YT YouTube2Text

YouTube2Text → Transcripts → The New Market Structure: AI Concentration and the Largest Hedge Fund Degrossing in 10 Years

The New Market Structure: AI Concentration and the Largest Hedge Fund Degrossing in 10 Years — Transcript

by Jordi Visser · 13,005 words · 1,833 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:02Saturday, Halloween day, which seems
0:04kind of apppropo giving the fact that uh
0:08it's a scary month, scary week, and
0:12there are more
0:14fantasy ghost stories that continue to
0:16be told every single day about how the
0:18world's going to end and AI bubbles are
0:20unwinding and everything. So, uh this is
0:23going to be a lot of slides. Uh I'm
0:25going to do this uh in a fast pace again
0:28like I did last week. Uh there's some
0:30other things I've added in here, but uh
0:32I did want to make just the one comment.
0:34I'm not going to go through these
0:35things. There's a lot to cover in here.
0:37Uh you know, just from the perspective
0:40of someone watching this who used to be
0:42on the other side looking for
0:43information like this. I covered so much
0:46in this week uh across so many different
0:49things that happened. It's almost
0:50impossible to remember everything that
0:52happened in the week. But if you can get
0:55all this from any one source, send it to
0:58me because I couldn't believe it when I
1:00did it. And this is one of the reasons
1:01why normally I try to do this on on
1:03Friday, but there was so much this week
1:05to go through. And I tried to give you
1:09all the data in a variety of different
1:11ways to tell the story, including the
1:13stuff that's been bad, the stuff that's
1:15been not so bad, and the things that I
1:18think are the most important to look at.
1:19So that's why I always give you this for
1:22the disclosure. I'm sure you guys have
1:23been through this plenty of times. Um,
1:25first of all, very importantly for th
1:28for the subscribers, I've now started to
1:31uh add the video slides. So, there's 160
1:34slides this week. Last week, there was
1:36about 145. Um, I highly recommend after
1:40you watch the video, download the slides
1:44and then go and this will be uploaded
1:47this week in the prompt section. So,
1:49this is what I added basically because
1:52I've gotten a lot of things. I know
1:53there's a lot of people that are purely
1:55crypto people. There's a lot of people
1:57uh that are younger at colleges and
1:58particularly, you know, parents have
2:00tried to get their kids up on AI and up
2:02on the markets, but there's also a lot
2:04of graduates that are trying to stay on
2:06top of what's happening in the markets
2:07and learn about the new market. Uh I'll
2:10say that again. If you listen to people
2:12my age and older, the market structure
2:14has completely changed. Uh there are
2:16things happening that have never
2:17happened before. We're going through a
2:18massive change in the markets and AI
2:20will only make this worse as humanoids
2:22and everything else come in. So trying
2:23to look at the past is completely
2:25useless to get a picture as to what's
2:27going to go on going forward and you
2:28will end up in situations where when you
2:30listen to bubbles by definition you are
2:33going to miss the home run pitches and
2:34I'll take you through that as we go
2:35through the market structure part. Um,
2:37but I uploaded this prompt and this is
2:40really going to be useful for RAAS, for
2:43FAS, for people who maybe are more
2:45focused on equities but don't understand
2:47that I'm doing here. So, this is an
2:50example of what the prompt will do.
2:51It'll give you an executive summary of
2:53what I talked about on the video. It'll,
2:56and you can see in there, it'll go
2:57through the slides that are impacted
2:59within there. uh the 10 key items that
3:02you should look at, how to go back to
3:04it, and then at the end you'll have
3:07this. If there's a particular topic that
3:09you want to learn more about, all you
3:10have to do is go press a button. So all
3:12you have to do is upload download the
3:13slides, upload them into an LLM, and
3:17then copy paste the prompt in there and
3:19let it go on its own. And that way
3:21whatever I've covered in here, you'll be
3:24able to go through it in there. So this
3:26is not the transcript. This is using it
3:27on the slides. So, as I said, uh the
3:30entire stock market looked like a house
3:32of Halloween horrors on Friday, driven
3:34by rapidly sinking tech stocks, swelling
3:36AI bubble. Okay, this is a fact. I I
3:41mean, I don't know how to go through
3:43this any more differently to start off a
3:45video than for this week, the S&P was
3:48down 60 basis points, which was less
3:50than the week before. For the month, we
3:53finished down 93 basis points. 93. You
3:57would think based on everything you've
3:58heard, everything that's gone on, that
4:00we were down a ton more.
4:02Nine of the 11 sectors for the month
4:04were up in the 93 basis points.
4:08300
4:10and
4:1327 of the 500 names, so I I mean I don't
4:18literally 60 plus% were up at least 2%
4:21for the month.
4:23This is what the breath looks like for
4:25the S&P 500. It ripped higher in the
4:28month. I can't say this enough. If
4:31you're looking at the market, this has
4:33been a rotation. This has been hedge
4:34fund deleveraging. Picking the end of a
4:36hedge fund deleveraging is not easy. And
4:39two weeks ago, I said we were getting
4:41further ahead last week. And on a
4:43technical announce thing, I said we're
4:44even further ahead. Maybe we've hit the
4:46bottom. But I also said that until the
4:49month ends and the P&L losses have been
4:53set and you get to start off a new
4:55month, especially with a week coming up
4:57that has
4:59enough important data points or worries
5:02that are still there. I think it should
5:04not be a surprise to anyone that the
5:06deleveraging continued into the end of
5:08the month. Here's the percentage of
5:10names in the S&P 500 as of Friday's
5:12close above the 50-day moving average.
5:15We started the month down at 20. We
5:17finished up at 61.
5:19Regardless of what people think, this
5:21has been a massive rotation.
5:24If you want the confirmation, I've shown
5:27you a bunch of deleveraging. Morgan
5:28Stanley's Prime Brokerage put this out,
5:30or the derivative desk, I guess, put it
5:32out from Prime Brokerage. The largest
5:36degrossing
5:40they've seen. This is the month of
5:43October. Now what I did was say hey uh I
5:47put this into chat GPT and I said those
5:50the other episodes where we had at least
5:52a draw down of 6%. Give me the time
5:54period. So it gave me these.
5:58Again this is the worst one. You had one
6:01during the tariff side this March of 21
6:04may not seem like a big time to you guys
6:07but this was right around the GameStop
6:09situation. So GameStop had started. So,
6:11this draw down actually was part of the
6:13GameStop situation. You obviously had
6:15COVID, you had uh Trump 1.0. Uh this
6:20ended up being a bunch of things
6:21surprising uh everyone at the beginning
6:24of 18 and then 2016 was really the end
6:28of the shale unwind when crude went down
6:30[clears throat] significantly. Um here's
6:32the S&P draw down. So, we are 4% off the
6:36all-time highs. And these are the other
6:38times we got down every one of the other
6:41times we got down at least 15% draw down
6:43in the S&P except for the GameStop one,
6:46which was a hedge fund unwind one in the
6:49midst of a good economy. We eventually
6:51got a draw down from the Fed tightening
6:53and the economy headed towards recession
6:55with the fastest tightening uh in terms
6:58of size and from zero that we had seen
7:00including in here SVB, all that. But
7:02here's where we are. So, this is most
7:04similar to the GameStop unwind. In terms
7:07of the VIX, the VIX has barely budged.
7:09It did get above 20 on Friday and then
7:11came right back down. This is the 20-day
7:13moving average. Uh, I wanted to show
7:16this because historically, these events,
7:18even in GameStop, uh, uh, yeah, even in
7:22GameStop, you were sitting up here
7:23already at those levels. So, we really
7:26haven't seen it in the VIX and IGCDX
7:29despite all the doom and gloom you're
7:30hearing and the unwinds that are
7:31happening and all the panic and blah
7:33blah blah. I'll go through that as well.
7:35Uh IGCDX and this includes a roll date
7:38here, this jump up otherwise
7:41and I'll show some charts of this later.
7:43This is the uh contract that doesn't
7:46have the rolls in it. It's a little
7:47shorter time period, but you can see
7:49that the investment grade market, the
7:51the panic is nowhere near in the side of
7:54what we've seen with inside the equity
7:56market, even though we have seen some
7:58panic in credit related to core, which
8:00I'll get into later because there was
8:02some news from Oracle. Um, Charlie
8:04McKilligot does great information, but
8:06what I really like is the uniqueness of
8:10what he's talking about here. Traumatic
8:12risk management unwind that saw 124 pair
8:15trades. This is a great way to measure
8:16it and a very unique way of going
8:18through it. So when I show you the the
8:19deleveraging happening at the prime
8:21brokerage level, that's taking it across
8:22portfolios. This is just picking pair
8:24trades out on a long short monitor and
8:26then showing and 47 had greater than
8:28three sigma move 264 65 uh a lot of
8:32stopout capitulation
8:36an unusual year of widespread high sharp
8:38winners. This has been the AI trade.
8:40It's been very concentrated. So you had
8:42a unique situation where all of the
8:44winners, the true winners were very
8:46related to AI, the ones that had the
8:48best sharp ratio in there and then you
8:50started to see these pair trades that
8:51just completely blew out that had great
8:54not just in stocks but in carry and
8:56everything. So carry and trend. Um and
8:59again we're at a very dangerous point of
9:01the year. So we enter a period now where
9:04uh I don't think there's going to be
9:06releveraging. So the dry powder is
9:09there. Uh we'll see if we get through
9:11next week with some uh you know with the
9:13midterms with the all the threats about
9:16Iran. We'll see. Um in terms of the
9:19market, if I was sitting and had a
9:21portfolio on and had to manage
9:22monthtomonth or dayto-day, which luckily
9:24I do not have to do anymore, uh I mean I
9:27would look at this chart, this head and
9:28shoulders type formation where we made
9:31new all-time highs and failed and you
9:33saw the same thing in the NASDAQ. I can
9:35see why people are bearish. I can see
9:36why people think this is going to extend
9:38lower. We'll see how it goes. you have
9:39this kind of island top formation where
9:41you gap lower below the the trend line.
9:44Obviously, we're going to need to get
9:46back above there until we get above uh
9:48the 50-day moving average again and the
9:51VIX breaks down to where it was before,
9:53which to me is going to be around 16. Uh
9:55I could see this staying jumpy and hedge
9:57funds still remaining in kind of this,
10:00you know, scary period. But the reality
10:02is we've been in this range for a long
10:04time, even if we made slightly new
10:05highs. When you look at it on a monthly
10:07basis, it looks completely fine and does
10:10not look like something that would
10:11suggest there's a top here. Uh at least
10:14at this point, it just wasn't that big
10:16of a month. And here's part of the
10:18reason why. This is the equal weight
10:19S&P, which ripped back higher. So, this
10:22is a monthly chart. Again, none of these
10:24look bad. Now, this one you could argue
10:26from a candlestick basis, you've got a
10:28big outside reversal month. And if you
10:31want to get bearish at this point and go
10:33through it technically, you know, you
10:35could say this high here. This is why I
10:38can see where people are starting to get
10:39a little bit more bearish. I'm going to
10:41take a different approach and just
10:42remind you the market structure has
10:44completely changed and it is forever
10:45changed. We will never go back to what
10:48we would see like this before. This is
10:51my opinion, but I feel very strongly
10:53that you have to get used to strikeout,
10:57strikeout, strikeout, home run. So when
11:01you look at that, I just want you to pay
11:03attention. So here's the chart again.
11:05During this period, for the last, you
11:08know, 30 months,
11:12you've had 30% annualized returns on
11:15Nvidia. So you've made if you just buy
11:17and hold you're getting 30% annualized a
11:20year total return of 87%.
11:24Here is one period of four months of the
11:2930. So less than 15% of the months made
11:32the entire move plus some. That's the
11:35world you're living in. Strikeout,
11:37strikeout, strikeout, home run. Which
11:40means if you're getting out here after
11:42you break something and technical
11:44picture looks bad, good luck, my
11:46friends. Good luck. That's why the
11:48bubble talk. If you're betting on a
11:50bubble and you're getting completely out
11:51of the market at this point because
11:53you're worried about the bubble
11:54unwinding,
11:56you don't got a really big window for
11:57being right. And if all of a sudden it
11:59turns higher, trust me, you start to buy
12:01into it again when it comes up here. And
12:03then all of a sudden you're back in the
12:05mush again. This is the S&P. Same exact
12:08story. the entire return plus since
12:11September. So we're dealing with 13
12:13months. This is 150% of the return from
12:17here to here all within this period. And
12:20all you had to do is buy it when people
12:21were telling you crude was going to go
12:22to 200.
12:25The war the straight of war moves was
12:27being shut. All you had to do was buy
12:28the market here. Same thing as COVID.
12:31Nobody wanted to buy during COVID. No
12:32one wants to buy it. This is the world
12:34we're in. This is the market structure
12:35shift. I put this out in March. In March
12:38of this year, I put out that we are
12:41leaving this world. I want you to think
12:43of this world as the SAS world, the
12:47Google world, the Amazon world, the Meta
12:49world. Guess what, guys? We're now in
12:51this world where they're borrowing
12:52money. Back here, they didn't even take
12:54any debt and their earnings beat every
12:57single month unless it was a macro
12:59event. Now in a incredibly strong
13:02economy and this is from March oil
13:05prices higher rising interest rates
13:08again rising credit spreads rising
13:10volatility trap money selling. This is
13:12from March before all of this stuff went
13:14on saying to you this is what we will be
13:17in. This environment is what I lived in
13:19in emerging markets. This environment
13:22was fake. This is the one that was fake.
13:24This was the QE world. We pegged rates
13:26at zero and these companies printed
13:29money with no draw downs. Here's the way
13:32it looked if you didn't live through it.
13:35That's it. Now, this is the hyperscalers
13:39since this period at the end of last
13:41year. Voom voom voom voom voom. This is
13:44the hyperscalers. They've gone nowhere
13:47since [snorts]
13:49October of last year. Nowhere.
13:52And yet you've had these violent ups and
13:54downs. So, if you treat this as a bare
13:56market when all it is is these companies
13:59had no debt, now they're making a big
14:01bet on this. If you want to bet that
14:03it's a bubble and they're not going to
14:04get it, good luck. They have the most
14:05pristine balance sheets despite the fact
14:07that their CDS is widening out. Their
14:08balance sheets are phenomenal. The
14:11concentration in the market of how big
14:13these guys are in both the equity market
14:15and now the debt market. That's what has
14:17happened this year. These guys were
14:18non-existent in the debt market. Now,
14:21they are massive in the debt market.
14:23Adapt or die.
14:25hyperscalers relative to the S&P. Again,
14:28what do you want to do? This has now
14:29been three years of this. If you take it
14:32back since 2020, the hyperscalers of
14:34which there this includes the four of
14:36them, the Amazon, Microsoft,
14:39uh, Google, and Meta
14:41unchanged relative to the S&P, but the
14:44waiting in people's portfolios is so big
14:47that when these names move like this,
14:50this is Google relative to the S&P, down
14:5312% this week. Biggest drop relative to
14:56the S&P. I had to do it relative to the
14:57S&P because we've had days that Google
15:00was down a lot, but if the S&P is down a
15:02lot, it doesn't matter as much. So, you
15:05got to go all the way back to the great
15:06financial crisis. This has never
15:08happened. This is new stuff going on.
15:10Remember this year, Meta races 2004 a
15:13trillion dollars in value, biggest
15:14wipeout in history. Google faces $500
15:16billion stock white out. One of the
15:18biggest market cap drops ever. Microsoft
15:19loses 357 billion. These are all daily
15:22moves. This is this year. Meta's this
15:24year. Google's this year. Nvidia was on
15:26DeepSeek. That was 589 billion. These
15:30numbers are massive. 589 billion is 20%
15:35of the Russell 2000 market cap of all
15:382000 names combined. So when they move
15:41and they have a month like this, this is
15:43two weeks where they fall 11%.
15:46And these types of moves are happening,
15:48of course, the market is going to be
15:50responding in a way. These are
15:51incredibly big moves. But here we are
15:54year to date. These things haven't
15:57budged. And it's not a surprise at this
15:58point that Microsoft is the best
16:00performing one now. Not because of their
16:02business, but because they've been the
16:05one that hasn't raised any real capital
16:07this year as I'll go through. Um here's
16:10the hyperscaler chart. To me, this is
16:13the most important thing when they got
16:14locked in with the AI trade. I mentioned
16:18back in this period when I was writing
16:20the stuff about
16:22um the fireworks show is over that it's
16:25some point the lack of correlation
16:27between these remember the hyperscalers
16:29were going down at the beginning of the
16:31year and were down on the year while AI
16:33trade was working that is because the AI
16:36trade was mispriced hedge funds mutual
16:38funds everyone that I knew was
16:40underweight semis the way they should be
16:42for the memory inference trade well that
16:45all changed during this period in here.
16:46And so now these things have become
16:48correlated. And I think they will pretty
16:50much become correlated. If you're
16:52doubting the hyperscaler's ability to
16:54build the capex, then you have to get
16:57out now of all the memory and all the
16:58stuff there. That was not happening
17:00here. But now everyone questions this.
17:02So they become locked together much more
17:04than they were before. That's the
17:06correlation change that's happened for
17:08everything for the week. How bad it
17:10felt. Here's my agentic uh portfolio.
17:13And this includes a lot of things that
17:15were down big for the week. I mean, I
17:17got crushed this week in Marll, which
17:19was down uh 20 plus percent. Uh I have
17:22silver that was down big. Uh I have EWZ
17:26and kind of retirement type money, which
17:28was down pretty big. Uh not pretty big,
17:31historic big. Uh at least not too
17:34historic, but for the week, the Agentic
17:36portfolio was up 1%. So it felt a lot
17:39worse than it was. Here's something I
17:42just want to make sure people realize
17:43because the AI pair trade Goldman Sachs,
17:46this is the white line. It gave back
17:47basically it gave back the entire year.
17:50So, this is the winners versus losers.
17:53So, this is a a market neutral uh or at
17:56least dollar neutral uh setup that
17:59Goldman Sachs put together uh down here.
18:01Now, the red line here is the SMH. So,
18:03you've already given back most of the I
18:05trade. The only way I can see this
18:06getting worse from here uh is if there
18:09is a bubble. So at a minimum I think
18:11you'll stabilize and I do think once the
18:13stabilizes semis will be fine. I for the
18:16first time since probably March or
18:20April. Uh I started making my sales in
18:22Micron. I did my rotation but most of
18:25the money went into silver and crypto
18:28and left the AI side and I left the
18:30Marll alone. Nvidia, I've added at times
18:33uh in small amounts, but this week I put
18:37on more AI names this week uh
18:40particularly in the optical side and my
18:42Micron position is back up to a
18:44decentsized position as of this week.
18:46I'll talk more about that as we go. Here
18:48is the winner side of that Goldman Sachs
18:50uh basket overlaid with the semis. And
18:53this again is meant to show how
18:55correlated the trade is. So semis, of
18:58course, they sell off with these other
19:00names. So if Intel goes down 20 plus%
19:03it's causing losses across things. How
19:06do I know it's causing losses? Because
19:08this scary chart and this is why the
19:09deleveraging is not only obvious but
19:11this is the hedge fund positioning guys.
19:14This is what has caused the AI unwinds.
19:17Hedge funds got over their skis. This is
19:19the Goldman Sachs VIP hedge fund index
19:23relative to the S&P overlaid with that
19:26AI long short basket.
19:30So this is the positioning. This is why
19:32when the deleveraging is happening, the
19:34AI trade goes down. So again, as I've
19:36said, hedge funds are still up for the
19:38year as an aggregate number, but they've
19:40given back a significant part. So you go
19:42back to this was spreading into other
19:45areas, and as I showed last week, we got
19:47into France more last week. We got into
19:50coreweave the week before. This was a
19:52horrible month for leverage because we
19:55spread completely out of what had
19:57happened in July. Remember when I showed
19:58you that Morgan Stanley chart? This is
20:00lower than what happened in situational
20:02awareness. Situational awareness was
20:03very isolated to the equity side. This
20:07has now spread to other parts of the
20:09market, which is why it makes sense for
20:11me for people that have seen this
20:12before, like the dot bubble, like the
20:15housing market, like whatever they want
20:16to talk about, which I live through as
20:18well, that this looks a lot like it. In
20:21some ways, it does. In other ways, as
20:23I've shown, it doesn't at all. Um, I did
20:25do a technical analysis. I highly
20:27recommend for the subscribers the same
20:29way you can use the prompt at the
20:30beginning. I would definitely go look at
20:32the technical analysis thing I've
20:34created because I have this hundred name
20:36index and because this index is the
20:39thing that represents
20:41across industries the agentic side
20:44having a technical side where I'm
20:46showing you the RSIs I'm showing you the
20:48Williams indicators a whole bunch of
20:49technical analysis I did this on Monday
20:52basically with the story being we are
20:54very close if not at a low theatic
20:56portfolio was up for the week it didn't
20:58feel like it but I'm looking a lot and
21:00doing a a lot more work. This is where
21:02we got on the average RSI. I'll show
21:04everyone this one. We hit the lowest
21:06level basically of the year. Again, this
21:09is where we hit it situation where this
21:10is the average daily RSI. So, this is
21:12taking the 14-day RSIs for the 100 names
21:14and taking a chart of the average of
21:16them. And you can just see where we are
21:17now. Um, I just want to highlight this.
21:21I there are two weekly sheets, guys,
21:24that I've been uploading for the past
21:26month. Um, I highly recommend using the
21:29weekly one and transitioning to it. This
21:32one to me has a lot of stuff in it that
21:34right now is important because this is
21:36using weekly RSIs and the daily RSIs. In
21:39particular, the care the columns I care
21:41about the stages and the MACD's. You get
21:44a whole bunch of other data in there,
21:46including the uh ATR. I'll be doing this
21:49for crypto as well at some point in the
21:52very near future. I highly recommend
21:54doing what I did here. Here I want you
21:55to focus only on the weekly technical
21:57sheets in the folder. Give me the
21:58breakdown of the column with the base
21:59and this. So again, put them in one
22:01folder. Every week you download them.
22:03Just upload them in a folder and then
22:05connect via whatever one you want. In
22:07this case, I'm using Claude Co-work. And
22:09then I just say go in and show me very
22:11quickly. So you're getting all the work
22:12done for you. I'm giving you the ability
22:14of doing this yourself. Which ones have
22:16an advance uh in the column? These are
22:18names that already are starting to move
22:21uh on the high side. And what I did to
22:24try and screen for names technically
22:26that look good is I just said I only
22:28want ones that are either listed at
22:29advance or pullback. I want the the MACD
22:32histogram turning up and I want above a
22:34200 day moving average upward sloping
22:36200 day moving average. So again it gave
22:38me only five fit the number. Uh these
22:42names are all actually hanging in there.
22:44So they're charts that should be okay.
22:45They've done well and that way if
22:48they've done well so far that probably
22:50means they don't have a lot of hedge
22:51fund risk. Marll was in that camp for me
22:53until they made the decision to uh I
22:57don't know possibly buy Lentum. I got
22:59asked a lot of questions so you guys
23:00hear it on this. I do not like being
23:03involved with a company if this ends up
23:05being true. I just think it takes away
23:08the possibility of uh a doubling as fast
23:11as I thought. I still think the name
23:12will go higher and I won't be out of it,
23:14but I probably will reduce if if this
23:16deal happens and once we get a bounce in
23:18the name. I just don't like being in
23:20there. So again, the AI names, they're
23:22basically just hedge fund names. You
23:24guys have to get the technical analysis
23:25to work on the AI stuff. I think it's
23:28the most important thing in the market.
23:29And just remember, if you look at things
23:31this way, and you remember that I did
23:34this video in March saying this is the
23:36world we were headed into forever. This
23:40is absolutely in my case. There is no
23:44chance or very little chance unless
23:46inflation were to drop dramatically and
23:49labor were to fall off dramatically,
23:50which will happen at some point most
23:52likely from AI, but we are nowhere near
23:54there yet. Rates are going to stay on
23:57the higher side. I think we probably
23:59made a high in long-term yields uh for a
24:02variety of reasons which I'll go
24:03through, but we are not going back to
24:05this anytime soon. I also don't think
24:07the hyperscalers or any of those
24:09companies should be as big a weight in
24:10the portfolio which I also talked about
24:12right now if you still whatever you have
24:14ETFs this is for RAAS and for FAS reduce
24:18the positions in the hyperscalers and
24:21increase the positions in the agentic
24:23compute side this is the world they are
24:25becoming more and more susceptible to
24:27this make sure that you have longevity
24:29in your portfolio like Eli Liy which
24:32I'll go through and make sure you have
24:33energy stocks as well uh For more
24:36information, go to Mark Whailing. He's
24:38been doing an awesome job. You guys have
24:40been great about looking for feedback.
24:41And just one more thing. I know we have
24:43a ton up there. I did three videos this
24:46week and I produced two papers. Uh I
24:48think it was two. It was definitely one.
24:50May have been No, it was two because I
24:51did an optical paper, too. Um I know it
24:54gets hard to keep up with everything. We
24:56will have an AI tutor group that'll be
24:58connected with us to make it easier both
25:00to get through the sites, but most
25:01importantly for for you that have
25:03purchased this for your kids to allow
25:06them to understand financials and use AI
25:08together. We are going to make this
25:10where you can reach out with them. So,
25:11this will be a service that goes up on
25:13on online soon. You can reach out to
25:15Mark for more information. So, let's go
25:16through the portfolio. Marll, like I
25:18said, was acting great and then all of a
25:21sudden the momentum deal uh destroys the
25:24stock for a day and then the second that
25:26Intel goes down, this goes. I still love
25:28the name and I I I've diversified into
25:32some other optical names as well. I have
25:33not reduced that position yet. Uh little
25:35island top here on Micron, but as I
25:37said, I did increase my Micron position
25:40for a variety of different reasons. Um,
25:42but most of them have to do with the
25:44fact that now that we've gone through
25:45this and we got the SKHEX numbers and
25:48we're closer because of Astra to
25:50humanoids, uh, the memory stiff is going
25:53to become even bigger. For those of you
25:54who keep reaching out in silver, silver
25:56is part of my portfolio. It was
25:58obviously not the best decision to
26:00rotate in there. Uh, from here, it's
26:02outperformed many, many of the AI names.
26:05Uh, I mean, Micron fell 40%. Uh, but the
26:09reality is this is a horrible chart to
26:11be trading. Uh, I own it for a variety
26:14of reasons. I think you might get a
26:16payoff in it this week. Uh, but I
26:18wouldn't be betting on it. But
26:19regardless, uh, silver I still like
26:22because it's the same story as before.
26:24It's a big part of the next phase, the
26:26embodied phase. Uh, it's where it
26:28accelerates. So, I do think once silver
26:30makes a technical bottom along with
26:31gold, you still have central banks
26:34buying, you still have tech companies
26:35buying it. it has just been destroyed
26:37against retail but also CTAs are shorted
26:40in a big way as well. Now, um, in terms
26:43of EWZ,
26:45just a horrible fall. Um, again, it's
26:48the biggest fall since I came back from
26:50Brazil. Uh, except for 2020, but again,
26:54this was a unique scenario where these
26:56two things were not Brazil specific. So,
26:57if I did it relative to the S&P, it was
26:59bigger on a on a monthly basis. It's
27:03just not that big of a move. So, part of
27:04the reason the day was so big for
27:06everyone panicked, um, is because it had
27:09been up so much in the month. So, this
27:11is the monthly move, which is nowhere
27:13near as big as some of the other ones.
27:14Um, Eli Liy, I'll take a uh a chest bump
27:19on this one in terms of the story behind
27:22it. I'll go more into the details, but
27:24I've said it before, I'll say it again.
27:26I think this will be the biggest stock
27:28before we get to 20 the end of 2030 uh
27:32in the S&P 500.
27:34So, I'll just go there. Even if it's
27:35not, I mean, it's moving up the ladder
27:36as it is. I think it's number nine now
27:38uh after this week. SMH got down to the
27:41200 day, got below it, held it. I'm
27:44expecting this to hold here. So, unless
27:46we're truly in an AI bubble, and there's
27:47a lot more unwinds, everything
27:50fundamentally in this week was better
27:51for the space. Um, aside from the
27:53idiosyncratic
27:55part of Intel, which I don't think we
27:58should forget. Intel was a $20 stock not
28:00that long ago. The government got in at
28:01around 20 bucks. Um, remember the
28:04government's in the company as well. So
28:05yes, they've given back a lot of gains,
28:07but they're still up significantly as
28:09most of the semi-names are. Uh, one of
28:11the themes that has gone on, which I
28:13think will continue to some degree, but
28:15I just wanted to highlight it, um, is
28:18quality. So this is the quality index in
28:21uh, in Bloomberg. The reason this has
28:24been a big story, it's been in play
28:26since we got into July, and that's
28:28really when the credit fear started.
28:29Whenever credit is an issue, you get
28:32this. So this is the IG uh the yellow
28:34line there is the IG
28:38uh option adjusted spread barely has
28:40moved but again I'm showing this because
28:42it's very correlated with quality. So
28:44the market has made a move into quality.
28:46Mike Wilson has talked about this. I
28:47actually think at this part of the cycle
28:49until rates were to go back down
28:50significantly as long as tenure rates
28:53stay up above four and a half which I
28:55think is likely that they will even if
28:57we go back down there uh in a short
28:59burst the demand needs and credit are
29:01just going to keep flying out. If rates
29:03were to go down I would expect that the
29:06AI companies would come with a lot of
29:08issuance. I think they'll slow down as
29:10rates go higher which we've seen this
29:12year. So, I think this becomes something
29:14that you're going to see large issuance
29:15if they can borrow money at 4 and a
29:17half, less interested if we get up to
29:19five and a half. Um, here is the Apple
29:21chart. If you wanted to know why Apple
29:23could trade at a 40p while the rest of
29:25the hyperscalers are getting hit, it's
29:28because of quality balance sheet. They
29:31haven't spent any money. Um, midterms
29:33coming up. Uh, Iran got to worry about
29:37that since he's making all these
29:38threats.
29:40possibly the most important but least
29:42talked about for everyone watching this.
29:45The fixed income guys know this but you
29:47just have to remember that the quarterly
29:49refunding announcement is going to come
29:50happen uh I think it's supposed to still
29:52happen on November 4th. Uh Bessant will
29:55likely cut US long bond sales next month
29:57says city. If for any reason he has
29:59tried to do other things uh it's just
30:02something to keep on your radar. We also
30:05have the anthropic IPO very likely now
30:08in a couple weeks. So, it's going to be
30:11a big month. Uh, but I think we took out
30:14most of the risk from the IPO this week.
30:16Uh, this is forgotten and I read a lot
30:20of things and if you didn't take this as
30:22a positive, you're just wrong. Um,
30:24there's just no way around it, at least
30:26for the near term. Again, I don't want
30:28to own this thing. [clears throat] I'm
30:30not sure that what the valuation will be
30:32in a few years, but I do believe that it
30:35is a very hard business to be fighting
30:36up in commoditization. As someone who
30:38uses ChatGpt more than Claude, uh, but
30:42also uses Grockbot now more than Claude,
30:45uses Muse more than Claude, I can just
30:48tell you from my own experience that all
30:49of a sudden's going and this week I did
30:51use Gemini a lot. not because it's able
30:54to do a lot of the stuff Anthropic has
30:56done for me, but the model has gotten
30:58very good and I trust it now far more
31:00than I did with Argon. Um, here are the
31:02highlights. The main one here I want to
31:04just go through, um,
31:07is 132 billion in cash. This was a very
31:10big surprise to the street that they had
31:12this much cash, that they hadn't already
31:14been burning cash. And when you're
31:15dealing with a credit market and with a
31:17capital market this year that has raised
31:19a trillion, yes, a trillion plus dollars
31:21for the AI trade um raised it not out of
31:25free cash flow. So free cash flow did a
31:27lot at the beginning, but then we got
31:28into raising money. That is equity, SPVS
31:30and and debt. It's a big number. Um so
31:35you can go through this on your own, but
31:36the revenue is growing much faster than
31:38expenses. Uh the enterprise adoption is
31:41broadening is broadening rapidly.
31:45the numbers are are just big and I asked
31:49uh I put everything in uh on the LLM
31:53into chatbt. I use the highest model and
31:56I just ask it where its scenario would
31:58be based on everything it took from the
31:59S1 and it has the base case of 100
32:01billion 95 to 110. So again, this is why
32:05you don't have to listen to X. just
32:07upload the actual S1, have it go
32:09through, and based on all of the metrics
32:12that have been in there and the nuances,
32:14just see where it thinks it's going to
32:15be. So, just as a comparison, so you get
32:18the success of this. This is Anthropics
32:20column, their S1 filing and Google's S1
32:23filing. I mean, the revenue 1.4 versus
32:274.5 in terms of the prior full year.
32:30Look at this.
32:32So, these numbers are staggering. But
32:36the staggering part is the growth. So
32:38the year-over-year growth. So when
32:39you're trying to compare the success of
32:41Google and again this is a fair
32:44comparison as far as I'm concerned and
32:46looking at companies and just what's
32:47gone through for people to doubt what
32:49Anthropic has done when we're talking
32:51first of all of something that is you
32:53know 13 14 times but in terms of the
32:57difference here it's just staggering how
33:00fast they are growing coming into it and
33:02yet people want to take the nuances of
33:04an up quarter or down quarter this is
33:06going to take this out and I just think
33:09this is a big deal and if you didn't see
33:11it, one of the most important parts of
33:13the S of the S1 was the fact that the
33:16adoption rate has been so low for
33:18companies at this point for anthropic
33:20except for the fact that they have half
33:22of the Fortune 500 companies already. I
33:24don't see that changing. I do see open
33:26source as being a part, but I'm going to
33:28keep saying this to people. I think Sam
33:30Alman talked about this on Cheeky Pint,
33:32which I'm I haven't gone through yet,
33:33but I saw some quotes which I've talked
33:35about on here. building data centers and
33:38having your own open source. That's all
33:40well and good, but that's not simple,
33:42especially when there are massive supply
33:44shortages in memory. I just think people
33:46are overestimating
33:49what's going to go on when they look at
33:52things and Opus 5.5 is a massive step up
33:56function. It is a cheaper model than the
33:58open source in many many ways because
34:00you have to get into how good the model
34:01is and remember they haven't released
34:04their better models which is preventing
34:06the distillation from happening on those
34:08models. So, I think you just have to go
34:10in if you don't trust what I'm saying.
34:12S1 all-in covered it. David Saxs who has
34:15definitely not been a fan of Anthropic.
34:17Um, so you got all these, you know,
34:19Anthropic bears who are saying they've
34:20made all these capex commitments in the
34:22future and they're never going to be
34:23able to live up to well actually this
34:24secures those commitments because
34:25they've got the capital now to basically
34:27b back up all those orders. And he's
34:30saying all the they're going to be able
34:32to do all the forecasted capex spending
34:34till I don't know 2030. You have to
34:37start thinking about this. This is
34:39really critical because this has been
34:40the year that everyone had to realize
34:43that they need to buy Micron out to
34:452030. So, we're out of all of these
34:48materials to 2030
34:50based on demand. At the same time, they
34:52said, "Oh crap, we got to go get as much
34:54capital as we can get in the door. So,
34:56let's go raise it." And these guys have
34:58130 billion before the capital raise.
35:02Chimamath who literally just went on and
35:03said they should be issuing this at a
35:05trillion dollars because they don't have
35:07enough capital. He said the fact that
35:10entering June already has$125 billion is
35:13unbelievable. So when I said they need
35:15to raise 20250 I was assuming they had
35:17only 30 40 50 billion balance sheet I
35:19was wrong. It's a lot to hear Chamas is
35:22wrong. Uh Enthropic needs significantly
35:24less financing than expected. And so
35:27he's basically saying the IPO should be
35:28somewhere in the 1.3 to 1.7 range. And
35:32oh, here's Hyperlid this morning on
35:33Saturday, right around the 1.7, just
35:35above it. That's where it's trading
35:37right now. So, the IPO is probably going
35:40to be in good shape now that we've got
35:42that. And there's a chance based on the
35:44fact that they left a uh a spot where
35:46they can put in the October number,
35:48which as Gavin Baker is talking about,
35:49probably means it's going to be a good
35:51number. Gavin Baker was on TBPN uh
35:56highly recommend listening to it. He
35:58went through and again less than.1% of
36:01the expenses for the clients that they
36:03have showing that the adoption has not
36:05picked up. The reason this is so
36:06important is I believe their agentic
36:08workflows are going to be a major major
36:10part of the companies that they already
36:11have. So you're going to see a step up
36:13function. Again, for people who don't
36:16use the technology, I'm saying again and
36:18again, if you're listening to people
36:19telling you the bubble is going to blow
36:21and they don't use the technology or
36:22even understand what's going on, I keep
36:24trying to direct you and your family to
36:26use the tools, use the prompts, go
36:29through it. That alone will put you in
36:31the.1% of the globe. Trust me, it is
36:34worth your time. Uh enterprise adoption
36:37is still very early based on these
36:39numbers. Uh the October AR likely
36:42accelerated significantly for the
36:43reasons that I mentioned. You guys can
36:45read that on your own. GPU rental prices
36:47are rising as compute has moved into
36:50contango.
36:52That's what's happened in October. So
36:54for all the bearish stuff, all the
36:55spooky stuff, the market's gotten
36:57tighter, guys. Computer's gotten
36:59tighter. A lot of the credit and the
37:01things you've seen, it's assetbacked and
37:03the GPUs are part of it. You can still
37:06bet that in the future this isn't going
37:08to happen. But every single thing you're
37:09having, including the now the capital is
37:12telling you the opposite, guys. This is
37:14an opportunity. This is not a time to be
37:16listening to people telling you you're
37:18going to lose all your money. AI is a
37:21transformative technology. It is an
37:22inevitability. There is so many good
37:24things happening. Compute has shifted
37:26into contango and that is based on
37:28silicon. Uh Baker argues that public
37:31investors are focused on the amount of
37:32physical compute capacity online during
37:342027, but the pricing data suggests
37:36something different. That future
37:38capacity is becoming more expensive.
37:40This becomes very important. That's why
37:42it's in green. When I put something in
37:43green, you guys know it's important.
37:46It's not about the demand side. It's
37:48about something else, which I'm going to
37:49get back to when I get into the credit
37:50side. Let's go through the earnings in
37:52the S&P. Everything's still good. You've
37:54got growth happening in the mid 25%. It
37:58is less than last quarter but last
38:01quarter was insane numbers and I think
38:04going forward if we settle in and keep
38:07declining but keep growing at 20% that
38:10is very very different as I've talked
38:11about particularly as profit margins
38:13continue to be strong
38:15uh and the guidance again here's the
38:18guidance of what happens
38:21in the first month of a quarter normally
38:24here is what happened
38:27these last three it's actually stronger
38:28than the last quarter. So the guidance
38:30is still incredibly good based on what's
38:32gone on. So you've got the earnings
38:34growth the S&P right now as it's 30%. I
38:39think the expectations are more for like
38:4025. Uh but you can see how many are
38:44growing and more importantly the revenue
38:45side is still higher than what was
38:48expected in September. Um it's not just
38:50the US and this is the important thing
38:52from a breath basis. It doesn't matter
38:54whether you're a big company or a small
38:56company or whether you get things
38:57overseas, you are still growing. So this
39:00shows the earnings growth based on all
39:03the S&P 500 companies right here. And
39:06then if you have greater than 50% in the
39:08US and greater outside. So we're getting
39:10lots of stuff. Europe's earnings growth
39:13to rival US for the first time in years.
39:15Now this has been where the
39:17disappointment is is the MAG 7 versus
39:20the other ones. This has mainly been the
39:22others. So when you're looking for why
39:24the breath is going out going in there,
39:27this is what was anticipated. This is
39:28what's actually happened. You're
39:30starting to see some of the negative
39:32signs of the, you know, the backlog, the
39:35amount of money they're spending, and
39:36the inability to get the ROIC in the way
39:38that they should at a big level, but
39:40their balance sheets are pristine. So
39:41you get multiple compression, but when
39:43they have bad months or bad weeks, like
39:46I said, it has huge implications. You
39:49can get the actual ones here. The
39:51majority of the stuff that ends up being
39:53the issue is Meta, but you can see that
39:56Alphabet, Amazon, all of them little
40:00little dangerous. Now, on the earnings
40:02news front, SKH Highex blew away stuff,
40:04reacted good that day, fell later in the
40:07week as the US market unwound. Um, just
40:10going through the side by side of what
40:12they said in each quarter. I'll leave
40:14you guys to this, but the progression is
40:16still getting stronger and stronger
40:18despite what the stock price is doing.
40:20Don't try to read the stock price to
40:21anything more than hedge funds got too
40:24levered. Retail got too levered. The
40:27unwind isn't finished yet, but the
40:29business is still growing rapidly and
40:31they've already raised their capital
40:32that they need for the near-term. They
40:34were a big capital raise. The investor
40:37the investor base is broadening also
40:38including non- tech companies investing
40:40in their own AI infrastructure,
40:42sovereign AI or onre demand. This is the
40:44most important thing. This is the Jensen
40:45Yuang story. 90 trillion not 10 guys.
40:49There is 90 trillion of spending that
40:52will happen. Capex is starting to show
40:54and broaden out away from just the
40:56hyperscalers. That is critical for what
40:59I'm about to show in the debt markets.
41:00It also means, like I said, you'll have
41:03more people looking to raise money
41:05through either free cash flow through
41:07debt, through equity. There will be more
41:10capital that's needed, which probably
41:11means more multiple compression,
41:13probably means rates stuck at higher
41:15levels, but it will be spreading. won't
41:17just be those companies and it won't
41:18just be the suppliers. It will also be
41:20the Goldman Sachs and Morgan Stanley's
41:21and everyone else. Um just so many
41:25positive news stories in terms of H uh
41:27HBM to DRAM. I'm not going to go through
41:29this. Nvidia talking about where
41:31everything is tight. This is from trend
41:33force of every single part of it. This
41:36is where it is in terms of supply
41:38demand.
41:40Elon Musk put this out. Robot RAM needs
41:42will exceed everything else on earth as
41:44you need the robots to be physical
41:46substrates in order to shape atoms. Why
41:49robots could become the largest
41:51consumers of RAM when I have talked
41:52about why memory and this is the major
41:54reason aside from the SKHEX news along
41:57with the Micron news. There is not one
41:58bit of slowdown that's happening inside
42:01the memory names. They're obviously
42:02extremely cheap for valid reasons, but
42:04they keep doing more contracts out to
42:062030 and the demand keeps going. But
42:08this part is dramatic. You will have new
42:11buyers coming on in massive numbers
42:13starting next year. The ramp up will be
42:16begin next year. But every year going
42:18forward for Optimus, the massive plants
42:20that he's building. Elon Musk is telling
42:23you that to work they need tons of
42:26memory. Robots will need onboard memory
42:29for processing visual and sensor
42:31information continuously understanding
42:32the environment and planning movements.
42:34I've written about this with VLMs. I've
42:35written about this so many times with
42:37embodied AI. This is why when I say okay
42:41all we did this year was the coding
42:42agents then we started the consumer
42:45agents in September. Next we start the
42:48embodied agents that starts next year.
42:51Remember Tesla said there were three
42:53things that mattered for when they could
42:55start scaling. Dexterity in the hand
42:58scalable manufacturing. his plant is
43:01very much already being built and
43:02Moonshots did an episode from it and
43:04just I remember when he said I'm
43:06completely changing how many robots will
43:08be built by 2030 by being inside inside
43:12uh the manufacturing plant but it took
43:14real world intelligence and I
43:16highlighted that unit out of China said
43:20it's poised for a chat GPT movement
43:23related to when they got the
43:24intelligence and then that was in the
43:27summer and then this month.
43:30Open AAI's GPT6 may lead to robotics
43:33breakthrough. Latest advances have
43:35accelerated prospects for a chat GPT
43:37moment have accelerated. Guys, [sighs]
43:40go by memory names. Trust me. Corning
43:43profit surges on AI demand. Corning
43:45reported this was a big report. I wrote
43:48a paper because of how it was. Demand is
43:50accelerating. Demand is pressing against
43:52the company's ability to add capacity.
43:53Optical connectivity could expand 10fold
43:56per GPU. That's a t-fold increase. We
43:59continue to be quite convicted that this
44:01accelerates in 2028. Supply is
44:03constrained.
44:05Data center in interconnect opportunity.
44:08This is mainly related to the needs that
44:10are coming with efficiency for per GPU.
44:13This is mainly a Vera Rubin story, a
44:15Fainman story, everything that goes
44:16going forward. So again, I've I've done
44:20so many reports on optical and I've been
44:22waiting for the step up moment.
44:24Corning basically said we are getting
44:27close to it right now and they have this
44:29demand stretched out. So the question is
44:31when are you going to start to see here
44:33are the main um uh quotes from there he
44:37talk even talked about the S1 uh talked
44:40about the their number one constraint is
44:42now will continue to be compute and this
44:43is for anthropic but again anthropic has
44:45the cash. So, I wrote this for the
44:47subscribers. Definitely go through. I
44:49showed a bunch of names in there that
44:51are the most likely to benefit from that
44:53report that I did three weeks ago. I
44:55would spend a lot of time on optical. We
44:57had a nice bounce in optical. I started
45:00buying here. I bought more in here and
45:03then I bought more in here. So,
45:05hopefully I'm averaging in. Uh none of
45:08these were Marll. I just haven't sold
45:10anything in Marll and may not if they
45:12don't do the momentum deal. But uh if
45:14they do do it, there's other optical
45:16names that I like that I will buy and
45:19have bought that are in there. 10ear
45:20rates uh sneakily
45:24peaked here around October, what is
45:27that? Uh that's October 7th and have
45:31broken below the 50-day. They hadn't
45:32been below the 50-day with a close. We
45:34closed there two days and we closed a
45:36week below it. Question is, is that a
45:39technical sign? I said I could see us
45:41get down to 450, especially if Bessant
45:43messes around at the quarterly refunding
45:45announcement. More importantly, um the
45:48fact that it's headed lower takes off
45:50one of the bare trades and now
45:51technically we're just in a situation
45:52where I would watch it because the CTAs
45:54are very vulnerable here to messing
45:56around. And of course, the infamous will
46:00bonds blow up
46:02economist cover that they put out
46:05right at the top. Good work. Uh this has
46:09been a major story this year. It's had a
46:11huge impact. Uh and again this to me is
46:13not about the debt. You know the
46:15government is running into trouble. Uh
46:18but it is definitely related to the AI
46:21trade. It's related to nominal GDP but
46:23is also related to the capital needs
46:25that have come in because there was
46:26really no need. We've had no housing
46:29market to speak of. So the long end of
46:32the curve has been dominated by the
46:35hyperscalers. Um, and so real tenure
46:38yields have gone higher. This has hurt
46:40Bitcoin. This has hurt silver. This has
46:41hurt everything. I don't think this is
46:43going to be the story next year. I'll go
46:45through that. Um, and the reason that
46:47real tenure yields, as you can see here,
46:49have gone higher. Uh, is because this is
46:52actually nominal yields. So, the real
46:53yields are are up here though as well.
46:55We haven't seen any change in inflation
46:58expectations on the swaps from where we
47:01were a year ago, which is unbelievable.
47:03So we did have the rise at the beginning
47:05of the straight horror moves then we
47:06came off. This is both one and two-year
47:09which is even more shocking. So we'll
47:12have to see for everyone saying the bond
47:13bear market violence. This just shows
47:15what happens when you have a month like
47:17October. This is total return for the
47:19TLT.
47:21It is unchanged
47:23basically since the hawkish speech that
47:26Kevin Worsh gave. Um Goldman put this
47:29out. I was surprised how many people I
47:31talked about this week that didn't had
47:33not seen this or looked at it. I think
47:35this is important for people to just
47:37think about and go through. They were
47:39showing the impacts of software, the
47:41impacts of energy, the impacts of
47:42tariff, and how they're all going to
47:44roll off this year, which is a fact.
47:45Unless energy prices go up to 240, we
47:49basically went through and have doubled
47:50this year starting most of that
47:54happening in March. So, you start to get
47:55through positive comps on that. The
47:58tariff effect is rolling off every
47:59single month now. And the software side,
48:02which is a, you know, an error or way
48:05that they measure it, has not been
48:06talked about either. But guess what,
48:08guys? Here's DRAM prices. The white
48:10line.
48:11Now, this is not DRAM prices. DRAM
48:14prices are still going up. This is
48:16year-over-year DRAM prices. We are
48:18collapsing here, and we will continue to
48:20collapse. This is JGB's 10-year yields.
48:23Don't think that this doesn't matter to
48:25Asian markets. We have seen what
48:28happened to Intel. They've been impacted
48:30by higher memory prices. Memory prices
48:32are slowing down in terms of the rise.
48:35So that component of PCE related to
48:38phones and the way memories are double
48:39counted in there will be rolling off.
48:42This has been a year about global bond
48:44sale shattering records for the AI bid.
48:46I highly recommend go back to August and
48:48listen to this Goldman Sachs exchange
48:51with Allison Nathan and two credit
48:53people. Highly recommend going through
48:55it. I'm going to give you some of the
48:56details in here, but they covered this
48:58extremely well. I listened to it in the
49:00summer. I didn't want to bring it up
49:01until we got to a point where either
49:02rates started to turn the other
49:04direction. But since we've had so many
49:06things go on, I think it is critical for
49:07people to realize. The main thing for me
49:10was waiting for something that Oracle
49:12did this week that Gavin Baker talked
49:13about, which I'll get through. Critical
49:15critical moment. AI debt issuance has
49:17become the dominant theme in credit
49:18markets. Hyperscalers can increase
49:20leverage significantly. This is really
49:22critical. Their balance sheets are
49:23pristine. They talk about it. they can
49:25go much much bigger. But
49:28there are some issues and this is one of
49:31the things that you need to understand
49:32and I'm going to cover this in detail.
49:35First of all, they expect their
49:37financing needs to peak in 2027.
49:40Everyone keeps looking at these charts
49:42of the amount of capex. At some point,
49:46the RO ROIC comes in and the capital
49:49needs which are funded for building the
49:50initial data centers for the compute
49:52where the money comes in. Then all of a
49:54sudden the fi you run into peak
49:56financing needs and that's what they're
49:57saying is going to happen next year. The
49:59constraint this year and this has been
50:01the most important thing. We came to the
50:02market with capital in massive size on a
50:05short amount of time. That's the biggest
50:07thing that happened. The constraint is
50:09investor absorption. It is also
50:11concentration.
50:13That's a big thing which I'm going to
50:14get in. Concentration limits are
50:16becoming the binding constraint for AI
50:18financing. So the reason CDS has been
50:20widening out guys is these deals are so
50:23big and they already have equity
50:26exposure that a lot of places insurance
50:28company demand for longduration debt
50:30fell approximately 50%. Not because they
50:33don't need more long-term debt, but
50:35because they have too much of these 10
50:38four to 10 companies. Uh it's it's
50:40probably less, but um so they talk about
50:43they're going to have to go to the
50:44private markets, which means SPVS, which
50:46you've been seeing. And there's talking
50:48about how there's a tremendous amount of
50:50buying power, 4.5 trillion. And
50:52remember, as these software deals
50:55continue to mature and people get their
50:56money back, these guys, all of the stuff
50:59they're doing will be related to AI. So,
51:01you'll get concentration there. They're
51:03just switching from one to the other.
51:04I'm not saying there won't be bad
51:05investments in five years on that. I'm
51:07just saying if you're looking for this
51:08bubble to end because we're not going to
51:10get the financing, I think you should
51:11listen to the Goldman Sachs one and go
51:13through it. Here is the total number, $1
51:15trillion. quarter 1, quarter 2, quarter
51:183, and now we're into quarter 4. Okay,
51:20there's a couple things worth noting.
51:22The debt Q1, Q2, Q3. So, as rates went
51:26higher, it went down. Microsoft just
51:28announced a deal of 25 billion. Where is
51:31the transfer going? Okay, you're
51:33starting to see SPV, SPV.
51:36This is the reason that you have to use
51:40your brain. Don't listen to people just
51:42saying one thing. Go do the homework.
51:45show where it is. A trillion dollars is
51:47a big number to come and that's through
51:49now and again we have we have uh more to
51:53come uh in terms of anthropic as well.
51:57So 16 companies this is what they did
52:01this year that's how concentrated it is.
52:04So if you take you know the metas and
52:06you take the SpaceX and you take SpaceX
52:08was mainly through SPV includes the SPV
52:10but you're dealing with big numbers. it
52:12made it very very hard for people to to
52:15take them. Um, so the CDS has widened
52:17out. Here's the thing I want you to
52:18notice. I got to show you the Meta one.
52:20So Meta up here, just move this over
52:23here for now. Here's the CDS of Meta,
52:25Amazon, Google, Microsoft. So this is
52:26the hyperscaler. So this is what has
52:28everyone panicked. Meta peaked after
52:31their earnings came down and then went
52:33right back up. Uh, Google and Amazon are
52:35what drove everything back up. Here's
52:37Microsoft who hasn't is now borrowing 25
52:40billion because they haven't bought
52:42haven't borrowed anything. That's why
52:43their stuff hasn't gone up. This is
52:45concentration. This is not people
52:47binging the bubble. Do you think
52:48Microsoft isn't in trouble if there
52:50isn't a bubble? They're just in less
52:51trouble than the other groups. Here is
52:53the CDX for IG. So, it's following it,
52:57but again, we're not even as high as we
52:58were in March in the IG side.
53:03Um, I wanted to include this because
53:05this is Oracle and Oracle had its best
53:09week for its CDS since September of last
53:13year. Let me repeat that. They had their
53:16best week for CDS since September of
53:19last year.
53:21Um, and if you want to go through the
53:23reasons why,
53:26let's first focus on the Goldman Sachs
53:28thing. And remember, okay, capex in
53:312025,
53:3433, 2026,
53:36I'm sorry, 27%
53:39then 33%, then 35. So, they've been
53:42growing in terms of their percent of the
53:44IG market, and they think it's going to
53:46peak in 2027.
53:49So these are the reasons why capex
53:52deceleration could strengthen credit.
53:55Monetization determines the borrow
53:57needs. There is a delay between when
53:59they invest and when they can monetize
54:00the investment. So again this is the
54:03most important story and Gavin Baker has
54:06talked about this as well. Now he made
54:08the comment on TBPN. So, this is two
54:10months after the Goldman one. And the
54:12reason I waited to bring this up is
54:14because I wanted to see one of the
54:15companies, and it was Oracle that said
54:17it, that would say that they're not
54:19going to need capital going forward, or
54:21at least nearly as much. He went further
54:23to say, "We may be approaching a peak
54:25external financing for AI." Baker
54:27explained that crowd providers side
54:29contract prices substantially below
54:31today's rates. At those renew higher
54:33prices, cloud revenue and operating cash
54:35flow should accelerate. I don't know if
54:38we've seen the peak of it, but I know
54:39we're getting but I think we're getting
54:41close. So again, he said something.
54:43Here's the argument. This is if you
54:46don't understand it, go use the LLM
54:48subscribers and go figure it out. But
54:50this comes on a week when
54:53Neocloud's face guppy level death. Musk
54:56warns. Now again, for people who are
54:58like, "Yeah, he's calling they're going
54:59to be gone." He was raising 40 billion
55:01this week. So he also wants to make sure
55:03that he crowds out people. Oracle eases
55:07AI spending worries with plan to turn
55:09cash flow positive. Oracle started this
55:11whole thing. They were the first CDS to
55:13widen out. It happened in the fourth
55:15quarter of last year after their
55:16earnings report when they said how much
55:18money they were going to need. They
55:20pledged to raise no more than 20 billion
55:23in additional financing through 2030.
55:26They said they would be free cash flow
55:29positive in the second half of 2028 and
55:32it would reach 30 billion in 2029.
55:35So you can doubt what they're saying,
55:37but the main point is here are the free
55:39cash flow and then here is what these
55:42companies the capex outlook is. I saw
55:44someone talking about that the main
55:45thing is the capex outlook. Guys, I
55:47could not disagree more. Um these capex
55:50numbers have become ridiculous. whether
55:52they're 196 or 185 is completely in it's
55:55completely insane. The market may
55:58respond that way for a month, a day, a
56:01week. The main thing is the buildout is
56:03happening and it is massive. It is
56:06spreading beyond the hyperscalers which
56:08is the only numbers we talk about and
56:10you heard that from SKHEX. You're going
56:12to see it repeatedly. Here's Oracle CDS
56:14again came down sharply. It widened out
56:17on Friday but this was a huge move. It
56:20was the biggest move in the history of
56:21their CDS in terms of tightening. So I
56:24like it when the market responds to good
56:26news in a good way. It did with Oracle.
56:29Coreweave also despite the fact that it
56:33opened it opened on the lows of the
56:35week. The stock the debt was the same
56:40thing. So again, if you're looking for
56:43good news, you got Oracle. If you're
56:44looking for Coreweave, it handled bad
56:46news well and then it handled good news
56:48well. So maybe we've reached a point. So
56:51it gets back to anthropic. Again, they
56:54also said they're going to have enough
56:56capital, maybe it'll 2030. That's what
56:58David Saxs is saying that you could
57:00argue. And again, the reason is they're
57:01going to have cash. They're also going
57:03to have cash flow because their revenue
57:06is running ahead of the expenses.
57:08Guys, that's a very different story than
57:10the panic that had come in. That's why
57:11it's important to pay attention. We had
57:13wash
57:15higher rates are affecting AI financing.
57:17interest rates get into the cracks.
57:19Financial conditions have tightened.
57:21This made the possibility of the Fed
57:24raising stay right around the 75% mark.
57:28It was slightly higher. The reason I
57:29bring it up is I really do believe that
57:32um the tightening fears have reached a
57:35point where with rates at this level, we
57:38know that the market's doing tightening.
57:39And he's saying long-term rates are
57:41doing tightening. It doesn't mean they
57:42won't tighten in December. If the
57:45economy is strong, if stocks are
57:47rallying sharply, if credit all of a
57:48sudden gets better, it's fine. But it's
57:50not a guarantee that they're going to
57:52raise rates in December. Um, and part of
57:54the reason is this. He mentioned that
57:56certain parts of the market were being
57:58hurt and we were seeing issues.
58:02This is what's happened to Triple C
58:04debt. Now, again, the Fed raising rates
58:06had a part of this, but this is also the
58:07crowding out. So, as rates have gone
58:09higher, you've got an issue here. Now,
58:11this is spreads for triple C paper,
58:14which is now at the wide, it's wider
58:16than it was in 2022, which is what Bruce
58:20was talking about. Here it is in yield
58:22terms. I mean, it's 16% now. So, the
58:26only time we've seen yields this high
58:28for triple C paper is during crisis
58:30mode. That's why to me, we are very
58:35likely going to see a lot of
58:37bankruptcies in companies because I
58:38don't think yields are coming down much.
58:40But I do think with mortgage spread,
58:43mortgage rates also up here doing the
58:46exact same thing up a 100 basis points
58:49since April. But most importantly since
58:51Jackson Hole, we're up another 60.
58:55That's also right here when Scott
58:58Bessant intervened in the quarterly
59:01refunding announcement announced higher
59:03buybacks.
59:04So he did the yen intervention in July
59:08when the AI trade was kind of unraveling
59:10and the dollar yen was there. Then he
59:13made the buyback announcement. So I get
59:15back he does the I'm the house now
59:18speech. You really don't think he might
59:20do this? I think this could be a very
59:21very uh missed event that people have
59:26not focused on if he reduces the auction
59:29sizes for longerdated bonds or cancels
59:33both the 20-year
59:35and 30-year bonds to be funded by T bill
59:38issuance effectively a twist or whatever
59:42you want to call it. you call it yield a
59:44version of yield curve control whatever
59:46it is I with everything going on voters
59:50caring about affordability mortgage
59:52rates higher triple C market I I just
59:55think you got to be careful uh Microsoft
59:59insane announcement if you didn't see it
1:00:01offer to buy out 60,000 workers offer to
1:00:05buy out 27% of the employees
1:00:09unprecedented
1:00:11even No, not all of them are going to
1:00:14take it. If you're one of those 27, what
1:00:16do you think is coming down the road?
1:00:19Here is what the employee headcount,
1:00:21this goes for everyone who doesn't
1:00:22believe in ROIC and doesn't believe in
1:00:24productivity. Uh I've talked about this
1:00:27relentlessly. GDP is a horrible measure
1:00:29for AI. It has no way of measuring the
1:00:31economist versions of productivity. It
1:00:33is a stupid thing. And if you're using
1:00:34that to invest in stocks, you're going
1:00:35to pay the price. It always gets revised
1:00:38higher. It did with the internet. it in
1:00:40the 90s, then it happened in the 2010s.
1:00:43There is no way to measure it except for
1:00:45profit margins and things like this. If
1:00:47m if all 60,000 took it, we'd be back to
1:00:512020 levels of employees.
1:00:56The revenue for next year for Microsoft,
1:00:58here's where it was in 2020. This is
1:01:01where it will be at the end of next
1:01:02year, which would mean that this is the
1:01:05way the revenue versus the employee
1:01:07would look like.
1:01:09incredibly correlated. I work for Morgan
1:01:11Stanley. That's the way it is. You make
1:01:13a lot of money, then they give you money
1:01:15to go hire people. Here's what's
1:01:18happening now. Nobody's being hired.
1:01:19It's going higher. This is the tech
1:01:21companies. This is stage one of the AI
1:01:23productivity boom. It will start
1:01:25happening in the financials. It will
1:01:26start happening in every single company.
1:01:28And this becomes the labor verse capital
1:01:31issue, which again is going to start to
1:01:33become an issue next year as the agentic
1:01:36workflows finally get used. So here is
1:01:39their revenue per employee. These
1:01:41numbers are staggering. This is profit
1:01:43margins. This is what is happening. This
1:01:45is why you're fighting against this. But
1:01:47this is also why we get back into things
1:01:49like Bitcoin, things like the basement,
1:01:51things like all this stuff where the
1:01:53yield curve is an issue.
1:01:56I talked about this in the price of
1:01:58abundance this week where I connected
1:02:00France and what's happening there which
1:02:02is dominated by affordability issues
1:02:05with what's going to happen in labor
1:02:06versus capital in the US with these big
1:02:08companies. It won't go fast but it will
1:02:10happen. Eli Lily reported if you didn't
1:02:13see the numbers I just want to highlight
1:02:14this 39% year-over-year revenue growth
1:02:18for the quarter 24.4 billion. I mean
1:02:22these are insane numbers.
1:02:25gross margins.
1:02:28How large could the GLP1 market become?
1:02:31This is these are the reasons why they
1:02:34have an insane cash cow business. Think
1:02:36of it as the iPhone before they got
1:02:39heavily into the service business. Eli
1:02:42Liy is a dominant market player in a
1:02:45incredibly important market with 110
1:02:48million people in the US with obesity or
1:02:50overweight. The US penetration is 10%.
1:02:53The international penetration is only
1:02:55two and this is with injectables.
1:02:58This is what has happened in the price.
1:03:00This is what has happened in the volume
1:03:01of GLP1s. The price is falling. Here's
1:03:04how much it's fallen. Here much is
1:03:06volume. This is Jevans paradox. Again,
1:03:09this is the kind of thing you want to
1:03:10see. Revenue, revenue, revenue growing
1:03:13violently. But also Q3 sales for
1:03:16Alzheimer's, immunology, oncology,
1:03:19immuno.
1:03:21Look at these numbers, guys. This is
1:03:23from the earnings report. And by the
1:03:25way, if you didn't see this late last
1:03:26night, Friday night, Dave Ricks
1:03:29challenging
1:03:31big VC guy on longevity and where the
1:03:36drugs will come from. The big bio, he
1:03:38said it unlikely will come from the big
1:03:40bio companies. Dave Ricks, who's the CEO
1:03:42of Eli Liy, chimed in and said he'd take
1:03:45that bet. And then Elon Musk. Elon Musk.
1:03:49What better selling point for Eli Liy
1:03:51than come out and say has some banger
1:03:53coming out. All right, finish up with
1:03:56crypto. So, we get the Binance hack, the
1:03:59largest hack in history, largest heist
1:04:02in history,
1:04:04and here's what Bitcoin did for the
1:04:06week. It was down 4%.
1:04:08Here's what it did for the day. You
1:04:11can't even tell that something happened.
1:04:13Uh, again, this wasn't a hack of
1:04:15Bitcoin. I read some people that again
1:04:18don't spend any time understanding it,
1:04:20so I get it. Um, I also had someone on
1:04:23this hacking thing. This has been a
1:04:25growing fear of quantum. And I finally
1:04:27got to the cancer part of this. Ionic,
1:04:30uh, a company who's in quantum, has
1:04:32basically been out there in conferences
1:04:35just talking about that they've solved
1:04:37it and that they'll be there at the
1:04:39latest of 2028.
1:04:42Okay. So everyone is now believing this.
1:04:44So September 8th is when this was a
1:04:47public report.
1:04:50I asked um chat GPT I gave it all the
1:04:53information from Ionic. I gave all the
1:04:56information from Google, from IBM, from
1:04:58Jackader at Sandbox, Adam Back, Nick
1:05:01Carter, and basically had it go through
1:05:03and see whether Ionic was being a salesy
1:05:08we need to raise money thing or going
1:05:10through it and basically it said the
1:05:12chance that they'll get there by
1:05:13December of 2028, not December of 2027,
1:05:16is 15%. And again, I gave in all this
1:05:19information and asked it to use all of
1:05:20it based on what Google has said. Who's
1:05:22more likely to be right? Google with
1:05:24infinite money, IBM with infinite money,
1:05:26or Ionic, which then announced a month
1:05:30after their conferences that they were
1:05:32pricing a deal. So, um I think AI
1:05:37hackings are going around the globe and
1:05:39this is not just a crypto thing, any
1:05:41application. This is happening in Forex
1:05:44brokers in Japan. A big deal. Here is
1:05:47Bitcoin's technical side. So, I'm an
1:05:49Elliot wave person, as I've said, and I
1:05:51think we're at the beginning of a major
1:05:52move. So, we had 1 2 3 4 classic
1:05:56fivewave move. And now, where we stand
1:05:59is exactly where you want to be on a
1:06:01retracement side. We've effectively done
1:06:03a retracement of 38% into the wave four
1:06:08move. That is classic. We'll see if we
1:06:09go higher from here, but so far it looks
1:06:11like a classical Elliot wave move in
1:06:13terms of Bitcoin relative to the NDX,
1:06:16which is my favorite trade and the way
1:06:18that I am also playing this, here's the
1:06:21way you go. So, if you guys want to
1:06:24lever something up in per I say get long
1:06:27Bitcoin, get short the NDX.
1:06:30It's moved nicely, but more importantly,
1:06:33very quiet in the month of October to
1:06:35show you Bitcoin. I showed this in the
1:06:38video this week for both the technical
1:06:39analysis but also my crypto video. So
1:06:41you guys can go check those out. Um
1:06:43here's the overlay of SMH.
1:06:45[clears throat] This is through Monday.
1:06:47I didn't update it through the week.
1:06:49It's not going to be that much
1:06:50different. Uh Bitcoin has been trading
1:06:53since the basically the top of that
1:06:56fifth wave. It's become correlated to
1:06:58the SMH. But the other line in here is
1:07:01oil and tenure rates. Oil and tenure
1:07:03rates have had a huge problem on the
1:07:06market since the September highs and I
1:07:08think the midterms have too. I just
1:07:10think that people because Trump is
1:07:11saying he's going to go back to war with
1:07:13Iran because tenure rates kept going
1:07:15higher and then we got the deleveraging
1:07:17in the middle of the month which just
1:07:19added to people not taking risk and I
1:07:21think that's been a major I uploaded the
1:07:24um the deck uh so the the index was down
1:07:28for the month 16 and a half% Bitcoin was
1:07:30down seven so it did get correlated to
1:07:32the tech trade again it's had a very big
1:07:35bounce and it's still up 25%. Uh this
1:07:37week was about tokenization in RWA
1:07:39Rails. Uh I want a reminder because I
1:07:42know a lot of crypto people have joined
1:07:43in the last couple weeks. You guys have
1:07:45to use the prompts especially when it
1:07:47comes to taking what I just showed you.
1:07:49If you take the names that I showed for
1:07:52the week for tokenization and RWA. I did
1:07:54two of them this week, but I'm just
1:07:56showing you this one. So Ando Centrifuge
1:07:58Figure and Providence. If you take this
1:08:01prompt, upload that screen. That's all
1:08:03you have to do. Just upload this to an
1:08:06LLM, then copy paste the prompt. You get
1:08:09a long, long, deep research on the
1:08:12names, okay, on the space. So, it will
1:08:14give you an executive read. You have
1:08:16many, many pages. I think this one ended
1:08:18up being 30, 40 pages. This way, you can
1:08:21do your deep dive. I said to do the
1:08:23Peter Lynch side. You can go through,
1:08:24you can learn about each one, how
1:08:25they're different, what's going on. That
1:08:27way, you can make your own decision on
1:08:29which ones you want. It goes through
1:08:30everything that way. Final slides.
1:08:32Google just announced they're developing
1:08:33a Bitcoin and crypto custody service for
1:08:35institutions.
1:08:37Tokenization continues. Salana's RWA
1:08:39real world assets hit a new all-time
1:08:42high. This was a big news item, too. The
1:08:45DTCC tokenization service is now live
1:08:48and this is only the beginning. That's
1:08:50it for this week, guys. Uh, a lot in
1:08:53there, a lot going on. Again, think
1:08:56market structure has changed. Home runs
1:08:58are going to be the way. If you get
1:09:00yourself out of the market for fears of
1:09:02a bubble, if you're wrong, you're
1:09:04probably going to be getting in at the
1:09:05wrong price on the way back up. It makes
1:09:07it very challenging. So, when the
1:09:09fireworks show, when you've made a lot
1:09:10of money, I think the rotation and the
1:09:12movement has to happen. This market has
1:09:14been about rotation, rotation, rotation.
1:09:17And I think a lot of that has to do with
1:09:18the instability that has happened with
1:09:20the hyperscalers. And I think that's
1:09:21going to remain. So, we'll leave it
1:09:22there. I'll see you guys next week.

About this transcript

This page contains the full transcript of The New Market Structure: AI Concentration and the Largest Hedge Fund Degrossing in 10 Years by Jordi Visser, generated from the public captions YouTube serves with the video. The transcript has 13,005 words across 1,833 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.