YT YouTube2Text

YouTube2Text → Transcripts → Momentum Is Crashing, Bitcoin Is Bottoming, AI Agents Are Rising

Momentum Is Crashing, Bitcoin Is Bottoming, AI Agents Are Rising — Transcript

by Jordi Visser · 9,238 words · 1,335 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00back in Maine. Uh 90 plus degrees for
0:04one of the two times a year. Uh holiday
0:07week. Happy 4th to everyone. Hope you
0:09got to enjoy it. I did this on Friday,
0:13so uh I'm seeing you guys Sunday. Uh
0:17midcycle slowdown continues within AI
0:20and as expected the narratives and the
0:23headwinds and what everyone's talking
0:25about. Uh I spent a lot of my time this
0:27week uh talking to people about
0:31particular
0:32uh podcasts expost. Again we are in a uh
0:37a point now where people are getting
0:38worried because it's not as easy to make
0:40money. So I'll go through that. A big
0:42volatility shakeout. Momentum is just um
0:45killing people with the volatility which
0:48I've highlighted. I'll go through
0:49podcast. Greg Brockman had a great
0:51podcast this week. the consolidation
0:54playbook uh for subscribers. Kevin Worsh
0:57had an interview. I think um people need
0:59to start focusing on that. Some of what
1:01I'll be going through is going to be
1:04focused on Worsh. Um the application
1:08layer, insurance and health care seem to
1:10be the two sectors right now benefiting
1:12the most as the consolidation is
1:14occurring with inside the
1:15infrastructure.
1:17Meta, another big uh week for them in
1:21terms of trying to get the market all
1:23enthused. Uh open source continues to go
1:28and do I say Bitcoin is actually
1:31giving me enough to talk about from a
1:33potential bottom. We're still below the
1:35200 day moving average, but I'm spending
1:37a lot of time on this because I believe
1:39the application layer matters. I believe
1:41wash matters for Bitcoin. We'll go
1:44through it. Uh, I just want to start
1:47with this equal weight S&P. If you're
1:49listening to anyone saying that the
1:52market is a bubble or that a correction
1:54is coming, equal weight S&P is breaking
1:58out and is strong despite all the
2:01volatility. There is a rotation going
2:03on. IWM made new all-time highs this
2:06week. The S&P is in a consolidation
2:09triangle at this point. So, for everyone
2:11who sends me stuff getting all beared
2:13up, got to short tech, got to short the
2:15AI trade. We're going to see capex cut.
2:18Everything is going on. Um, the S&P
2:21spent the month of June consolidating.
2:23Uh, it's basically the same level it was
2:26since the beginning of May, which means
2:29we're taking some air out of the bubble.
2:30While it's an earnings driven recovery
2:33and bull market cues, same triangle
2:35formation.
2:37The New York Stock Exchange cumulative
2:40breath all-time high this week on
2:42Friday, Thursday. Hyperscalers continue
2:46to stay in a range they've been in. They
2:48are unchanged almost effectively since
2:50last year. Again, this is Meta, Amazon,
2:54Microsoft, and and Google. Uh it's
2:57multiple compression. Their earnings are
2:59growing. Everyone keeps trying to pick
3:01the bottom of them. Everyone keeps
3:02trying. I I think these again are just
3:06like software has become and the
3:08aggregate for SAS. This is just going
3:11through multiple compression. Every now
3:13and then two of them will do well, two
3:14of them won't do well. I still like them
3:17as the funding side for this, but
3:20I don't I I I think we've gone through
3:22enough at this point where the V so
3:25high. This is my thematic portfolio.
3:28We've been consolidating like the market
3:29has. just again shows that the AI
3:31infrastructure trade has been doing
3:34everything. Now, for any of the
3:36institutional clients that I've met with
3:38over the past couple weeks, when asked
3:39the question, well, what kind of
3:41consolidation? I keep saying the same
3:43thing. I think a 50-day moving average
3:46uh retracement either through the 50-day
3:49moving average moving up or prices
3:51coming down or a combination of both
3:53would be good. This is the Cosby uh two
3:56uh 200. So again, this is weaning out
4:00all of the people on margin. There's a
4:02bunch of ways to lose money when people
4:04are gambling. And one of the ways is
4:06just by the market stopping to go up and
4:08then gradually they they lose their
4:10attention. It doesn't have to collapse
4:12in a bubble. Uh this is SKH touching the
4:1450-day and then rallying uh on the day
4:16off. Uh this is the financial side. I'm
4:20going to bring this up as I go through
4:21in the US, but the financial side in the
4:24Cosby uh has been consolidating. If this
4:26is a GDP driven market, if GDP around
4:30the globe is improving, as we're seeing
4:31with estimate revisions, you should
4:34start seeing the financials be part of
4:36the bull market. They were lagging this
4:38entire year, especially in the US on the
4:41back of the blue owl situation, and this
4:43was impacted. Well, now we've
4:44consolidated. This is another
4:45consolidation chart, which looks like
4:48it's going to break higher. So, we'll
4:50see if financials become a space people
4:52want. the megaphones. Um, we got one in
4:56the socks rotation. So, this just again
5:00shows the massive increase in
5:02volatility. So, this is the same sort of
5:04the same thing as momentum, but it's
5:06just a rotation between the
5:07semi-software or the semiconductor ETF
5:10and the software ETF and just how wide
5:12the ranges are going. This is a way to
5:14get hedge funds out of the trade. It's a
5:16way to get uh retail momentum traders
5:19out of the way. They don't want to be
5:21looking in something that keeps costing
5:22them money when they keep buying a
5:24breakout. For the hedge funds, they also
5:26get stopped out or if V goes higher, the
5:28risk managers tapping them on. You can't
5:30be long this much VAR. This is the way
5:33to get rid of things. And these
5:34megaphones, this is the Morgan Stanley
5:36momentum uh index. And you can see this
5:38is all the way back to 2010. Just how
5:40wide it is and just how many times we're
5:43trading outside of this 5% band. These
5:45things are just moving uh every day.
5:48This is TMT. So this is the tech side
5:51and this goes back to 2010. You can just
5:53see how wide these have gotten. I mean
5:55this is just dramatic and I would say
5:58beginning from when the opus uh opus 4.6
6:01kicked in and people or 4.5 which is
6:04back here we've seen more and more
6:06people going. This is a sign that it's
6:08just crowded. There's just too many
6:09people in the trade and this is the
6:11volatility of that uh M tech Morgan
6:15Stanley Mek. I mean this is 5% a day.
6:18This is higher than the.com bubble. Um,
6:21this has just gotten to the point where
6:25it's flushing people out. Now, this
6:28month has over the last five years has
6:30just been horrible for momentum. So,
6:33we're off to a horrible start. Uh, so
6:35far in 2026, momentum's gotten crushed,
6:37but you can see that this has been a bad
6:39month, just like December has been a bad
6:41month for momentum. So, the seasonals
6:43are in there. Um, I happen to subscribe
6:46to this as well, not in terms of market
6:49meltdowns, but I think the more and more
6:50that individual investors, particularly
6:52out of Asia, are trading the markets
6:54using AI agents, AI agents are going to
6:56continue to push this stuff. Those
6:58charts are bullish. They're not bearish.
7:00The only people getting hurt are the
7:02people that have draw down limits like
7:04hedge fund managers that have
7:07monthto-month risk for liquidity
7:09purpose, which goes back to hedge funds.
7:11But retail traders and people that are
7:13there, they're trying to make money and
7:15the VAR is not part of their job in
7:18this. So they might get chopped up a
7:20little bit, but I think the people that
7:22are more likely to get chopped up are
7:23the ones that are worried about the
7:24marktomark perform mark tomarket
7:26performance where they're borrowing
7:28money from people. So, I think the
7:30better way to put this is we're going to
7:32see more volval, and I've been saying
7:34that for a couple years, but I think as
7:36the agents get involved, their number
7:37one goal is to make money, not to meet a
7:39certain sharp ratio. Uh, I just want to
7:42bring this one up because I've I've
7:44repeatedly seen people show this chart,
7:47which is the margin debt, and say that
7:48this is a bubble. And I'm just sick of
7:51watching it. It's just such lame
7:54analysis to sit here and say that debt
7:56is getting bigger without looking at the
7:59size here. Just go look at the debt of
8:01the housing market. If the I mean we get
8:03into the point that if the value of the
8:05houses are going up well then the debt's
8:07going to go up. This is it relative to
8:08the market cap of the US which is a
8:10better way to look at it. Now it is
8:11rising right now but again we're below
8:14the midpoint of the range over the since
8:17the great financial crisis. So I don't
8:18see debt as being a big story. So, let's
8:20get back to what I showed last week, the
8:22different types of consolidations. And
8:24like I said, this would be great if the
8:26AI infrastructure trade went up, you
8:29know, 20% and then you consolidated in a
8:32tight range. The consolidation is going
8:34to be somewhat representative of two
8:36things in my opinion. One is how much of
8:38it was fueled uh or how much what were
8:41the earnings growth during the time
8:42period and is this something that uh in
8:46price movements was big? This is
8:48basically something where I would expect
8:50a very good sharp ratio with something
8:53that's a little more boring. This is
8:56when it's more speculation driven. I
8:57don't believe the AI trade is
8:58speculation driven. I believe much most
9:00of it is earnings. There is some
9:02speculation and that leads to this
9:04action and that's where I think we are.
9:06We are resetting now the thought process
9:09on this and these are all real and I've
9:12talked about all of them every week.
9:13It's the reason why I go through the
9:14headwinds, not because they're going to
9:17stop the AI trade. Headwinds don't stop
9:19things. Planes don't hit a headwind and
9:22then go straight down. You get
9:23turbulence, and that's what we're going
9:25through at this point. So whether it's
9:27the data centers, the power and cooling
9:28side, which is causing some of the data
9:30center side, the memory bottlenecks, the
9:32regulation, all of these things in here,
9:34of which there's many more, too. They're
9:37all part of the headwinds that are now
9:39becoming the bigger story. The reason
9:41they matter more now is because we've
9:44gone through the earnings. This is no
9:45longer as big a surprise and people are
9:47crowded into the trade. There's no way
9:48to refute the fact that they're crowded
9:50into it. Um I thought this week's all-in
9:53with Gavin Baker covered a lot of the
9:55headwinds and some of the topics uh
9:57fairly well. Um they also went through
9:59the memory side. Uh they talked a lot
10:01about GLM 5.2. I do think you should
10:04listen to this one on it. Um Gavin Baker
10:07says the future is composable models.
10:09This fits in very importantly with GLM
10:115.2. You need to have your mindset on
10:14this. It is silly to think that it's all
10:16ornone, meaning, okay, I'm going to use
10:18open source. I'm not going to use these,
10:19but that's never going to happen. The
10:21best models still have been in the
10:23frontier side. They've still been in the
10:24US.
10:26Think of it as those are the IV leagues
10:29or the highest IQs out there. um the
10:33companies that have the most money,
10:36which spend the most money, which drive
10:37the economy, and I'll say it this way,
10:40in the Russell 2000, the total market
10:42cap is about three and a quarter to
10:45three and a half trillion dollars. Well,
10:47that's one company at the top of the
10:50S&P. Pick your pick. So, one company
10:53equals 2,000. So, if 2,000 people use
10:55open source that are on the smaller
10:57side, but one company use Okay. Well, if
11:00all the big enterprises do, the S&P is a
11:02massive side. So, I do think over time
11:04they'll be there. But don't fall into
11:06the trap of people saying that no one's
11:07going to use the higher models and
11:10anthropics revenues are going to stop
11:11growing. I think that is just again
11:13silly talk. Um, and I'll get into the
11:16composable models later. And then they
11:18go through the AI regulation. And then
11:20the final section, they cover the
11:22modular compute and a whole bunch of
11:24other things which again get into where
11:25we're going on this. Uh, on the GLM 5.2,
11:28two, they put it as a serious openweight
11:31challenge to US frontiers. And for any
11:34small businesses, anyone who's
11:36sophisticated enough to be able to use
11:38open source,
11:40it's going to happen. I've been using an
11:41open source Chinese model on one of my
11:44setups since uh April. So again, this is
11:48going to happen. I don't have enough
11:50hardware right now to deal with GLM 5.2
11:52too in terms of on my own computer. But
11:56go I thought this was the most important
11:58line. Gavin Baker says the model's
11:59performance on GLM 5.2 the open source
12:01forced him to rethink part of his
12:03framework. I do think how good GLM 5.2
12:06has challenged some of my beliefs. He
12:08talks about it. Those beliefs have to do
12:10number one with could open source keep
12:12up. I talked about that with Leopold Ash
12:14uh Leopold Ashbrunner um who also got
12:18that wrong in his um his situational
12:20awareness. Uh, but this is the other
12:23part. The future is composable models.
12:25And I think we're all starting to
12:26realize that to keep the cost down
12:29because no one expected the cost of
12:30these things to go up so dramatically
12:33that you're going to have to make sure
12:34that the smarter models are making the
12:37decisions. The smarter models are being
12:39used on the most important decisions in
12:43a company and certainly the most
12:44important things like drug discovery,
12:48science, energy, all of that stuff. So
12:50you're still going to need the models.
12:52You're still going to have people paying
12:52a lot of money for them. We still have
12:54scarcity on them. Uh but I think the
12:57open source is going to take up most of
12:59the let's say workflow jobs. So think
13:02anything that you're using co-pilot for.
13:04Copilot is not this. So think of open
13:06source as being behind copilot. And
13:08don't remember don't forget I
13:09highlighted that Microsoft was talking
13:11about deepse possibly for for that. So
13:14the frontier I left that up.
13:17Frontier tokens are capturing 90% of the
13:20economic value and open source tokens
13:22are probably 80% plus of tokens
13:24processed. That's the thing that
13:26matters. Everyone is starting to look at
13:28this and going this is going to kill
13:30anthropic tokens are capturing 90% of
13:33the econ economic value. So I don't
13:35agree with the people that are getting
13:37bearish on this side and trying to make
13:39some kind of cracks in the AI which I
13:41heard repeatedly in podcasts over the
13:44last three days. Um, I I didn't
13:46highlight this last week. It came out
13:48the the week last week when I did the
13:52video, but I think Dean Ball, who was
13:54hired, a former Trump official on the AI
13:56side, was hired by OpenAI. I think the
13:58paper that he wrote is kind of important
14:01what should be done. I I'm not going to
14:03go through a lot of the details on it,
14:06but he's just saying that the government
14:08regulation um it creates a dangerous
14:11limbo where these models are not being
14:13used and it's allowing the open source
14:16models to catch up. It's creating other
14:18issues. It's allowing countries around
14:19the world to think that maybe we
14:21shouldn't be using this. So, he's just
14:23going through this as a risk. Again, a
14:25headwind that I think is important. The
14:28administration and the states have not
14:29figured out how to deal with this. We
14:31still have a midterm election coming up.
14:32Again, I think these are headwinds. I
14:34don't think these are stoppages. Same
14:36thing with the data center, the
14:37backlash, front page, an economist
14:40article
14:42last week. JP Morgan put out that more
14:45likely to have a cyber attack causing a
14:48banking crisis than a credit loss. I've
14:50talked about this before. Agent swarms
14:52are bigger risk to the fiat system than
14:54quantum is to Bitcoin. Book it. Remember
14:58it. Stay with it. Um, this is a memory
15:01wrist that came out and if you didn't
15:03see this, it definitely got pushed
15:05around. It's starting to feel a little
15:07like the crypto market always in the
15:08equity market where there's this pump
15:10and dump thing where people are just
15:11posting stuff. Now, Andrew Curran is a
15:13very respectable um person with inside
15:16the tech and AI community and I
15:18highlighted a paper he wrote last week
15:20or the week before which was excellent,
15:22but he put this out which is probably
15:23true. There has been a significant
15:25breakthrough in architecture,
15:26specifically around memory efficiency,
15:28not by one of the big labs, but a team
15:30that was spun out. 2.1 million views.
15:33So, it was floating around, I think.
15:35Let's assume it's true. Um, don't go
15:38sell your memory stocks because of this.
15:41But as I've said since the day that I
15:44got out of my micron and said, "Hey, I
15:47just don't think these things can
15:49continue." If you're lucky enough ever
15:51in life to make a 7 to 10 bagger, don't
15:54expect there to be that much more going
15:56forward. Maybe there's a double over the
15:57next three years. Maybe there's uh 200%
16:00over the next three years. It isn't
16:02going to be like what it was. And I just
16:04thought there were better places where I
16:05still could get triples on the same
16:07theme. So for people that are long
16:09memory, I wrote a paper on it this week.
16:11I still believe memory is is a scarce
16:13issue. And even if they've come up with
16:15something, the question is until we get
16:18details on how fast they can get it up,
16:20how fast this can be brought through,
16:22whether it works for everything. We
16:24don't have any details. So, I would not
16:25spend a lot of time on it. Okay, those
16:28were the headwinds. Now, let's get into
16:29the podcast for you guys to listen to on
16:31where we are in AI and stop listening to
16:34the noise. Um, the future of software
16:37interview with Scott Woo, who's the
16:40co-founder and CEO of Cognition AI
16:45believes most people underestimate AI
16:47because they pattern match from history
16:48instead of reasoning from first
16:50principles. He highlights that AI has
16:51gone from handling seconds of work to
16:53hours of work and asks why that cannot
16:55become days, week, months or eventually
16:57a year of uninterrupted work. The most
16:59interesting future agents are not just
17:00tax task executors, but mission agents.
17:05I I'm I'm bringing that up because
17:07that's getting back into something I
17:09talked about last week, which I'll show
17:11in the next slide. Um
17:14uh AI may be one of those rare periods
17:17where pattern matching fails because the
17:18underlying variable is improving
17:20exponentially, which I completely agree
17:22with. Um the key constraint he focuses
17:24on how long a can AI can work without
17:27human interruption. We are already
17:29moving into a world where agents can
17:31perform hours of work. And I just want
17:33you to remember this hours of work. That
17:36is a huge change. Once an AI can take a
17:39well-escribed task, work through
17:41multiple steps, debug errors, search for
17:42context, test its output, and come back
17:44with a useful result, it starts to look
17:46less like a tool and more like a junior
17:48coworker.
17:50Combine what Scott Woo said in this part
17:53of the interview with what
18:04trying to give you guys more.
18:06Oh, with the Boris uh interview I showed
18:09last week. Boris Churnney uh I showed an
18:12interview on loops last week. So
18:16the story is the same in both
18:17interviews. AI is moving up the
18:19abstraction stack from writing code to
18:21completing tasks and running loops and
18:22eventually owning missions. Scott Woo
18:24describes the philosophical arc. Boris
18:27Churnney describes the operating reality
18:29inside Claude code. The industry has
18:31already moved from humans writing source
18:33code to agents writing code to agents
18:35prompting other agents. His concept of
18:37loops is the operational version. The
18:40unit of work is expanding.
18:44That leads into the Greg Brockman
18:46interview. If you haven't listened to
18:47Greg Brockman, he was interviewed with
18:49Alex Caner Witz on his podcast. Uh I
18:52think it's called the Yeah, it's the big
18:53technology podcast. Definitely worth the
18:5640inute listen. Um
19:00it's evolving from a chatbot into
19:02something closer to a unified super app.
19:05Rapid adoption inside open AI. Tools
19:07like Codex are seeing explosive internal
19:09usage. um a so the broader ecosystem
19:14likely to embrace AI agents rather than
19:16resist them. Now I've said this before I
19:18use codeex more than claude code. I use
19:21codeex more now than chatpt
19:25than I had in the past. So every day
19:27codeex is taking over more of my more
19:29and more of my work. So if you guys have
19:31been uh stuck with clawed code and using
19:33it it's great and you're used to it. I
19:36highly recommend using both of them. Um
19:38I use still use all of them but my the
19:40amount of time as I've mentioned I use
19:42on Gemini and on uh Grock has gone down
19:46significantly. Same thing with
19:47perplexity.
19:49Uh compute is the bottlenecks. Demand
19:52for AI will exceed supply.
19:56Access to compute infrastructure will be
19:58decisive. He basically went into whoever
20:01has the most compute wins. So you can
20:04sit there and doubt whether the compute
20:05is necessary. You can somehow say the
20:07capex is going to be cancelled, whatever
20:09you want. The reality is every single
20:13person who's at the heart of AI and is
20:15the ones seeing the models get better,
20:17Greg Brockman and uh anyone at Claude,
20:20but Dario in particular, it never
20:22changes. Uh AI and healthcare is already
20:25impactful and expected to become
20:27mainstream. I highly recommend listening
20:30to the final seven minutes of the
20:32interview. For those of you who do not
20:35agree with what I have said or have
20:38family members with cancer, they go
20:40through more stories in there of people
20:43that have used LLM to diagnose what's
20:46going on. And
20:48there just unbelievable stories. And
20:51when he was asked, is this something
20:52that's going to become the standard? And
20:54he said absolutely it will be the
20:55standard. Um here are the key insights.
21:00Enterprise adoption we are at the
21:01inflection point. This is critical guys
21:03to the application layer. This is
21:04critical to the profit margins. This is
21:07actually critical to the AI
21:08infrastructure trade. And in my opinion,
21:10this is also critical to Bitcoin. The
21:12phase shift is happening now. Now it
21:16didn't happen before. So phase one to to
21:18early 25, we need AI. We're falling
21:20behind. Now it's show me the ROI. So now
21:23they're bringing in these companies to
21:24do it. The constraint is supply not
21:26demand. Can we supply enough compute and
21:29energy? We don't have it. And companies
21:32are not adopting AI tools. They are
21:34adopting AI workers. This shifts the
21:37model to labor substitution,
21:38productivity leverage, organizational
21:40redesign. It's a red a redefinition of
21:43how it work. Productivity leverage. So I
21:45want you guys to think about when I get
21:47into the Kevin Worsh part as well. Just
21:49remember what Greg Mochman started.
21:52These are the quotes. There's not just
21:54not going to be enough compute in the
21:55world to satisfy all the demand. Don't
21:58know how he can make it any more clear.
21:59And he is at the front of the best
22:02models in the world. There's not there's
22:05just not going to be enough comput in
22:07the world to satisfy all the demand.
22:09Every provider sells out all their
22:11compute. That's the world we're heading
22:13towards. Even at tens of millions of
22:15agent users, just look at that number.
22:18Tens of millions.
22:21We haven't brought this to the planet
22:23yet and that's for agents. Not enough
22:26compute, guys. It can't happen.
22:28Regardless of what you listen to and
22:30hear, what you read in X, there will
22:33never be enough compute. Never, ever,
22:36ever, ever. Just like there's never
22:38enough intelligence necessary to solve
22:41the world's problems right now. So, you
22:43can sit there, you can argue, you can
22:44try to go through it, but I think it is
22:45a mistake to get into it. There will be
22:48capex air pockets. There will be time
22:50where bottlenecks are happening and I
22:52believe we're in one of those now or
22:53where the narrative gets too far or the
22:55prices get too far and we're going to
22:56get a pullback. But those pullbacks,
22:58those consolidations, those are
23:00opportunities for you to get involved.
23:04Customers are now asking how do we
23:05control spend? How do we observe what
23:06the system is doing? That's the
23:08enterprise adoption. We're investing
23:10heavily in enterprise readiness and the
23:11tools for customer. This is where it's
23:12going to start to see productivity,
23:15the size of the market. None of us were
23:16anticipating how steep that curve is
23:18going to be. This is a vast
23:21transformation far bigger than I think
23:23most people realize today.
23:26Okay. So, we're in a consolidation.
23:28We're in a narrative driven point. I
23:30think it's going to continue. You these
23:32things are going to break out eventually
23:33during the course of this year. All of
23:34them in my opinion. Um or at least the
23:36majority of them. I've talked about I've
23:38shown you where you can use the stuff
23:41that I'm giving you every weekend to
23:42figure out where the trends are. Over
23:4585% of the names are above the 200 day.
23:47Over 80% are above the 50-day. The 50
23:49days are point or pointed upward. Sorry,
23:52the slopes are pointing up. You get all
23:54of these healthy things. So with inside
23:56of consolidation, that's why I created
23:57this AI consolidation playbook. So had a
24:00lot of people reach out. Where was the
24:02consolidation? I put it back in again
24:04this week. Um remember for the pullbacks
24:07and everything there. This gives you I
24:09gave you a prompt for the pullback
24:11hunter. It'll come up with a list of the
24:13names that are the best ones based on
24:16the sheets that you have. Here's part
24:18two of the AI skill architecture went up
24:21on the site this weekend. That way for
24:23the hedge fund analyst skill, you can go
24:26build your own. So for all of you that
24:28have done the knowledge brain that are
24:29feeling more comfortable on either
24:31claude or codeex, you're going to be
24:33able to just upload those and do them.
24:35Now for the sign of just whether you
24:38should be worried, again, I'm going to
24:39say I'll show this again and again. uh
24:41not every week but at least every month
24:43when revisions start turning negative
24:46when half of the companies all of a
24:49sudden are getting negative revisions
24:51and nobody's getting upward revisions or
24:53if no one gets an upward revision and 20
24:56companies get a downward reision re
24:57revision you're going to start seeing
24:58some red bars right now we are still
25:02great this is the US estimate revisions
25:05again S&P peaks here revisions start
25:08going down red starts showing up. That's
25:12what needs to happen.
25:15Here's Europe. Europe just posted its
25:18biggest one in the last two years. Um
25:22Europe was suffering. They didn't have
25:24it. But look at this. Going higher.
25:26Starting to see the benefit in Europe.
25:28And on the global basis again, sitting
25:31near the highs in just every week is
25:34just strong. Um payrolls this week. Uh
25:37again, I I I I read the payroll. I go
25:40through this. I'm so bored of people
25:43just not taking this. Especially with AI
25:45guys, the aggregate weekly payrolls is
25:47the most important thing. Weekly hours
25:50are not turning higher. They are sitting
25:52below the levels that they were from the
25:54period after the great financial crisis
25:56into COVID. Ours are not expanding.
25:59We're having a nominal GDP acceleration.
26:01I listen to a lot of macro people, which
26:04I respect, that are saying wage
26:05inflation is going higher, the jobs
26:07market is tight, blah blah blah. None of
26:09that's true yet. Could it get there? I
26:12guess. I don't think it will. But one
26:14thing is certain. If we're at the point
26:16of AI agents, I wouldn't be getting too
26:18optimistic. Particularly when again
26:19these jobs are being created in health
26:21care. This is health care jobs,
26:23healthcare jobs, healthcare jobs. It is
26:25dominating the monthly numbers. This is
26:27the aggregate aggregate uh payrolls
26:30which takes the weekly hours, the wages
26:32and the number of people hired. And we
26:34are still sitting at very very boring
26:38levels. The same levels that were going
26:40on past the great financial crisis. So
26:43there's no acceleration here as the AI
26:46boom is happening. Here it is on a
26:48rolling three-month basis of the
26:49aggregate. So again, we're just around
26:52the same level we've been during okay
26:54times. There's nothing great going on.
26:56There's nothing horrible going on. All
26:58right. To Worsh, if you can get a
26:59chance, you can go into YouTube. You can
27:01see him interviewed by Sarah Isen from
27:03CNBC.
27:04Uh Kevin War spoke at the ECB forum. So
27:08let's go through what he said. We've
27:09seen this rejected forward guidance.
27:11You're just going to see a lot less
27:12information from the Fed. It's just
27:14funny that the first press conferences
27:16he does, he's hawkish
27:18and now we're stuck in this limbo where
27:21we have no idea. But he did speak. The
27:23Fed should not spoon feed markets. If
27:25people thought the Fed would tolerate
27:27inflation above 2%, they would be
27:29disappointed. Of course, he has to say
27:31this. This isn't a He doesn't want
27:33inflation higher. I don't know why this
27:35was a a big story in some hawkish thing
27:38or people that talk about it. The Fed
27:40independence. That's obviously important
27:42with what went on the last couple years.
27:44This is an important part on AI and
27:47productivity. He said the improvement in
27:48AI models is moving at an exponential
27:50level and called it hyper Moore's law
27:53stuff. He suggested that current
27:54business surveys may be underestimating
27:57how quickly AI adoption will show up in
27:59the economy.
28:00Again, this means he is in he's on top
28:04of this. He follows AI. He gets AI. I I
28:08really would be listening that if he
28:10believes AI is about to do this, how
28:13quickly AI adoption will show up in the
28:14economy, why would he raise rates? And
28:18I'm going to go through this whole thing
28:19because you had banks saying three times
28:22as of I think Bank America said three
28:24times. I could not disagree more and I
28:26don't even understand where it came
28:27from. And I'll go through again why this
28:29makes no sense to me at with uh
28:31inflation and wages. But regardless of
28:34what you believe will happen if he's a
28:37pro- AI person, why would he raise
28:39rates? I just don't even get the sign.
28:41Why would he raise rates to get ahead of
28:44something when he believes deflationary
28:46pressures and job impacts? Those are the
28:49two things that AI brings. Deflationary
28:51pressures and at the same time jobs
28:53disruption. That's what it brings.
28:55There's no question about that. If he
28:57believes we're entering into that, this
28:59is gets back to the Allen Greenspan
29:00thing. We are in the first or second
29:02inning of this revolution. We haven't
29:04even started, guys. He expects jobs to
29:06be greater and prosperity to be stronger
29:08over time though the timing matters for
29:11the Fed's dual mandate. So he's
29:13acknowledging the fact that and again
29:16what they have to focus on is inflation
29:19and wages and what's going to go if we
29:21have higher inflation but the jobs
29:23market is weaker because the
29:25displacement is happening before the
29:27rehiring gets he's got a dual mandate.
29:30He's got to focus on it. So both of them
29:32have to be moving in the same direction.
29:35sees AI as a major paradigm shift for
29:38central banking itself. Okay, again I'm
29:42favorite economic indicator. He said his
29:44hope is that in 9 to 12 months the Fed
29:46will be using new technologies to
29:48understand the real economy in a much
29:50more contemporaneous real-time way. It's
29:53very important stuff.
29:55Uh Fed is going back to first
29:56principles. Nothing new there except for
29:59the fact that central banks got used to
30:01suppressing volatility. uh higher VA
30:04higher letting things you know if the
30:06market corrects 10% not stepping in and
30:08doing something as long as it's not
30:09something big um I mean honestly they
30:12haven't really had to do that uh aside
30:15from SVB uh but we'll go see what's
30:17going on um task forces are not meant to
30:20prejudge outcomes this is on the task
30:22fund side but it's meant to bring the
30:24best minds in he wants interest rates to
30:26be the monet he wants monetary policy to
30:28be the tool because he wants the balance
30:30sheet to be reduced used. He made the
30:33point that it took 18 years to get into
30:35the current large current state of the
30:36large balance sheet and that it would
30:38take more than 18 weeks to get out of
30:39it. So, he's just telling you he's going
30:41to do it when he can. Um, he was upbeat
30:43on US growth. Worsh came across as a
30:46hawk hawkish reformer, but not in a
30:49simplistic rates now way.
30:53It needs to stop acting like it's still
30:55in the post208 crisis and overguiding
30:57markets. So, that's his thing. Now the
31:00reason I want to bring this up is
31:01because one of the stories and I think
31:02one of the the the important stories has
31:05been real two-year rates. Now this is
31:07real two-year rates inverted and I
31:09highlighted here because this is the end
31:11of March. Now on here are Bitcoin, gold,
31:15uh silver, the dollar and then this blue
31:18line here is inflation expectations
31:21inverted. So at some point this white
31:25line started to impact the debasement
31:28trade.
31:29And what started happening is, oh, okay,
31:31the Fed's going to raise rates. And this
31:32is basically when because inflation has
31:36come down and rates stayed the same and
31:38actually went higher in terms of
31:39expectations what the Fed would do, all
31:41of these trades started to sell off
31:43together. Um, the dollar was rallying.
31:45So that's inverted. but between the
31:47dollar, gold, silver, and bitcoin.
31:51And yet inflation
31:53expectations on the swap for two years
31:57are now at the lowest level of the year.
32:01Yeah, I'm going to guess that this is
32:03all based on belief that the Fed is
32:05going to do something. Now,
32:08this is the chart of rate cuts out to
32:11the end of this year. So we've gone to
32:14where the December
32:16uh end of this year what has happened
32:19into the expectations. Well, now we're
32:21into the hiking side. So I wanted to
32:23just show where we broke this and
32:25especially here. This is exactly where
32:27it was. This is the May period where up
32:29the Fed's going to hike. That had an
32:31impact on things. And so if people are
32:33right and the Fed's going to hike, I
32:35don't think that's going to happen. But
32:36if we get three hikes, I don't think you
32:37want to be near gold. I don't think you
32:39want to be near silver. And I don't
32:40think you want to be near Bitcoin. I
32:41don't think the Fed is going to raise
32:43rates. So, if that message is going to
32:45be delivered somehow or if the market is
32:47now positioned for it, which is what
32:49we're seeing because we've got crowded
32:50positions and all of those trades the
32:52opposite way now, meaning they're all
32:54playing for the Fed to hike rates. It
32:57took a while, but if you go through all
32:58the com uh coot positioning in dollar
33:01positions, uh you go look at what
33:03happened in dollar yen, people are just
33:05very crowded now into long dollar
33:06trades. So, if we get an unwind the
33:08other direction, I think it's good for
33:10gold, good for silver, good for Bitcoin.
33:11You guys know that that's the majority.
33:13That's where I've rotated the majority
33:14of my AI uh infrastructure money,
33:17particularly on the semi side. Uh I
33:19still have Marll and I still have
33:21Enterrris and I still have Fluence and
33:23some other ones that I've I'm in.
33:24They're not as large as the memory side
33:26was except for Marll. Uh but I've moved
33:29into Eli Liy. I've moved into uh a bunch
33:31of silver and Bitcoin names. So, here is
33:35the overlay of the 10-year swap rate uh
33:38inflation swap overlaid with 10ear
33:41rates. So, the question is when people
33:43say and I listen to people say this that
33:44the reason rates were going down is
33:47because the market is sensing peak
33:48growth.
33:51The market is following inflation. The
33:53only thing that's not right now is it's
33:55not playing exactly with it. So, if we
33:58get inflation data that starts to
34:00suggest that things are different, maybe
34:01that'll change. Here's crude sitting
34:04down here and the white line here is gas
34:06futures. So the crack spreads have
34:07widened because the gas futures here.
34:09Gas at the pump is down to 380 something
34:12from up near 4 and a half. So you've
34:14seen a 17 18% decline. But if crude
34:18stays down around this level around 70
34:20bucks, you're going to see gas futures
34:21come down. The reason that's important
34:23is because we're already going to get
34:24the negative inflation side most likely
34:26in the next print for headline CPI.
34:29Currently for the June number we have uh
34:32slightly below zero and currently early
34:36at this point for July we've got minus.2
34:39from the Cleveland Fed. Well, if we get
34:41those numbers, we're moving the CPI
34:44which was at 420 down to three and a
34:46half year overyear. So, I just find it
34:49very hard with that going on for people
34:51to get all excited when they can
34:53basically jump on one thing which is PCE
34:56core which Oh, that's right. that's the
34:59one the Fed prefers. Well, let's go
35:01remind ourselves. Back in May, Kevin
35:03Worsh spoke in front of the Senate
35:05Banking Committee. And during his
35:07confirmation, hearing that the Fed's
35:09current preferred gauge, the core
35:11personal consumption easy. So hearing
35:15that the Fed's current preferred gauge,
35:17the core PCE, offered only a rough swag
35:20on inflation being short for scientific
35:23wild guess. Okay, this is Worsh mocking
35:27the PCE core. He has stated for a
35:30preference for trimmed gauges,
35:32especially the Dallas Fed's trimmed mean
35:35and the Cleveland Fed's median PCE.
35:39Well, here is the Cleveland Fed median
35:41PCE sitting down here. Not too much
35:45higher than it was in the period after
35:47the great financial crisis. So, yes. Is
35:50it about say 50 basis points higher?
35:53Yeah, that's a lot different. But here's
35:54the other thing. This is where the Fed
35:56funds rate was during this time. This is
35:58where it is now. We have the Fed funds
36:00rate way above this at this point. It's
36:0280 basis points above. We've got
36:04restrictive monetary policy if he's
36:06serious about this compared especially
36:09to the prior decades. So why would he
36:12hike? Let's continue wages. The white
36:16line here is the Atlanta median wage.
36:18Back to the median again, 3.5%. Here it
36:22was when the Fed started hiking up at
36:26six and change. Here's where we are
36:27here. Basically, we are exactly the same
36:31on the wage side as we were from 2010 to
36:342020. I put this over the quits rate
36:37because the quits rate lines up with it.
36:40And the quits rates is is important. And
36:43the reason it's important is because
36:44it's a good sign of whether people feel
36:46it's okay to jump to another job.
36:50the mar the labor market is not strong.
36:52It is not strong. So using the backward
36:55way of just saying the number of people
36:57hired or using the claims whatever it is
37:00everyone has a job but the ability of
37:03going out and getting higher wages AI is
37:05having an impact. If you strip out
37:07health care there is no jobs market. So
37:09again look at where you have the quits
37:13rate.
37:15We're in the middle. I mean, I
37:18the wage right here, the orange line,
37:20quits rate here. Like, we're in the
37:22middle, guys. There's nothing there. If
37:25you go look over here, the effective Fed
37:27funds rate over the Atlanta Fed median
37:30wage. Again, here's where the median
37:32wage was from 2016 to 2021. This is when
37:36we saw an uptick. This is when they
37:38raised rates. Okay, so here's the Fed.
37:41They raised rates. We were here at the
37:44peak here. So, we're not too far from
37:46where they are. I just don't see a need
37:48at this point based on the data that is
37:50there. And oh, by the way,
37:54how we have forgotten the narrative that
37:56was in play for the prior two years that
37:58we ignore now. The federal debt path is
38:01worse than reported.
38:04And here is the CBO.
38:06How do people forget this? This is why
38:09you're long gold. This is why you're
38:10long silver. This is why you're long
38:12Bitcoin. the situation on this. This is
38:15the math. The math just gets worse. Now,
38:17as an AI guy, do I think this number is
38:19overstated? Yes, because I think we will
38:21be curing diseases and the mandatory
38:23spending won't be as much. But we're not
38:26there yet. And I don't know if
38:27everyone's going to get it, even if we
38:29are able to do that. So, the
38:32discretionary spending is very little.
38:33So when you get back to it, the
38:34mandatory spending for 2036,
38:37$7 trillion
38:39in net interest,
38:42and this is assuming that the Fed funds
38:44rate is here. If Kevin Worsh raises
38:46rates three times and fights in the the
38:49dreaded inflation, which is nowhere,
38:52I I
38:54people are this is PTSD on what happened
38:57before. All right, here we go.
39:00Application layer. Remember this stuff,
39:04everyone's long. Nobody's in this stuff.
39:07This is where Greg Brockman's talking
39:08about. This is where Scott Woo is
39:10talking about. He's talking about
39:11applications. Remember, companies are
39:14not adopting AI tools. They're adopting
39:15AI workers. Productivity, productivity,
39:18productivity. You start seeing profit
39:19margins go higher. Scott Woo again, we
39:23are already moving into a world where
39:25agents can perform hours of work.
39:29So, the KIX, the kicks, the insurance
39:34ETF
39:36or index. Yeah, this is not the ETF or
39:40maybe it is. I can't remember. Um,
39:41regardless, we're breaking out here on
39:43the insurance side after a long
39:46consolidation. Remember what I showed
39:47you on Korea?
39:51KRE regional banks. I think you're going
39:54to have a lot of consolidation with
39:55inside the regional banks. I think
39:57they're going to be buying up the
39:58smaller banks. Uh again, I think the
40:00bigger banks will be the AI ones. I
40:03think companies that can do this on
40:04their own books are going to buy up
40:06smaller banks. Put the same thing in.
40:07They're going to be able to immediately
40:09turn them. You have to think about this.
40:10If you're big enough to put AI into your
40:12business, and you're able to cut
40:14workers, you're going to be able to do
40:15the same thing with smaller banks that
40:17don't have the budget, don't have it.
40:18You're going to see a lot of
40:19consolidation with inside these
40:21medium-siz uh banks into smaller banks
40:23in my opinion. And if not, it'll come
40:25from the bigger banks. But you're going
40:26to see consolidation uh M&A. Uh I wanted
40:29to show this on the breath side. So
40:31remember all the Hindenburg stuff that I
40:33was highlighting about the market. Well,
40:34now that we've done enough shakeouts,
40:36we've done enough kind of consolidation
40:38here. We're back to 8%. This was on
40:41Wednesday. 8% of 52- week new highs in
40:44the S&P. So about 40 names out of the
40:46500 with only one
40:4952- week new lows. Now
40:5213 of those 40 were in financials. Nine
40:55of those are in healthcare. You're
40:57looking for breakouts in them. You're
40:59looking for this. You see a lot of
41:01things in here. Let's go into the next
41:03one.
41:06Uh inside the insurance side, CB TRV,
41:10you've got some of these on there. You
41:12guys can go look at the charts yourself.
41:14With inside the PNC insurance side, AI
41:18dominates agendas. You're starting to
41:20see the same thing that I talked about
41:22with inside the infrastructure. They're
41:23starting to see it showing up on many,
41:26many levels. It's worth reading about
41:28it. In particular, I would go read on
41:30Travelers and go read on on All State,
41:32which are both two names that showed up
41:34on the 52- week new high list. So, the
41:36KBW insurance index worth looking at. I
41:40don't think you're going to see massive
41:41moves in these, but you go back and look
41:43at McKessen. If you can improve just a
41:45little bit on these bloated companies
41:48and get rid of some of the paperwork and
41:51make them more efficient, get rid of
41:52some of the call centers, you can bring
41:53a lot to the bottom line very quickly.
41:56All right, let's get into meta. Building
41:58a business to resell AI compute had
42:01everyone excited this week. Oh, they're
42:03going to that was this is this is not
42:06something to pay much attention to other
42:08than the fact that Meta looked at SpaceX
42:11valuation and recent deals to sell exit
42:13capacity and came to the conclusion that
42:14investors will look favorably.
42:17I happen to agree with this but remember
42:19in March Meta signed a 27 billion cloud
42:22service deal. In April they signed a 21
42:24billion expansion of their cloud
42:25services. Uh two weeks ago Bloomberg
42:28reported that Meta signed up rent one
42:30point.
42:32now all of a sudden they have excess
42:33capacity. I mean, it's really stupid to
42:35to to to leap to that side and not see
42:38that maybe it's because their company's
42:41under pressure because the stock is
42:43acting so badly. They've got a company
42:46backlash going on. So, let's just go
42:49through what they said in their
42:50earnings. We are still compute
42:52constrained. This is from their Q1
42:54earnings, most recent one. And even as
42:57we aggressively add capacity, demand
42:59keeps outrunning our own expectations.
43:02Compute is now strategic, not optional.
43:04They aggressively trying to secure
43:06capacity. The biggest need is inference
43:07capacity, not just training.
43:10Again, I don't I don't know. Continued
43:12to underestimate our compute needs even
43:14as we have been ramping capacity
43:16significantly. She ti tied that to
43:18faster AI progress, new product ideas,
43:20and growing internal use. they are
43:23building with flexibility in case they
43:25overshoot.
43:28That didn't make investors happy. So
43:31that is the reason why that I'm sure
43:34they're saying, hey, if SpaceX got away
43:36with it, maybe that'll allow us as we're
43:38doing this to say that we have that
43:39option so people at least put a put
43:41underneath. Um Meta's potential mega
43:44equity raise puts AI ambitions in
43:46delution focus. Remember after Google
43:48raised money, they talked about going
43:50out to raise more money. Why would they
43:52go raise more money if they already have
43:54excess compute? Why would any of these
43:56companies, why would Google do it if
43:58we're at the beginning of the end of
43:59needing uh compute? Uh this is more
44:02likely having to do with it. I think
44:04this is a good way to look at it is just
44:05for all of these companies, the three
44:07the three bad boys, Microsoft, Meta, and
44:09Oracle, the free cash flow, the price to
44:12free cash flow, negative over here,
44:13negative over here. I mean, at some
44:16point, if they're not getting it in
44:17quick enough, it's hurt. And here are
44:19their stock prices for Oracle, Meta,
44:21Microsoft. So you get this bounce in
44:23Meta. It came all the way right back
44:24down to here. These companies bounced
44:26slightly with it. All the companies that
44:28were hurt with it, the Neoclouds,
44:30they're still down. This was just a
44:32story again that got taken out of
44:35proportion and you'll just uh I'm sure
44:38not have to deal with it. And it's
44:40possible Meta starts showing uh revenue
44:42gains. I I doubt that's going to show
44:44up, but don't be surprised if we don't
44:46get some bounces in these names. they've
44:48been hurt so much. Um, this is the more
44:50likely thing is that the physical
44:53capacity could limit hyperscaler capex
44:56surprises. If they don't announce big
44:58bigger numbers, it's more likely to be
45:00because of the bottlenecks. And that's
45:01basically what I've been saying now for
45:03a couple months is the rate of change is
45:05going to go down. You're going to see
45:06disappointments or you're just not going
45:08to see the amount of surprises that you
45:10saw before. The worst case scenario for
45:12the US is becoming increasingly
45:14realistic and it's basically that we
45:16have few too few data centers.
45:19In particular, semi analysis put out a
45:21chart just basically showing we don't
45:22have enough of this. I'll get into that.
45:24I'll show you that chart in a little
45:25bit. The BIS put out something just
45:27showing the risk involved in it. And
45:29here is the semi analysis side just
45:32saying um how much gigawatts we need
45:35based on the growth in AI and chip
45:37supply. And we're just down here. We're
45:40not anywhere where we need to be. Now,
45:43just a quick thing on GLM 5.2.
45:46Um,
45:48again, American and European enterprises
45:50will ditch open AI and anthropic and
45:52adopt Chinese models. I don't agree with
45:54that, but again, uh, case concerning the
45:57worst case scenario for USA, Chinese
46:00open source keep getting this is from
46:01Mark Andre. So, again, you don't want to
46:03ignore Mark. Many smart people are
46:05saying GLM 5.2 2 is the first Chinese
46:07model to match and often beat the
46:09American big watch model. Incredible
46:10timing given our current events.
46:14This is the clawed moment.
46:16This is from the head of data bricks. So
46:19a company that would know the demand
46:21we're seeing at data bricks is
46:22astonishing. The world is going to see
46:24massive adoption of open-source
46:27LLMs.
46:29More companies will shift towards
46:30post-training their own models on top of
46:33the OSS models and owning the weights.
46:37Um, reminder, Gavin Baker said that
46:40we're going to go to more composable
46:42markets and said he had to challenge his
46:44thought process because of GLM 5.2. So,
46:47in my opinion, this does create
46:51a potential issue for the slowing down
46:54of the ARR for Enthropic. I believe that
46:57is a possibility. Now, remember, they
46:59have been a 10 timeser in terms of the
47:01revenue. You heard many many places say,
47:04"Well, if I multiply it again by 10 and
47:06then they'll be at a trillion dollars at
47:08the end of next year." Anything that
47:10starts showing that they're not going to
47:11hit 10 times, I think this will become a
47:14major story, which I think is realistic.
47:18I I do think that open- source models
47:21are going to be used by a greater
47:23portion of the country uh than people
47:26realize particularly as these expenses
47:28go and particularly when we get
47:31the hardware out there to be able to run
47:33it. All right, now to finish up the last
47:35few slides for Bitcoin.
47:38Reminder, here is the AI tech part V
47:46momentum. This gets really important. As
47:48this goes higher, it's getting harder
47:50and harder for people to speculate with
47:52inside the AI infrastructure trade. All
47:54of the headwinds I showed, I think
47:56you're going to make this not as easy.
47:58So, if going forward we're getting 40% a
48:00year, that's still a great number. It's
48:02better than the S&P will do, but it
48:04might disappoint people that are looking
48:06for eight and 10 baggers.
48:08So, now we go over to Bitcoin. Bitcoin
48:10sitting on its 60-day V at basically the
48:13lowest V it's been. um and it's been
48:17sitting down here now for quite some
48:18time. So, we've had a restructuring
48:20involved. You can thank Michael Sailor's
48:23um job for that. As people, as I said
48:25last week, when people are calling me up
48:26saying Michael Sailor is going to blow
48:29up Bitcoin, he's horrible for it. That's
48:31usually the point that the bottom comes
48:32in. So, I start looking a little more
48:34closely. Here is the rate cut chart
48:36overlaid with Bitcoin. So regardless of
48:39whether you like it or not, as we've
48:41taken out the rate cuts and actually
48:42gone to minus
48:45cuts, meaning hikes, Bitcoin has
48:47followed it lower. So if we get any kind
48:49of change right now or stabilization
48:51where this trend doesn't get as worse,
48:52maybe we'll see something. I just want
48:54to let you guys know that over the last
48:56100 days,
48:59we are back where we were. So we're at
49:03minus 5.8% as of Thursday.
49:07That's pretty impressive. Meaning for
49:09all this bare market, the rate of change
49:11has been negative the entire time, but
49:12now we're getting less bad. And I think
49:14that's important if we start to break
49:16higher. You can see that the 20-day
49:18moving averages here. We need to see
49:20some positive technicals. And here we
49:21go. So, for the first time, we have
49:26divergences. And what I mean by that is
49:29we actually have
49:32lower lows and higher highs. And this is
49:35going back to July of last year. It's
49:38been a long wait. I've been looking at
49:40this. I have this up on my Bloomberg.
49:41I've been waiting for divergences and
49:43then I wanted to see a week where it
49:45happened on
49:47bad news.
49:48Clarity Act passing falls to 39% and
49:52President Trump discloses 1.4 billion in
49:54income. All right. Um I do want to leave
49:57you guys with one more thing on the
49:59Bitcoin side. again, now that we have
50:01hikes built in, if they just stay status
50:04quo, and Kevin Worsh keeps saying that,
50:06you've taken the pressure off. Trump
50:09hired Kevin Worsh, and even though he
50:11wants him to quote unquote fight
50:13inflation, fighting inflation with talk
50:16is very important, but the midterms are
50:18coming up. So, I find it highly unlikely
50:21for him to be too hawkish ahead of the
50:23midterms other than just say that
50:25inflation shouldn't be here. And if we
50:27get negative CPI prints, I do believe it
50:30gives him the ability to be more
50:31patient. That might be the side that
50:34ends up happening. So, I'll leave it
50:36there, guys. Again, I hope you had a
50:37great fourth. I'll see you next week.

About this transcript

This page contains the full transcript of Momentum Is Crashing, Bitcoin Is Bottoming, AI Agents Are Rising by Jordi Visser, generated from the public captions YouTube serves with the video. The transcript has 9,238 words across 1,335 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.