one.
[13:31] uh it was part of a private wealth fund
[13:34] fundraising bonanza and again a lot of
[13:36] this happened over the last 5 years and
[13:37] you can see that 24% of that fund is in
[13:40] information technology. The problem is
[13:43] when you're in a position where there's
[13:44] not much trading in those and you have
[13:46] to sell something for redemptions you
[13:47] typically sell the other ones or the
[13:49] ones where you can get the cash in if
[13:50] there's no liquidity in the others.
[13:53] And the problem is you're starting to
[13:55] see more and more of these cracks. And
[13:56] if you sell the other ones, that's what
[13:58] takes the credit down wider. And then
[13:59] you have people buying hedges. And
[14:02] that's what we're starting to see. You
[14:03] had Mark Rowan from Apollo who probably
[14:05] has a pretty good idea of this. They
[14:07] must be in a decent position because
[14:08] he's saying private cracks are widening.
[14:12] Lloyd Blankfine said it's due for a
[14:13] reckoning. I watched the p or I listened
[14:16] to the podcast. Uh I didn't use any of
[14:18] the quotes from it, but um private
[14:20] credit is definitely an issue. And his
[14:22] point was when you go this long without
[14:24] an issue, you're probably going to have
[14:25] one. And remember we did have an issue
[14:28] which was Silicon Valley Bank and I keep
[14:30] saying to people especially on
[14:32] conversations uh that Silicon Valley
[14:36] Bank if we learn one lesson in the
[14:37] digital world of social media world if
[14:39] something starts spreading very quickly
[14:41] then everyone tries to get their money
[14:42] out at the same time. It's not like
[14:44] you're losing a lot of money if you try
[14:45] to get your money out of private credit
[14:47] at this point. Yes, it was giving you uh
[14:50] returns above, but clearly the risk that
[14:52] you're taking going forward with the
[14:53] tech names there, I think this is
[14:56] spreading quickly and so PIMCO said
[14:59] should face full-blown default cycle.
[15:02] If you want to go learn more about how
[15:04] this extends even further from private
[15:06] credit and it starts to get into how did
[15:08] this actually happen that retail was
[15:10] able to buy it. Steve Eisman who um is
[15:13] obviously famous from uh from his
[15:16] financial work and he's a very smart
[15:18] financial guy. He has the Steve the real
[15:20] Eisman playbook. He did talk about it
[15:22] this week and he had on in particular a
[15:26] person who knows the insurance industry
[15:28] real well and that's where this ends up
[15:29] guys. So, this all extends in the
[15:32] connection between private credit,
[15:34] insurance, and the retail connection.
[15:36] These are all the things in it. I'm not
[15:37] going to go read them. I highly
[15:39] recommend listening to the about 30
[15:41] minute podcast uh that was there. Uh if
[15:44] you guys don't know where it is or you
[15:46] can't find it, uh as is for all the
[15:48] subscribers and 22V people, there is one
[15:50] uh one that goes up on the website that
[15:52] has a link and a summary of all of those
[15:55] podcasts. And then Goldman uh was trying
[15:59] to get this uh this buyout loan for
[16:01] DuPont done. We've also got the same
[16:04] thing with um EA, which is the largest
[16:07] buyout uh maybe in history, but
[16:11] regardless, they're just having trouble
[16:12] getting this stuff to the market. Now,
[16:17] um I listen to Ryan Dietrich every week.
[16:19] Not picking on him, but I do think we've
[16:22] reached a point where you have to make a
[16:24] decision in your head. Again, if we're
[16:26] in a different playbook than we've been
[16:28] in the past, the there's no doubt in my
[16:31] mind that right now um sentiment is on
[16:33] the negative side. We haven't seen in
[16:35] some indicators. I'm going to show you
[16:37] some things that I've built. Uh, but I
[16:40] think people trying to pick the bottom
[16:42] like Scott Rubner and just saying with
[16:45] oil moving higher and no certainty on it
[16:48] with the credit situation and with AI
[16:50] going in an incredibly fast pace which
[16:53] is where my focus is. I believe the a
[16:55] disrup AI disruption is causing all of
[16:57] this and again unless people are using
[16:59] AI all day long and they're on OpenClaw
[17:01] and they have some knowledge of how much
[17:03] the stuff is progressing more than just
[17:05] a chatbot. I think you're living in a
[17:07] world right now, which is why I'm trying
[17:09] to help people, but in particular
[17:11] spending more time, not just with the
[17:12] institutional clients at 22V on what's
[17:14] going on, but also spending time with
[17:18] advisors. Um, I think this is going to
[17:20] be a very complicated time for people,
[17:22] and everyone seems to be looking for
[17:23] when the bottom of this is. Now, we're
[17:25] coming out of one of the tightest
[17:27] ranges,
[17:28] especially for this point of the year.
[17:30] So, this is specifically taking the
[17:32] tight range we've been in, which I
[17:34] showed on the webinar with the Ballinger
[17:35] band for the S&P, but this is it for the
[17:38] beginning of the year. And normally,
[17:40] when you get tight ranges, you get a
[17:42] break. And I just don't see this being
[17:44] something that is going to break higher
[17:46] with the oil situation, with the AI
[17:48] situation, with the credit situation. I
[17:50] don't see all three of those going away.
[17:53] So, even if the oil thing were to go
[17:54] back and we were at $65 oil, I think
[17:57] we'll go back to the highs. But the
[17:58] question is, will financials be able to
[18:00] get out of here? And will the credit
[18:02] unwind that's been happening just stop?
[18:04] I don't believe any of those are the
[18:06] most likely scenarios. So again, I put
[18:08] this out. Um, I just wanted to make sure
[18:10] that people saw, but when my turbulence
[18:12] model is telling me that there's
[18:14] deleveraging happening, at the same time
[18:16] that financials are below the 200 day
[18:18] moving average and the 200 day moving
[18:20] average turned down for the first time
[18:22] since 2023.
[18:25] Like I said, the conditions of the
[18:26] market have changed. Um this is
[18:28] something that I will start uh putting
[18:31] up also for the subscribers. It's going
[18:33] to be a combination of just a technical
[18:36] side of things to make sure that you
[18:38] have kind of a playbook of the how
[18:40] things are trending. These are all the
[18:41] different indices right now. You can see
[18:43] how everything right now is in a
[18:44] downtrend. When you get a breath of
[18:46] things that are in a downtrend,
[18:48] particularly since they were all up in a
[18:51] uh uptrend and when you look at the
[18:53] weeklies, we're losing most of the buy
[18:55] signals. We still have one for EWZ. We
[18:57] still have one for TLT. We're neutral on
[18:59] it's neutral on EM. This is not John
[19:02] Ro's work. Whenever I'm doing the
[19:04] thematics is when I'll get John's when I
[19:06] put out a new piece. But for a lot of
[19:08] the names that I've recommended to
[19:09] people, I do think it's important to
[19:11] stay on top of the technical picture on
[19:13] those. Uh Palanteer Act great this week.
[19:15] I'm going to write a paper this week uh
[19:18] on Marll for subscribers along with uh
[19:20] at least one other name that I think
[19:22] fits into kind of trying to find names
[19:24] that maybe are going to be part of the
[19:26] thematic picture that I've been talking
[19:28] about. But that'll be going on. Then the
[19:30] other thing I'm I I've kind of finished
[19:32] up on and I just wanted to highlight
[19:33] this. So I did want to have an oversold
[19:36] indicator that basically takes into
[19:37] account the way that I look at oversold
[19:40] and overbought. I want to see something
[19:42] on the breath side. We have not had a
[19:44] panic in breath yet. So, we haven't
[19:46] really had any big negative days in
[19:48] breath. And I think when we finally do
[19:50] make a bottom, you're going to see that.
[19:52] So, what this thing is is this is a
[19:54] model thing. When it actually gets to an
[19:56] oversold condition, you're going to get
[19:58] these dots. Uh you can see that like the
[20:01] last one we got was here. Here. They
[20:03] generally happen as kind of an ending
[20:05] point. We got a lot of them during
[20:07] liberation day. Uh so you know a lot of
[20:11] signals in there which meant we got
[20:13] oversold and we sat there. It uses the
[20:15] VIX. We closed outside the Ballinger
[20:17] band. So we have a uh oversold on the
[20:19] VIX. We have an oversold on the VIX term
[20:21] structure meaning front month fall is
[20:22] above three threemon out the demar
[20:25] signal not there yet from an exhaustion
[20:28] basis. It's the breath and the RSI of
[20:30] the breath that is just not there at
[20:32] this point. Uh so we'll get there. Uh
[20:36] we're not there yet but I just wanted to
[20:37] bring that up. And this is just to
[20:40] highlight to you what has happened
[20:42] basically on the muddy track analogy
[20:44] when the financial this is the financial
[20:47] 200 day moving average. It just turned
[20:49] down here. So that's where this line is
[20:52] turned down here. These are the S&P draw
[20:56] downs. Every time we've had a draw down
[20:59] in the S&P of at least 15% or at least
[21:02] 10% let's say but eventually 15. And
[21:05] these were all periods that in 15 this
[21:08] was related to oil which was a credit
[21:11] event. And eventually you had the US
[21:13] Shanghai accord that happened here.
[21:16] That's what had the market go higher.
[21:17] Here is the 18 period. This low here
[21:22] where it ended that was when the Fed
[21:24] pivoted. Meaning they were they had just
[21:26] started to go through the point of
[21:27] tightening and then they stopped right
[21:29] after that. Uh that was after a again a
[21:32] 20% correction in here. COVID, we
[21:34] obviously got the massive stimulus and
[21:36] then in 22 we didn't bottom until the
[21:38] Fed pivoted and said they were done with
[21:40] rates. So again, I think you're in a
[21:42] situation where normally when you get
[21:44] the financial sector turning down, think
[21:47] of it as financial conditions.
[21:50] Financial conditions are bad right now.
[21:53] You've got financials down below the 200
[21:55] day and the 200 day is pointing down,
[21:57] which means it's not just a short-term
[21:58] thing. It's been going on for a while.
[22:01] At the same time, you have credit
[22:03] showing tons of strain and you have the
[22:04] VIX trending higher. The S&P hasn't
[22:07] fallen yet, but I think the best case
[22:09] scenario in my mind is that we're kind
[22:11] of playing with a ceiling and we don't
[22:13] have a floor. We could bounce from here
[22:16] just because we're oversold like I
[22:17] showed on some of the indicators. But I
[22:19] still think you're going to find supply
[22:20] into this because I don't think the
[22:22] negative stories completely built in.
[22:24] Now, the S&P was only down 2% for the
[22:26] week. It was the worst week since
[22:28] November.
[22:30] NASDAQ barely down one and a quarter.
[22:33] IWM's down worse, worst week since
[22:35] August. Um, they've obviously had a
[22:37] tremendous amount of good weeks, but
[22:38] again, you started to run into
[22:41] momentum unwind. So, Bitcoin was up
[22:43] because Bitcoin had been down. So,
[22:45] everything now was kind of a screw you
[22:48] week. Uh, the VIX on a weekly basis was
[22:51] up 10 points aside from this period in
[22:55] 2025. It was the biggest since going
[22:57] into the rate hike cycle. So again, if
[23:00] you think, well, the VIX is up, that's a
[23:02] panic, just remember and go look at
[23:03] equities back here. That's why I'm
[23:05] saying if if financial conditions have
[23:07] changed, which right now with oil, which
[23:09] is an inflation part, credit, which is
[23:11] already widening, which has been driven
[23:12] by first software, but then also by a
[23:15] structural unwind of private credit
[23:17] where retail wants their money back and
[23:19] no liquidity in the market for the
[23:20] things that they're in. Financial
[23:22] conditions have changed. We've barely
[23:24] budged on CDX. So CDX for IG or sorry
[23:28] high yield has not even gone anywhere
[23:30] yet. Uh I don't think we're near the
[23:32] point of the levels that are going to be
[23:35] representative of people actually being
[23:36] scared on things. Uh I don't think we're
[23:38] there. The advanced decline line like I
[23:40] said breath. So this takes decliners uh
[23:44] uh relative to advancers. So I look for
[23:47] days where you're at least going to get
[23:48] up close to 10 to one. We haven't had a
[23:50] day yet 5 to one. We haven't had a big
[23:53] sell day. There just hasn't been a day
[23:55] where everything went down, which means
[23:57] hedge funds are doing well so far year
[23:59] to date. It might I'm sure they're
[24:00] losing uh gains, but they're still up
[24:02] for the year. Semis and all these trades
[24:04] that people had clearly moved into
[24:06] finally sold off and that's why you got
[24:08] the momentum trade unwind. So this is
[24:10] the week for momentum. It was the worst
[24:11] week since liberation week. But even
[24:14] before then, you got to go back all the
[24:17] way again to 2022. So when you start
[24:19] seeing weeks like this and remember you
[24:22] start seeing them that's not the bottom
[24:24] if if this is an unwind now I happen to
[24:26] think we are going to make a bottom down
[24:28] lower but for now the conditions have
[24:31] changed this is one month mo so this is
[24:34] basically trades that people have had on
[24:35] so this is the winner momentum created
[24:37] just for the last month this is the
[24:39] worst since covid
[24:42] technology so this is just techmo again
[24:46] worst since around the covid period
[24:49] And that's because of semis relative to
[24:52] IGV which now has given back about uh
[24:55] looks like about 40% of the gains year
[24:57] to date which was a massive number. So
[25:00] it's not anything structural and if this
[25:01] is the trade you had on yeah you're
[25:03] feeling pain over the last whatever
[25:04] days. It may not feel good but you're
[25:06] not down for the year. So you're still
[25:08] sitting there looking to play offense.
[25:10] The Cosby I mean
[25:13] this is the daily move and I don't want
[25:15] you to just look at this. I want you to
[25:17] look at this, too. I mean, we're talking
[25:20] big sizable moves that typically when
[25:23] this kind of stuff is happening, there's
[25:25] usually something some sort of
[25:27] deleveraging going on. Um, gas at the
[25:29] pump is one thing. I think this is a
[25:31] major major problem. Uh, this is jet
[25:33] fuel. Uh, this is a, you know, this gets
[25:37] into a incredible move that has just
[25:40] huge implications for travel costs. this
[25:44] needs to come back down, but this is
[25:46] even worse than anything we saw leading
[25:48] into the inflation side. So, just be
[25:50] very careful right now with following
[25:52] people that are saying nothing's going
[25:54] on. This is the gas at the pump move
[25:55] just to show you the chart. Uh inflation
[25:59] swaps, they've moved out, but really the
[26:02] one year and the two-year, which is what
[26:04] you have here, they're they're they're
[26:06] out to 290. They're still under 3%. It's
[26:09] not that big of a move. Um so they are
[26:11] building it in but not huge yet. We got
[26:14] a negative payroll number. Um in the
[26:16] same way that I show that health uh if
[26:18] you strip out health and uh I'm sorry
[26:21] yeah health and education almost all the
[26:24] numbers in this for the last year have
[26:26] been negative. In this case the reason
[26:27] we had a negative 92 is because we
[26:29] didn't we had a we had a minus 95 or
[26:33] minus 160 something relative to last
[26:35] month for health. If we don't get health
[26:37] jobs there's no jobs. So for everyone,
[26:40] again, I say this every week. If you're
[26:42] in the debate of AI is not impacting
[26:44] jobs, you're wrong. You're absolutely
[26:47] positively wrong. Is it going to destroy
[26:50] the economy? No. Is it having an impact?
[26:53] Absolutely positively. There is no way
[26:55] anyone can convince me that at this
[26:57] point we are not seeing it. So when you
[26:59] see these things of like, hey look,
[27:00] software engineer postings are up. The
[27:03] reality is just go back to here. There
[27:05] is no job creation ex healthcare.
[27:08] Healthcare is the one place if you go
[27:10] look at the the the clawed code where
[27:12] it's going to disrupt jobs. It leaves
[27:14] health care alone. This is an
[27:16] unprecedented time and it shows up on
[27:19] the diffusion side. So here's the
[27:21] three-month diffusion for the labor
[27:22] market. We got the payroll numbers.
[27:24] Three-month diffusion. This didn't just
[27:26] start. This has been going on since
[27:28] 2024. We literally have no hiring and
[27:32] that's the way to go. This isn't about
[27:34] firings. I don't think we're going to
[27:36] have an unemployment rate where Satrini
[27:37] put it. Um, but I do think hiring is
[27:40] going to remain muted and I think
[27:41] there's going to be a churn when people
[27:43] do get fired. If they get fired, they're
[27:45] probably going to go to a startup or go
[27:47] go to another place. They're going to
[27:48] get another job. There's going to be a
[27:50] lot of displacement, but because AI is
[27:53] not being adopted by everyone, we still
[27:55] have a labor shortage overall. So, I
[27:57] wrote a paper on this with investor
[27:59] implications. I think this is really
[28:01] important especially for macro people
[28:03] just to have a sense of AI and how it's
[28:05] disrupting the same way I went through
[28:07] and said that AI is causing all of these
[28:10] issues. I know most people don't get a
[28:12] chance to use it as much as I do by not
[28:14] using it all the time. You have to read
[28:16] something that converts it back into a
[28:17] macroeconomics. If you're reading
[28:19] Goldman Sachs and Morgan Stanley and all
[28:20] of this stuff and I've seen the work
[28:22] sent to me, I can't even read it
[28:24] anymore. Same thing goes for people on
[28:27] the sell side research for names. AI is
[28:30] changing so rapidly, guys. And now with
[28:32] OpenClaw, I don't see OpenClaw mentioned
[28:34] from anyone. We're going to get into
[28:36] that. It has a huge disruption to the
[28:39] economy. Here we are year-to date. So
[28:41] again, energy, staples, industrials,
[28:43] materials, utilities still lead. And
[28:46] again, you've got consumer discretionary
[28:47] tech underperforming. Most people to
[28:50] some degree have this trade on. I don't
[28:51] think people went into the year getting
[28:53] long a bunch of these. They may have
[28:54] gotten caught, but again, financials the
[28:57] worst performing group. Can't say this
[28:58] enough, guys. When the financials are
[29:00] down this much, financial conditions by
[29:02] definition are worsening.
[29:04] And if you want another example of it,
[29:07] here's the 15-year chart of financials
[29:11] relative to the S&P, the white line, and
[29:13] 10ear rates. So, for everyone playing
[29:15] for tenure rates to go higher because of
[29:17] the oil situation, we have a credit and
[29:20] a labor situation, which to me is the
[29:22] dominant thing, and oil is not going to
[29:25] help the situation. Oil is a temporary
[29:27] thing. it will be demand destruction and
[29:29] if it's demand destruction and inflation
[29:31] goes higher but we also see real GDP
[29:34] come down I I just think as people go
[29:37] through this and they realize that the
[29:38] credit market is already tightening and
[29:40] energy is not going to help that that
[29:41] this all relates back to what is going
[29:43] to happen to the economy so open AI
[29:47] funding
[29:50] just to put it in perspective they
[29:51] raised $110 billion this private market
[29:54] funding round is about four times larger
[29:56] than the biggest IPO.
[30:00] So you're just talking about a massive
[30:03] private raise and it surpassed the March
[30:05] one which was 40 billion. Then you got
[30:07] Enthropic, you got XAI. So all of this
[30:10] the AI spending
[30:12] it just continues. Now we got more
[30:15] information on
[30:17] the ARR for OpenAI 25 billion is the run
[30:20] rate. Okay. valuation 840 annualized for
[30:24] anthropic with just hyper speed. So
[30:27] they're growing much faster. They're
[30:28] catching up to open AAI and at this pace
[30:30] they're going to surpass them. Their
[30:31] valuation seems like a bargain compared
[30:33] to Open AI. I worry about this situation
[30:37] cash burn all the stuff in there. I
[30:41] think you have to start keeping all this
[30:42] in mind and I think you have to go back
[30:44] and look at some of the things which
[30:46] I'll get into later. Jensen Yuang, Open
[30:49] Claw is probably the single most
[30:50] important release of soft for probably
[30:52] ever. If you don't know what OpenClaw
[30:54] is, I'm assuming most of you haven't
[30:57] used it.
[30:59] [snorts] If you don't know what it is,
[31:01] when Jensen Yuang says it's now the
[31:03] single most downloaded open source
[31:05] software in history, and it took three
[31:06] weeks,
[31:08] you have to start paying attention. And
[31:10] if you don't think it's real,
[31:13] just I'm I'm just including these all of
[31:15] these meetups, viral meetups that are
[31:18] going Open Claw Builder Day in Toronto,
[31:21] London,
[31:24] New York,
[31:25] LA,
[31:28] and of course, 1500 people gathered
[31:31] within three days notice for an open
[31:33] claw meetup in Shenzhen.
[31:35] This one I'm not going to even bother
[31:36] showing. It's another one that just
[31:38] shows how big it is. This one, people
[31:40] are lining up in China to get help
[31:42] installing OpenClaw.
[31:45] Again, if you don't know what it is,
[31:47] this podcast I highly recommend going to
[31:49] listen to. It's not all on OpenClaw, but
[31:52] he goes through the macro implications
[31:54] of this for the tech side. He frames,
[31:57] this is uh Peter Murdoch, who's part of
[32:00] Insight's partner, so he's a VC person
[32:02] out there. Uh frames the current
[32:04] awesome.
[32:05] It's harmless out at sea, catastrophic
[32:07] at the beach. The big real wave isn't
[32:10] generic AI. It's autonomous agents
[32:12] arriving in multiple messy waves. This
[32:14] is open claw. Uh, bolt-on AI may by
[32:18] time. Think of Salesforce and agent
[32:20] force, but AI native thinking is the
[32:22] only durable path. I keep saying this to
[32:24] everyone. As someone who builds his own
[32:26] stuff, many of the things I showed you
[32:28] today, I just build them in clawed code
[32:30] myself. I used to have data scientists
[32:32] to do that. If I want to change it, I
[32:34] just go change it. You cannot do this
[32:37] with software. You cannot go into it
[32:39] into buying software off the shelf and
[32:42] recreate something that you have. You
[32:44] have to reach out to them. For those of
[32:46] you in Bloomberg who think Bloomberg is
[32:48] some magical tool, the reason I have to
[32:51] build all this stuff is because I can't
[32:52] do it in Bloomberg. They finally came
[32:54] out with their AI. I used it for one
[32:55] day. I can't use it anymore. A, it's too
[32:57] slow. And B, it's too stupid. So, I
[32:59] don't even really know at this point
[33:01] what to do with people who want to use
[33:02] this data and go through it. I'm sure
[33:04] things will get better over time, but it
[33:06] is very difficult to take a bolt-on AI
[33:08] to any software and actually get it
[33:10] there. You need to have your own data
[33:12] and then you need to use it. And if you
[33:13] need to use it on an Apple Pro MacBook,
[33:17] go do it that way. Uh, he claims many of
[33:19] the toolings that write code like cursor
[33:22] are obsolete. I only bring this up
[33:23] because he's saying the same thing I am,
[33:25] which is we're getting companies that
[33:26] come out, they get to massive valuations
[33:29] and then they stop. The next step is an
[33:31] orchestration layer that root roots
[33:35] house across multiple models. I think
[33:36] you have to go read my Palunteer piece.
[33:39] Uh he predicts orchestration will lift
[33:41] open source models and drive a shift
[33:42] toward cheaper, more tunable AS6 style
[33:44] chips. This is really important. I'll
[33:46] get into the reason why. Autonomous
[33:47] agents, they're choosing things
[33:49] probabilistically.
[33:51] Agents will be doing the choices,
[33:54] humans will be okaying the choices. Very
[33:57] different thing. And this is again my
[34:00] belief too. Systems of record like
[34:01] Salesforce won't melt overnight, but
[34:03] their value depends on adaptation. It
[34:05] also depends on growth. Uh and the buyer
[34:08] of software will shift from humans to
[34:10] agents that as employees. It's one of
[34:11] the reasons why again if there's a
[34:13] billion pieces of software and humans
[34:16] aren't the ones buying them, humans are
[34:18] lazy compared to agents. You have to
[34:20] think about that if you're building a
[34:22] company and you always want the best
[34:23] model, the newest model. I do this all
[34:26] the time. You just go out and find it.
[34:28] Uh, OpenClaw NanoClaw represents the
[34:30] breakthrough enabling autonomous agents.
[34:32] That's why this is so important. The
[34:34] first impact is not h is on hiring, not
[34:36] layoffs. We've talked about that. Small
[34:38] medium businesses will adopt agents
[34:40] first. This is more a stock startup uh
[34:42] conversation, but it's anything in
[34:44] there. I've already seen small
[34:46] businesses that have done this.
[34:47] Enterprise adoption will lag but
[34:49] eventually dominate.
[34:52] The lag, the question is how long will
[34:53] the lag be? And you have to think about
[34:55] this. It will lag. This is becoming more
[34:57] of a story. And I'm going to bring back
[34:59] why that matters. The adoption speed
[35:00] depends on how fast agent frameworks
[35:02] improve. Agents act more like employees
[35:05] than tools. This is the issue is that
[35:08] open claw cannot work in an agent. As
[35:10] someone who uses it, it makes mistakes.
[35:11] It does things you don't want it to do,
[35:13] which is why I run it in a sandbox, but
[35:15] it still does tons of work for me. It
[35:18] can build me 10 models overnight. I
[35:21] there's just no way that you can compare
[35:23] it. It's like having a mediocre employee
[35:26] at this point. And then at some point,
[35:27] it'll be having a very good employee.
[35:29] Maybe it'll never get to a great
[35:30] employee, but to get to a great employee
[35:32] is what the major companies need. So,
[35:35] they're going to be slower to adopt
[35:37] because Open Claw is not there yet. his
[35:40] argument it's the end of GPU for
[35:42] everything. It's very important. I wrote
[35:43] a paper on this which I'll go through.
[35:45] Uh cloud verse edge and all of this
[35:49] stuff that happens. There are major
[35:51] investment things that go there. The
[35:53] move away from GPUs and to AS6 is a
[35:56] major thing. A6 A6 A6 and again they can
[36:00] be tuned for specific infr tasks or
[36:03] models improving performance and energy
[36:05] efficiency. Open source models
[36:08] accelerate edge compute. This is the
[36:09] China story again. This is the way I'm
[36:12] running my stuff on Open Claw. And
[36:15] remember, the real bare case for this AI
[36:17] boom isn't a bubble recession. It's your
[36:19] iPhone. In three years, a bulked up
[36:21] iPhone will be able to run a prune
[36:22] version of a model. I think we're
[36:24] getting closer than most people realize
[36:26] to that time. I think he's going to be
[36:28] off in terms of the three years, and
[36:30] that means his bare case is coming
[36:32] faster.
[36:34] combine what Peter said, the guy on the
[36:36] podcast with the thematic paper I wrote.
[36:39] Does his view support this? And
[36:41] basically it goes through where it
[36:43] supports it. Uh agents drive hardware
[36:45] specialization. That is the thing again
[36:47] where you're getting into a different uh
[36:49] layer. AS6 and model on chip align with
[36:52] the memory centric architecture thesis.
[36:55] AI hardware evolving around model plus
[36:57] memory plus compute code design. And the
[37:00] reason that's important is that's what
[37:02] this whole thing is about. So again, for
[37:05] you subscribers and for the people
[37:07] there, the names on here, you should go
[37:09] through and you should start spending
[37:11] time. I will write a paper on one of
[37:13] them that came through. I want to
[37:15] highlight this again because I've shown
[37:16] you guys this before. This is the way
[37:18] that I go from a podcast to ideas. I
[37:21] will be releasing the final four videos
[37:24] this week for the subscribers on this to
[37:28] show you how to do this. It will include
[37:29] the prompts for each one. So, even if
[37:31] you don't want to spend the time truly
[37:33] going through this, I will give you
[37:35] basically instructions on how to do
[37:37] this, how to use the LLM. I highly
[37:39] recommend paying for all of them. But
[37:41] regardless of the point, even if you do
[37:43] it for one month, for those of you
[37:44] looking to figure out how to train your
[37:46] analysts and do things, and I've done
[37:48] these presentations before, I know that
[37:50] the majority of people are not using
[37:51] them to this manner. This will help you,
[37:53] and you will get the prompts as well.
[37:55] This way, you can go do it on your own.
[37:57] That's why from the subscriber basis,
[37:59] regardless of who you are, if you do
[38:01] this for mutual funds and you just want
[38:03] to go through and analyze which mutual
[38:05] funds are right, you can do this with
[38:06] anything.
[38:08] Um, and these are the prompts and the
[38:10] stuff just to give you an idea of the
[38:11] things that'll go through. So,
[38:14] um, I also remember for those of you who
[38:19] who reached out this week, we do upload
[38:21] all of the video the podcast that I'm
[38:24] about to go through. Uh, so this way all
[38:27] you have to do is go copy this from the
[38:28] website. You can go watch it if you
[38:30] want. I include a summary. I include the
[38:33] timestamps. That way it makes it easy
[38:34] for you. If you don't know where it is
[38:36] and you want to go read about the
[38:39] Moonshots podcast, you can just go in
[38:41] there and get it. So that way when I go
[38:43] through the summary, it's not in in this
[38:45] particular case. The Moonshots one
[38:46] really got into real one important thing
[38:49] for me. Um, you can read in here what
[38:51] it's getting in. It's completely
[38:53] changing the computing stack. The
[38:54] foundation models are the new
[38:55] computational base. AI is beginning to
[38:58] act like an operating system. The reason
[39:00] this becomes important is because this
[39:02] is what they're saying. And I think this
[39:04] becomes important for anyone trying to
[39:05] again understand the the the software
[39:07] side. I don't think you can say software
[39:10] is going to be built if you're not using
[39:12] AI. If you don't understand everything
[39:14] that's changing and you don't understand
[39:16] Andre Carpathy saying software 2.0 is
[39:19] completely different than this one.
[39:21] So this is the traditional one. If you
[39:24] go through the shift that are happening
[39:26] through agents, how agents change it and
[39:29] the autonomous systems in terms of what
[39:31] goes on, all of the software stuff, if
[39:34] you're just saying it's not killing
[39:36] this, it is killing this. It's
[39:38] completely changing the computing stack.
[39:40] The operating system is going to be
[39:42] completely different. So unless you can
[39:44] describe and go through this and
[39:45] understand it and go ahead take a photo
[39:48] of it, upload it into an LLM and ask it
[39:51] to describe to you and ask what it means
[39:52] for SAS companies. This is what all of
[39:55] the people in Silicon Valley are seeing
[39:56] and talking about now. So if you're
[39:58] trying to go out and pick software
[40:00] stocks that have been repriced, the
[40:02] reason they've been repriced is because
[40:03] this doesn't exist anymore and that's
[40:05] where they thrived. Now it's this. Some
[40:07] of them may survive this. Most of them
[40:09] will not. And so remember, Palunteer is
[40:12] built for this, not for this. So when
[40:16] you guys are questioning Palunteer and I
[40:18] saw someone put something out, whoa, how
[40:19] can you buy 50 times more? I
[40:23] you got to go look at the way the world
[40:24] is changing and not just make these
[40:26] grandiose binary statements. It's
[40:28] expensive. You can't buy it. Um
[40:30] consulting firms. So if you guys look at
[40:33] the uh IT services space, it's been
[40:36] destroyed with software. I think you can
[40:39] go I'd rather pick out some of the the
[40:41] names in there that are also technically
[40:43] in the software index the names like
[40:44] Asenture and thing and some of the
[40:46] companies in there and the reason is
[40:48] because the enterprises have to figure
[40:50] out how to adjust to this and to adjust
[40:52] to this it's fairly complex and their
[40:54] organization is incredibly complex. It
[40:56] was much easier to just go buy software
[40:58] products but to actually get your
[41:00] operating system to work you need to
[41:03] have consultants. So I think they're
[41:05] going to have a boom. That's what they
[41:06] talked about in in the near term the
[41:09] transformation could sign significantly
[41:11] increase demand for advisory services.
[41:13] So I happen to agree with that. I don't
[41:15] see how it's not going to go any way.
[41:16] And this was also brought up on this. If
[41:18] you if you don't listen to this every
[41:19] week, particularly if you're not in the
[41:21] finance business and you're trying to
[41:23] figure out how to use AI, aside from
[41:24] just subscribing to my stuff and getting
[41:26] my videos, listen to this podcast every
[41:28] week. I think these this this is the
[41:30] best one for people who are not looking
[41:32] to necessarily invest in it. I think
[41:35] there's investment ideas that come out
[41:36] too, but I think for using it, it's a
[41:38] great one. They covered the anthropic
[41:40] versus US defense department for about a
[41:42] half hour. I think it's important
[41:44] actually for 40 minutes, I guess. Went
[41:46] through the funding round and then uh
[41:48] they talked about an interview with with
[41:50] the Claude code person and how he's not
[41:53] done not done his own code at all since
[41:56] Opus 4.5 came out. uh but they also get
[41:59] into the enterprise point and why it's
[42:02] taking so long and how open AAI is now
[42:06] in a push to get inside big companies.
[42:09] So this is where the competition So
[42:11] Anthropic clearly has won the
[42:13] enterprise. OpenAI has been having
[42:14] trouble monetizing with consumer
[42:16] customers. So now they're making a big
[42:18] push to get inside companies. That does
[42:19] two things. One is it creates
[42:22] competition for anthropics. So they're
[42:24] fighting for the same pie much more so
[42:26] when it was apparent that they weren't
[42:28] fighting for it. But now they're
[42:29] partnering up and doing a whole bunch of
[42:31] things meaning Open AI. So this changes
[42:35] things. Frontier Alliances, the
[42:36] enterprise adoption machine, they
[42:37] highlight Mckenzie, BCG, Assenture as
[42:39] partners for OpenAI. So they're really
[42:42] trying to get the implementation because
[42:44] the models are are ready. It's the
[42:47] organizations that aren't there. And
[42:48] that's what we're talking about is the
[42:50] capability overhang. How do they get
[42:53] more people companies to use AI when the
[42:57] models are already at levels that are
[42:59] good enough? And so they're taking an
[43:01] entire approach
[43:03] to get involved with this. Um, in this
[43:06] case, they're trying to replace
[43:08] Salesforce and Workday. But again, this
[43:09] gets into that that image I showed you
[43:12] in terms of how the computer stack
[43:13] inside an organization is changing. It's
[43:15] so complex that it's very difficult.
[43:17] Remember, all I showed was what the new
[43:19] setup would look like. That does not
[43:21] solve them having all these data issues
[43:24] that they have to clean up. It is going
[43:25] to take a lot longer time in my opinion
[43:27] for the adoption at the enterprise level
[43:29] to have a meaningful impact on their
[43:30] margins especially while the cost of all
[43:32] this stuff is going dramatically higher
[43:34] in the short term. Uh again more open AI
[43:37] into their frontier stuff. Why
[43:39] enterprise AI ambition is slowing down
[43:40] in 2026. That cannot happen. This is
[43:43] where all of them need. They've made a
[43:45] huge bet. This is again from Dennis and
[43:47] the team at 22V. Great chart. I may show
[43:51] this a lot, but again, cash return as a
[43:54] percent of net income is at very low
[43:56] levels. They've made a huge bet. And we
[43:59] go back to this interview that Jim
[44:01] Chenos did. If you guys didn't see it,
[44:03] it's from months ago. I highlighted it
[44:05] uh before, but I bring it up again, and
[44:07] you can go get this in that video
[44:09] podcast info that I put in. Why Oracle
[44:12] is the fragile link. Oracle reports
[44:14] earnings this week. So does Adobe, but
[44:16] Oracle is the only one that I care about
[44:18] from an importance perspective. And the
[44:20] reason is when you start connecting all
[44:22] the dots. Um
[44:25] I told you that Jim Chenos was not super
[44:27] bearish on this, but he was very honest
[44:29] about the Oracle situation and the fact
[44:31] that they have negative free cash flow
[44:33] already and they're making a huge bet
[44:35] and they assume AI revenue inflection by
[44:37] 2027 2028. The problem is again you need
[44:41] the enterprises to adopt on this. If
[44:43] it's all being used by open-source China
[44:46] models that we're downloading onto
[44:48] Apple, Apple stuff is selling off the
[44:50] shelf. To me, that is an indication at
[44:52] this point that something bad is
[44:55] happening. You have startups using
[44:58] open-source.
[44:59] You don't have the enterprise is going
[45:02] at the pace that you would like for
[45:04] issues that they have.
[45:06] And here's the other part. Oracle's
[45:08] customers are part of the problem set.
[45:11] All this is fine until VC tightens, uh,
[45:16] credit tightens, sentiment turns, or the
[45:18] market starts funding losses. We're
[45:21] getting closer to that because of the
[45:22] credit issues. So, just remember
[45:24] Oracle's free cash cash flow yield and
[45:27] the trajectory, the bet they've taken.
[45:28] So, they're going to be reporting.
[45:30] Remember, this is already starting to
[45:32] show up in terms of how are we going to
[45:34] change the situation. So, as he tried to
[45:37] raise more money, clearly this became an
[45:39] issue. remember this Brad Gersonner
[45:42] and him basically walking off after Brad
[45:45] said, "How are you going to pay with
[45:46] only, you know, at the time 14 billion
[45:49] in revenues for 1.4 trillion in spending
[45:51] by 2030? How are you going to pay for
[45:53] it?" So he walked off stage. That was
[45:55] back in November, I believe. And then in
[45:57] December,
[45:59] you got Oracle and Blue Owl in there. At
[46:01] that point, Blue Owl was not a household
[46:03] name. Okay. Well, now they're a
[46:05] household name.
[46:07] OpenAI reset spending expectations. So
[46:09] they went from 1.4 trillion. He walks
[46:11] off the off the set with Brad Gersonner
[46:14] and now he's down to 600 billion. That
[46:16] clearly has something to do with the
[46:18] fund raise and what was going on.
[46:21] Oracle announced thousands of job cuts
[46:23] in face of an AI cash crunch and Oracle
[46:26] drops after report on Friday that it and
[46:29] OpenAI scrap plans to expand a data
[46:31] center site in Texas. Now, when you go
[46:34] through this, they're breaking the the
[46:37] the plan broke down after negotiations
[46:40] dragged over financing and OpenAI's
[46:42] changing needs. It's a sign that the
[46:44] financing demands of the AI boom are
[46:46] causing some of the biggest players to
[46:48] make some tough business decisions and
[46:49] pair their investment ambitions. I'm
[46:51] telling you, this gets important because
[46:53] in a lot of cases, they've had to
[46:55] pre-spend on on on other things. This
[46:58] gets into all the circular deals that
[46:59] we've seen. Again, I do not believe in
[47:02] any way, shape, or form that AI won't be
[47:04] built out. I also am not so sure that
[47:06] we're going to need all the data centers
[47:07] based on everything that's happening and
[47:09] the move that's going with open source
[47:10] and all the edge movements. But I do
[47:12] believe at this point that the market is
[47:14] not ready for any negative news on this.
[47:16] And oh, by the way, we're seeing the
[47:19] issue from the cash return on net income
[47:21] that I showed you from Dennis and the
[47:22] team because the buybacks continue to
[47:25] decelerate. This is another reason to
[47:27] just not be as
[47:30] [sighs] driven for these bouncebacks on
[47:32] things. You know, there's buybacks
[47:34] happening and so whenever the market
[47:35] gaps down in the morning, we seem to
[47:37] close near, you know, the middle to high
[47:39] of the day. And I think a lot of that
[47:41] has to do still with the buybacks which
[47:42] are buying on down days. But eventually
[47:44] you reach a point where the CTA starts
[47:45] selling. We are close to the first
[47:48] flipping sell signal uh with where we
[47:50] closed on Friday. So, if we do break
[47:52] down this week, uh I think you're going
[47:54] to start seeing a lot more supply that
[47:57] will offset the buybacks, but in
[47:58] particular, if it happens at the end of
[48:00] the day,
[48:02] uh the All-In podcast this week, uh the
[48:05] reason I'm bringing them up, they did
[48:06] talk about SAS, and I think there were,
[48:08] you know, there were there were
[48:10] important things in there, uh again, as
[48:12] to why spending your time on software is
[48:15] just a waste of time. I will say the
[48:17] same thing I say to clients when they
[48:19] call. There's a reason why Ford is the
[48:22] exact same price today as it was in
[48:23] 1990. SAS is not going out of business
[48:26] tomorrow. Ford has not gone out of
[48:29] business for 36 years. Yet, their stock
[48:31] price is unchanged. You can sit at the
[48:34] same stock price while your multiple
[48:35] compresses because the future is not
[48:38] yours. That is what I think will happen
[48:40] with a good amount of software. So, you
[48:43] have to make your decision. I do like
[48:45] the IT services consultant names for
[48:47] now. I do like the cyber names because I
[48:49] think there's going to be cyber attacks
[48:50] this year and they're still going to be
[48:52] necessary and they have to do the
[48:53] buildout. And I do like Palanteer. If
[48:55] you're looking for AI adoption, I think
[48:57] you want to focus on those names. But I
[48:59] think you also want to be wary that
[49:01] Anthropic and OpenAI and Palanteer are
[49:03] all competitors for the SAS model
[49:05] despite what everyone believes. Uh they
[49:08] are, even if it's just from a growth
[49:10] perspective. AI introduces structural
[49:12] uncertainty. SAS used to be easy to
[49:14] predict. It's not anymore. It goes from
[49:16] a growth annuity to tech technology risk
[49:18] asset. Um, that was from Chimath. And
[49:22] again, I think David Saxs put it
[49:24] perfectly. It was one of the easiest
[49:26] business models to value. And that's
[49:27] where I showed you where SAS relative to
[49:29] the S&P was like a linear increase. It
[49:32] never outperformed by a dramatic amount.
[49:34] It was kind of like a nominal GDP chart.
[49:36] Then you got into COVID and all of a
[49:39] sudden it went parabolic. And the reason
[49:40] was because people were sitting at home
[49:42] and it raised a ton of money in the VC
[49:44] world. valuations went through the roof.
[49:46] Well, now we're in the overhang period,
[49:48] not just in moving the valuations down,
[49:50] but now you have something that's
[49:51] changed, which is SAS is no longer an
[49:54] annuity with growth.
[49:56] So, it broke that certainty. So, the
[49:58] market compresses multiples, and I think
[49:59] that's a fair way to put everything. Um,
[50:02] it's not the end of the world. They're
[50:03] not going out of business tomorrow. And
[50:04] if you spend your time writing about it
[50:06] and taking a binary approach that this
[50:08] is too bearish, you have to buy them.
[50:09] This is too bullish, you have to do
[50:10] this. I just think at this point they're
[50:12] dead money. and if you want to get
[50:13] involved with them, I think you want to
[50:14] pick the companies that you think are
[50:16] going to thrive over the course of the
[50:17] rest of this year. And I gave you kind
[50:19] of my my my belief on those. Um, he also
[50:22] said that something important that you
[50:26] have to deal with that as the year goes
[50:27] on, if they haven't come back, a lot of
[50:29] these companies have been based on
[50:31] stockbased compensation. It'll be very
[50:33] difficult to retain employees. Will they
[50:36] lose the drain? And unfortunately, the
[50:38] software business is not like an auto
[50:40] business.
[50:41] uh they really don't have any assets uh
[50:44] except their brand and the technology
[50:46] they have in there. So if people start
[50:48] leaving because they can't attract
[50:49] talent, what happens to those types of
[50:50] places? Uh all right, let's just go
[50:52] through now a few things that were in
[50:54] there this week uh from people that are
[50:56] respected. Uh Kevin Rose basically said
[51:00] the traditional mode of we have better
[51:02] better engineers disappears because of
[51:04] the coding agents and again the coding
[51:06] agents get into the open claw
[51:07] disruption. Josh Wolf, uh, very smart
[51:10] guy. He basically shared this publicly,
[51:13] sent out this memo to all of the, uh,
[51:16] the family founders yesterday to be
[51:18] ready [sighs]
[51:19] [gasps] for difficult times. Um, and
[51:22] again, he said this isn't a macro call.
[51:24] It's a set of observations about
[51:25] correlated risks that are worth your
[51:27] attention. Got a lot of press because
[51:28] he's a respected person, but he just
[51:31] said you have to prepare for market
[51:33] risks. And I think that includes the
[51:34] macro situation in my opinion. It also
[51:36] includes this. And this is what I want
[51:38] to just make sure you realize. Here is
[51:40] the software market cap in the S&P 1500.
[51:43] It is at $5.4 trillion. Energy is 2.4
[51:48] and materials is 1.5. The combination of
[51:51] these two should be bigger than this by
[51:53] the time this is done. And this is the
[51:54] reason why I want to be long these and I
[51:56] want to still be short these as a trade.
[51:59] Uh it's run its course for the time
[52:01] being. So you get mean reversion. But
[52:03] remember these this is the the way it
[52:05] looked before software started and I
[52:08] think we're going to get back to that
[52:11] point because we're in a commodity super
[52:13] cycle. Uh and that's because you can go
[52:16] back and listen to this if you want to
[52:18] go back as to uh that podcast. So just
[52:21] to finish this up uh I said it last
[52:23] week, I'm going to continue to say it.
[52:25] You must pay attention to all of this
[52:27] stuff going on with Iran because this
[52:29] just adds another layer for the
[52:31] midterms. AI is a part of every single
[52:35] thing of the midterms. Whether it's the
[52:37] job situation, whether it's the war
[52:40] situation, whether it's the fact that
[52:42] these things are such uh dangerous tools
[52:46] than their nuclear weapons should these
[52:48] big companies actually be able to make
[52:51] decisions with them. It's not just a
[52:52] Republican thing. It'll be a Democrat
[52:54] thing. There's no way that AI will be
[52:56] allowed to run when the risks around it
[52:58] from people like Elon Musk who is 25%
[53:00] that it actually destroys everything. Um
[53:04] AI is the companies here represent a
[53:07] huge part of the market cap. The
[53:08] hyperscalers in all of them. So I think
[53:11] this is an issue where you have to go
[53:13] through and realize that everywhere on
[53:15] the tech side the more this gets between
[53:19] the Pentagon and the AI. What happened
[53:20] with Anthropic? This the Pentagon saying
[53:23] it told anthropic the firm is supply
[53:25] chain risk and how they're going to deal
[53:27] with it.
[53:28] The AI enabled weapons have been used in
[53:31] Iran.
[53:33] It's scrambling the politics of AI.
[53:36] It was used in the Maduro raid.
[53:40] It was used in Mexico. [laughter]
[53:43] And I wanted to highlight the fact that
[53:46] Mexican the cartel has drones and AI
[53:50] autonomous drones. The AI sovereignty
[53:53] paradox at home counter foreign
[53:54] relations is a sovereign power truly
[53:57] sovereign power if a private firm can
[53:58] constrain the ability to use what it
[54:00] could be decisive military technology.
[54:02] Um I'm telling you if you're long these
[54:05] companies in size which all of you are
[54:07] and this is the hyperscalers. This gets
[54:09] into all of them. The government is
[54:11] getting more involved and I think will
[54:12] continue to get more involved in AI. I
[54:15] don't know how these things keep their
[54:16] multiples at where they are over the
[54:18] course of the next few years if this is
[54:19] becoming more of a political issue. It
[54:22] is the exact reason why it is the most
[54:24] bullish case for Bitcoin. If the economy
[54:26] is growing and you're having a hard time
[54:28] buying growth assets,
[54:30] these things will be an issue.
[54:33] Uh again, more just anthropic with the
[54:36] military feud. Um open AI workers. So
[54:40] basically what you ended up with is
[54:41] OpenAI decided to do what Anthropic
[54:44] wouldn't. Anthropic got a tremendous
[54:46] amount of uh demand over the weekend.
[54:48] Open AI saw some people cancel. You've
[54:50] got the workers uh in argument about it.
[54:53] And again you've got the data centers uh
[54:56] which are getting delayed across the
[54:57] country uh and slowed down. And I think
[55:00] all of these lead to the same thing
[55:02] which is there's a massive bet on uh AI.
[55:05] The progress is going faster than the
[55:07] adoption is going. The progress going
[55:10] this fast is being mainly used now for
[55:13] AI agents. AI agents allow companies to
[55:16] grow rapidly on the startup side. And
[55:17] that's where most of this is happening.
[55:19] It disrupts any companies at this point
[55:21] that are both not able to use them. And
[55:24] then the bolt-on approach to just take
[55:26] AI and connect it to something. Well,
[55:27] those costs are going higher and higher.
[55:29] And part of it is because the data
[55:31] centers aren't being built enough to
[55:32] keep up with the capacity. And now we're
[55:34] getting into the Oracle side where the
[55:35] credit is starting to impact the space.
[55:38] At a minimum, things are getting more
[55:40] confusing, and I'll keep saying it. The
[55:42] financial sector is something you need
[55:45] to pay attention to. Financial
[55:46] conditions are a lagging indicator. They
[55:48] fall when stocks fall and when credit
[55:50] spreads rise. They are tightening every
[55:52] day and financials are telling you
[55:54] they're going to tighten more. I'm going
[55:56] to help you guys uh get through this war
[55:59] and uh create products and I'll probably
[56:01] do uh more webinars and maybe one this
[56:04] week depending on if we get a day where
[56:06] it looks like maybe a short-term bottom
[56:08] has come in or if we start seeing any
[56:10] kind of change. But for now, hunker
[56:12] down, have a trading mentality, and I'll
[56:14] see you guys next