AI Is Changing Time: Those Who Can’t Compete, Compute

[0:00] All right. Uh
[0:03] just a little bit of notes um because
[0:06] people have reached out. First of all,
[0:08] this um because this may be uh the last
[0:12] week that I'm in Maine. Uh not sure yet.
[0:14] I'm not sure whether I'll be recording
[0:15] from here or Maine next week, but uh
[0:19] I've been asked a lot of questions
[0:21] particularly about uh Maine. And I just
[0:26] want to give people who are getting to
[0:27] know me, have seen me on podcasts. Uh
[0:30] the reason I'm in Maine is because on
[0:32] September 11th, 2001, and we're
[0:35] approaching 25 years since September 1st
[0:38] or September 11. Um many of you know the
[0:42] best my best friend growing up and the
[0:43] best man at my wedding died 911. Um I
[0:46] ended up uh going back to Morgan
[0:48] Stanley. I was a partner there already.
[0:50] I was on the trading floor the day after
[0:51] when the markets were closed uh
[0:54] basically with other partners figuring
[0:55] out how we were going to operate and run
[0:56] business and uh very emotional obviously
[1:00] and I got a phone call. I had made an
[1:01] offer on a house in Maine because I had
[1:03] taken my daughter up there, my oldest
[1:04] daughter up to camp there. Fell in love
[1:06] with the place. Reminded me of my father
[1:09] um being the son of a a bluecollar
[1:11] worker and being the son of a core
[1:14] driller. Uh I just fell in love with the
[1:15] place. I did not like the shores uh of
[1:19] the area that I grew up in in New Jersey
[1:21] and definitely not the Hampton's my
[1:23] style. So, I fell in love with Maine. I
[1:25] bought a home in Kenny Bunkport. Um
[1:27] moved up there and I did the commute
[1:29] there for about 15 years. I've moved
[1:31] further north now to get away from the
[1:34] people that found Maine. Uh but in the
[1:36] background, people have asked about the
[1:37] painting. The painting is a reference to
[1:40] uh Shawn Christopher Shawn Kaitton who
[1:42] was a caner who died. Uh, I bought that
[1:45] painting after I moved to Maine. The
[1:46] wine bottle behind me is the bottle that
[1:49] I bought in Napa Valley, which is a
[1:51] remembrance of 911. I turned the bottle
[1:53] around this week. It has remembrance if
[1:55] you guys want to blow it up. You've got
[1:56] a Bernard King signed basketball there.
[1:57] You've got my Nicks hat on. Sean and I
[1:59] went to um many games. And then also
[2:02] behind me is a photo of him and I at a
[2:05] bar in the Bronx many hours after the
[2:07] Yankees won the World Series uh where
[2:09] Wade Bogs was riding the police horse.
[2:12] Uh Maine is a special place for me. I
[2:15] come up every year and since you guys
[2:16] asked about it, I just wanted to go
[2:18] through. Now on other notes for you guys
[2:20] that have reached out, number one, I
[2:21] will never reach out to you in WhatsApp
[2:24] offering you trades or doing anything. I
[2:26] will never do that in X. If you reach
[2:28] out to me in LinkedIn, I may or may not
[2:30] respond because I get hundreds of people
[2:34] reaching out. Same thing goes in my
[2:36] email. So don't take it personally if I
[2:38] don't respond. Don't I'm never reaching
[2:41] out to anyone. So, unless you're on the
[2:43] subscriber emails, you're not getting
[2:45] anything from me. I say that because
[2:47] there's AI bots everywhere and people
[2:49] impersonating me on every single
[2:50] location. I tried to get rid of them for
[2:52] a while. Didn't work. Um, on another
[2:56] note, people that are sending me stuff
[2:57] about what do you think about this
[2:59] Goldman Sachs person or this Morgan Sax
[3:01] person, the Gold Morgan Stanley person,
[3:02] they have a different view on XYZ. Guys,
[3:04] if you haven't learned anything watching
[3:06] this, I don't care about anyone's
[3:07] opinions. I care about everyone's
[3:10] opinions, but not anyone's opinions. I
[3:12] do my own work. I take all of the
[3:15] information that I can together and then
[3:16] I make a decision in a paramutual manner
[3:20] where I go through the odds and I'm
[3:22] trying to find things that people either
[3:23] don't agree with me on. So, I'm
[3:25] generally trying to be somewhat of a
[3:27] contrarian, but at the same point, I'm
[3:29] trying to find ideas that have good
[3:30] riskreward. I will always be an
[3:32] independent thinker. I will always be
[3:34] flexible on my views. So, I'm going to
[3:35] take you guys through that today cuz
[3:37] some of you said that I'm only positive.
[3:39] And if that's what you believe from the
[3:41] subscriber list and if you ended up
[3:43] buying the top of the market because you
[3:45] subscribed and said, "Oh, this is a bull
[3:47] market in something." I'm going to go
[3:49] through and show you how to take all of
[3:50] the transcripts I do and I'm going to
[3:52] give you a prompt to incorporate it into
[3:54] a way where you can take the
[3:55] information. I'm just speaking from the
[3:57] heart every single week. Guys, one thing
[4:00] I will say repeatedly, we are in a
[4:02] structurable market in the demand for
[4:04] compute. It is insatiable. And until I
[4:07] start hearing people that I respect on
[4:08] Wall Street start recognizing that there
[4:11] is no end ever coming up for the demand
[4:13] for compute, just like there is no food
[4:16] end. If we keep increasing people on the
[4:19] planet by a billion every year, that we
[4:21] will never catch up. The same thing for
[4:23] compute. It is impossible to get enough
[4:26] compute for all of the billions of
[4:29] agents that are coming online that are
[4:31] just starting. This is investing in
[4:33] compute. So, this whole video today is
[4:36] to some degree going to be about
[4:38] compute. And that's why I'm going to go
[4:40] back to the agenda side for this week.
[4:43] Um, my job is to help your business,
[4:46] whether you're an RAIA, whether you're
[4:48] an asset manager, help you with
[4:50] research, help you with independent
[4:51] thinking, but also get you ready to help
[4:54] your children adjust to this world
[4:56] because it's coming fast and it's going
[4:57] to be accelerating over the next
[5:00] certainly 24 months. Crypto is going to
[5:02] be a bigger part of the discussion. And
[5:04] if you have no knowledge of crypto or no
[5:06] understanding as to where it's going to
[5:07] be, you're going to suffer in the next
[5:08] 12 months for sure. But certainly 24
[5:11] months, you're going to have to adapt.
[5:13] So we'll go through the recap. We'll go
[5:14] through the compute. We'll go through
[5:15] the token factories, Coreweave, Google,
[5:17] SpaceX, their commentary. The AI regime
[5:21] change, which has to do with the stuff
[5:22] that I'm doing. And probably most
[5:25] importantly, two things there. Number
[5:27] six. Number nine. Number six, we are
[5:29] going back to work, guys. The summer is
[5:31] over. I'm heading back to New York. I
[5:33] have spent a lot of time working. I've
[5:35] tried to create prompts. I've created a
[5:37] new website. There's a whole bunch of
[5:38] things I've done um with AI agents,
[5:42] which hopefully I've been able to give
[5:43] you guys some stuff that has helped. But
[5:45] the 2027 outlook is what I'm going to
[5:47] start focusing on because next year is
[5:48] about consumer agents. This year was
[5:51] about agents, the agentic economy
[5:53] opening. Next year is about consumer
[5:54] agents and what they need, which is the
[5:56] financial guardrails, which is where
[5:57] crypto shows in. Number nine, be an
[5:59] entrepreneur, not an academic. Academics
[6:02] will continue to cost you money. Anyone
[6:05] that is giving you history and analoges
[6:07] and blah blah blah, especially
[6:10] economists, I will take you through why
[6:12] that is not worth your time. S&P up 40
[6:14] basis points for the week.
[6:16] IWM up 1%. Q's up 1%. The JV AI thematic
[6:23] portfolio outperformed again, up 3% for
[6:25] the week. We made new all-time highs in
[6:27] most of the important indices and also
[6:31] around the globe. uh the earnings and
[6:35] again I'll continue to say it earnings
[6:38] growth 51% for the S&P sales growth
[6:41] probably the more important number up
[6:43] 15% year-over-year
[6:47] I can't remember a time when we were in
[6:49] a mature part of the economic cycle and
[6:50] global UPS is something like 47%
[6:53] year-over-year everywhere you look EPS
[6:56] has been moving smartly higher
[6:58] this is the critical thing if you start
[7:00] believing in people that tell you this
[7:01] is is an isolated bubble to a few tech
[7:03] names in the US. It's not true. You've
[7:06] got the whole globe breaking out in
[7:07] terms of earnings per share, and that's
[7:08] because time is changing. You're going
[7:11] to hear me talk about this a lot. This
[7:13] is going to provoke thought in you guys.
[7:14] I will have a very important Substack
[7:16] coming out. For you Substack readers,
[7:19] the time is coming where I'm going to go
[7:20] down to about once a month in my papers
[7:22] because I'm entering in the crypto side
[7:24] and most of the stuff is shifting to the
[7:26] subscriber site for my crypto work. Um,
[7:30] this is the MSCI World Bank index.
[7:35] This is parabolic and going higher. This
[7:37] is a reflection of what's happening with
[7:39] AI. All of these charts are a reflection
[7:41] of what's happening with AI. It is a
[7:44] reflection of what's happening around
[7:45] the globe. AI is disrupting all academic
[7:48] research that you know. All the things
[7:51] that you've believed in are breaking
[7:52] down. And the reason is because
[7:54] productivity is happening. more and more
[7:57] of the smarter people with inside the
[7:59] Fed that I read and the IMF and the
[8:01] they're starting to recognize now again
[8:03] it's got to be the younger people the
[8:05] productivity boom is happening you have
[8:08] to put it into context though earnings
[8:10] are up 30% year-over-year in the S&P
[8:13] forecast for this year this is our job
[8:16] hiring zero
[8:19] economists tell you the labor market is
[8:21] strong they want the Fed to raise rates
[8:22] this is all BS the labor market relative
[8:26] to the way things move over time. We
[8:29] should be hiring lots of people because
[8:31] we're growing earnings, not coming out
[8:33] of a recession like these periods at a
[8:35] record pace.
[8:37] So why aren't we hiring people? When I
[8:39] was at Morgan Stanley, if we had
[8:40] earnings growing rapidly, we'd be hiring
[8:42] people. That's what you did. You threw
[8:43] more money at more people. Not happening
[8:46] anymore. And the reason is because we've
[8:48] got enterprise agents which are coming.
[8:50] Not having started yet. They're coming.
[8:53] This does not mean we'll be losing
[8:55] millions of jobs. But what it does mean
[8:56] is we have job losses already. X
[8:59] healthcare. We're already seeing small
[9:02] job losses. That is productivity. We
[9:04] don't need as many people. We're seeing
[9:06] businesses set up. Startups are
[9:09] happening. We're getting more business
[9:11] formations. Jordy has a business.
[9:14] All of this stuff. I don't work now for
[9:16] a Fortune 500 company and I don't work
[9:18] for a country a company with 100 people.
[9:20] Started my own business. So consumer
[9:23] agents are coming next year. This is the
[9:25] token needs for them. This is what
[9:27] they're paid in. This is what they eat.
[9:29] This is what they need to survive. They
[9:31] work 247
[9:33] 365.
[9:35] They're shrinking time. So when you say
[9:37] year overyear for anything now, that is
[9:40] a human thing. Agents don't sleep.
[9:43] Agents don't go to the bathroom. Agents
[9:45] don't go to lunch. Agents don't go to
[9:47] the pantry to flirt. Agents don't do any
[9:50] of that stuff. So what you're left with
[9:52] is token consumption is growing rapidly
[9:55] and time is shrinking. So in the case of
[9:58] year-over-year, the reason we're seeing
[10:00] earnings grow so fast, the reason
[10:01] nominal GDP at this point is at least
[10:04] growing where it is because the math on
[10:06] that is not correct, you have to focus
[10:08] on the profit margins and you have to
[10:09] focus on the earnings growth. Time is
[10:12] shrinking. So we're now doing
[10:14] effectively two times what it used to
[10:16] take in one year and then eventually
[10:17] it'll be three times and four times and
[10:19] five times. And the reason is because
[10:21] the agents are working all the time. Now
[10:25] this doesn't mean that things go
[10:27] parabolic in terms of the earnings. What
[10:29] it means is the earnings continue to
[10:30] outpace the prices because private
[10:32] companies are doing a lot of this. They
[10:35] are going to steal from the public
[10:36] companies. We'll spend more time on
[10:38] this. If you don't believe compute is
[10:40] insatiable
[10:42] right now, all of you know what coding
[10:43] agents are. At least you should. If you
[10:45] don't and don't know what I just said
[10:47] about 247365,
[10:49] well then we're going to get into
[10:50] finally them actually being deployed.
[10:52] They haven't even been deployed yet. We
[10:54] just have the capability of using them.
[10:56] Enterprise agents become persistent
[10:58] digital labor. This will be happening
[11:00] next year. Consumer agents will be
[11:02] following in the latter part of next
[11:03] year. This is your Jarvis. This is
[11:05] people ordering things for you. This is
[11:07] like your AMX uh uh assistant who does
[11:13] everything for you. this is coming and
[11:15] they'll be doing things while you sleep.
[11:17] They'll be booking your travel. They'll
[11:18] be ex orderering things for you.
[11:20] Everything will be happening
[11:21] instantaneously and we'll start reducing
[11:23] the friction in the economy. Then you're
[11:25] going to get into autonomous driving and
[11:27] for that you need a world a world model
[11:30] problem. We need a hell of a lot more
[11:32] compute for that. We need a hell of a
[11:34] lot more compute for humanoids. These
[11:36] are all coming every single year. These
[11:38] innovations used to take decades to
[11:40] come, but now they're happening in such
[11:41] a compressed amount of time because of
[11:43] intelligence.
[11:44] The numbers are staggering. Right now
[11:46] we're in this part. So this year we
[11:48] needed 10 to 100 times for the coding
[11:50] agents relative to last year for
[11:52] chatbots. That's why everything went
[11:54] parabolic. Look at the stepup function.
[11:57] Another 10 times. Then another 10 times.
[12:01] It never ends guys. The amount of
[12:03] compute we need never ends. We are
[12:07] transitioning from systems that give
[12:08] instant cash responses chatbot to
[12:11] systems that plan, reason and simulate
[12:12] outcomes before taking an action. This
[12:14] is so complex. This is demos sabas. We
[12:17] need magnitudes more compute. That's why
[12:19] Jensen Yuan comes out and says, and he
[12:21] only said this at the beginning of this
[12:23] year, it's going to be AI factories, the
[12:25] new infrastructure intelligence. We need
[12:26] this is what we need to produce. This is
[12:28] why there's compute futures. This is why
[12:30] everything you're hearing is about, oh,
[12:32] we need compute.
[12:34] We need it endlessly to feed the new
[12:37] billions of employees on the planet.
[12:40] Jensen Yuong says comput is the new
[12:42] economy. He said this at the Morgan
[12:43] Stanley event in March.
[12:47] Nvidia this week, AI factory computing
[12:50] investable asset class. The reason we
[12:51] know this is a fact is because Apollo,
[12:54] Black Rockck, Blackstone, Brookfield,
[12:55] Goldman Sachs, KKR to establish AI
[12:58] compute infrastructure financing
[13:00] platforms to mobilize over half a
[13:02] trillion dollars of third party capital
[13:05] to build the stuff that Nvidia needs.
[13:07] They don't want to use their balance
[13:07] sheet anymore. And people all signed up
[13:10] to do this. Morgan Stanley to facilitate
[13:12] $1.5 trillion in infrastructure
[13:13] initiative.
[13:15] The capital's there. Alphabet raises 25
[13:18] billion this past week. Intel announces
[13:20] a 15 billion stock offering for a
[13:23] company that was worth less than a
[13:25] hundred billion a year ago and they
[13:28] raised it to 20 billion in the stock
[13:31] traded higher than where the issuance
[13:33] was. China is in the same race which
[13:36] means the compute spending is happening
[13:38] around the globe. You can go read the
[13:39] little yellow lines but China and the US
[13:42] are in a race. The government is still
[13:44] trying to make sure that we don't lose
[13:46] to China. So, exempts data center bonds
[13:48] from key securization rules to make sure
[13:51] that it's easier for data center owners
[13:54] to sell asset back securities
[13:55] potentially open the door for more debt
[13:57] sales.
[14:00] If you're looking for data center
[14:01] delays, they're minor. We're still
[14:03] moving forward on everything on the data
[14:05] center side. Despite what you're reading
[14:06] in the media, there are delays. I've
[14:08] shown you guys my work on this. It's
[14:11] already built into the way things are
[14:12] being done. Nothing is a new surprise.
[14:15] All of these are going on. They will be
[14:17] delayed to some degree, but the capacity
[14:19] is still coming online. It's not as fast
[14:22] as it otherwise would be, but to be
[14:23] honest with you, it doesn't really
[14:24] matter because we don't have enough
[14:27] memory anyway. So, all of these delays
[14:30] are actually good because then we're not
[14:32] over buying things or at least overusing
[14:35] things before we get it in. So, the
[14:37] delays allow the revenues to go a little
[14:40] bit further out even though they've
[14:41] committed to them. coreweave this week.
[14:44] Very important in terms of going
[14:46] through. You don't have to believe
[14:47] anything they they say. You can go
[14:49] through it, but again, I've got tons of
[14:50] data that supports everything they say.
[14:52] I think the most important thing to go
[14:54] against things that you've heard from
[14:55] Michael Bur and from Jim Chados in
[14:57] particular, which I'll call out on this
[15:00] because they don't ever turn around and
[15:02] say, you know what, we were wrong. They
[15:03] just keep saying, "No, this is this is
[15:05] BS." They never change their mind ever.
[15:07] It never there's no like it's a 70%
[15:10] chance. It's always this is what's going
[15:12] to happen. So if you pay for that at
[15:14] least hopefully you pay for my stuff as
[15:16] well where I'm going to spend time where
[15:17] at this point I'm saying demand is
[15:19] insatiable.
[15:21] The A1 A100 is being contracted all the
[15:24] way to 2029. This was introduced in
[15:26] 2020. They're saying they're completely
[15:29] sold out. You can go through the
[15:30] physical infrastructure itself has
[15:32] scarcity value. So I've talked about you
[15:34] want to invest in scarcity. So they were
[15:36] originally saying two to three years may
[15:37] be realistic. They're talking about it
[15:39] being six to nine years now. New
[15:42] generations will be gone because the old
[15:44] one, we don't have enough data centers
[15:46] being built that we can replace it
[15:48] because the demand is greater than the
[15:49] supply. It's like walking into a
[15:51] supermarket. If they didn't have any
[15:54] food but cans of sardines, you would
[15:56] immediately love cans of sardines.
[15:59] Even if that was the last thing you
[16:00] wanted, that's what's happening is the
[16:02] demand is too great. we have too many
[16:05] agents buying or wanting and demanding
[16:07] too many tokens. You guys can go through
[16:09] the rest of this. The reality is it's
[16:12] just the fact that on every level
[16:14] whether it's pricing whether it's the
[16:15] stuff we hear from them what we hear
[16:16] from SpaceX what we hear from Google
[16:18] what we see from Google on that point
[16:20] with Google the article the uh podcast
[16:24] to listen to this week Google's AI brain
[16:25] drain from
[16:27] the all-in podcast they talk a lot about
[16:30] Google's reorganization framed it as a
[16:32] capital allocation shift I'm going to
[16:33] get into that a little bit more um
[16:36] two-horse race according to David Saxs
[16:38] completely agree AI infrastructure may
[16:40] be the cleaner business than model
[16:41] development. So, think of it this way.
[16:44] Google's making a
[16:46] decision similar to what Meta pushed out
[16:50] there and SpaceX did, which is maybe we
[16:52] should sell some of the compute since we
[16:54] can't use it all right now and our
[16:55] investors are getting angry with us
[16:57] because free cash flow is going
[16:58] negative. It's forcing you into
[17:00] businesses that have a guaranteed ROIC
[17:03] as opposed to chasing AGI. The problem
[17:06] is the talent wants to chase AGI. So you
[17:09] get Jeff Dean, you get Deis Sabis
[17:11] possibly leaving. We've hit an important
[17:14] inflection point now. And where this
[17:16] shows up is if talents leaving,
[17:17] multiples compress. Plain and simple. I
[17:20] I don't see any any other way around it
[17:22] unless Google can find some other
[17:23] business. But the businesses that they'd
[17:25] be creating that are new in my opinion
[17:27] would have to be from the frontier model
[17:29] side or at least that's my opinion to
[17:31] higher probability.
[17:33] Brad [clears throat] Gersonner gave a
[17:34] huge com huge talk on this. Again,
[17:37] Nvidia has financing limits, which is
[17:39] why they rolled out all these companies
[17:41] on there. Um, he talked about all of
[17:44] these different things. Frontier Labs
[17:45] are willing to overpay for compute
[17:47] because there's shortages. Four times,
[17:49] even five times market pricing to secure
[17:52] large-scale compute. It gives them a
[17:54] competitive advantage because other
[17:56] people can't get it. So, the buyers are
[17:58] concentrated because they have the
[17:59] revenues coming in and they want to
[18:00] secure the compute. And that's what he's
[18:03] saying. This is very concentrated. So no
[18:05] one's disagreeing that anthropic and
[18:07] open AI are the cyclical nature. At the
[18:09] same time, you have Nvidia doing a lot
[18:11] of the lending to make sure that these
[18:12] people can b buy their chips. It is
[18:14] circular.
[18:16] That's just the reality of what it is,
[18:18] guys. And unfortunately, AI is a
[18:20] completely different thing than anything
[18:21] you've seen. Dennis Sabas wants to leave
[18:24] Google, but he's staying. But they're
[18:26] predicting he could leave within a year.
[18:28] Again, it gets into the AGI side.
[18:30] Moonshots went through this in detail in
[18:33] two podcasts this week. One with Immad
[18:36] Mustach, one with a kid named Kush.
[18:40] I think Alex
[18:43] said it best. Those who can't compete on
[18:46] the frontier models go towards compute.
[18:49] That's what Meta, that's what XAI, and
[18:52] that's what Google did. And their models
[18:54] are way behind the frontier models. The
[18:56] frontier models are going to be
[18:58] important. Open source models are going
[18:59] to be important as well, but they may
[19:02] Google may be winning distribution. So
[19:04] they cover this as well. They talk about
[19:05] the Sergey Brin side. You guys can
[19:07] listen to it. I'll give you a couple
[19:09] summaries towards the end. But the main
[19:11] story is that Google's real business may
[19:13] be the cloud and it may be
[19:14] infrastructure. If that's the case, what
[19:16] multiple should they have if it's not
[19:18] Gemini? They may lose the model prestige
[19:20] race and still make a lot of money from
[19:22] AI infrastructure, but they may become
[19:24] the Saudi Arabia OPEC
[19:27] of compute while OpenAI, Anthropic, XAI,
[19:30] and others compete on the intelligence
[19:31] side.
[19:34] Google's talent brand has deteriorated.
[19:36] And this is where Kush comes in. He says
[19:38] that for his MIT cohort and younger
[19:40] builders, Google is no longer the dream
[19:42] destination. The dream destinations are
[19:44] now OpenAI, Anthropic, XAI, and Frontier
[19:46] Labs. Unfortunately, that is the
[19:48] reality. So, these companies
[19:52] are different.
[19:54] This is SpaceX. This is a private
[19:57] company that's going public. This is
[19:58] another private company. These are the
[19:59] Frontier Labs.
[20:02] This is the way it was for Google and
[20:03] them before
[20:05] they invented a lot of the modern AI but
[20:07] the spiritual center of AI has moved
[20:09] elsewhere. This is an important dynamic
[20:10] and one of the reasons why in the
[20:11] hyperscalers I will continue to say I
[20:14] want to belong the receivers of the
[20:15] capital because the infrastructure
[20:16] demand is insatiable. These guys are
[20:19] proving it because Google is seeing
[20:21] talent leave because they don't have
[20:22] enough compute. If they had enough
[20:24] compute, they could do both the frontier
[20:26] models and
[20:28] the business side. But they can't. They
[20:31] don't have enough compute. So they have
[20:33] to allocate the resources, the scarce
[20:35] resources somewhere. Google committed
[20:37] roughly 200 billion capex this year to
[20:39] compute infrastructure. That capital
[20:41] earns a documented 30% ROIC is taxed
[20:44] advantage through accelerated
[20:46] depreciation is near certain demand
[20:47] because the whole industry is compute
[20:49] starved. Model development by contrast
[20:51] requires tens of billions with a high
[20:53] variance and no guaranteed return.
[20:56] Again, this all gets into the argument
[20:58] that was there. Now, I asked them to
[21:00] take the transcript of the
[21:05] Brad Gersonner one with the allimp pod
[21:07] and then the the moonshots one. So,
[21:10] these answers here and if you guys
[21:12] haven't done this before, take all the
[21:14] smart people and now you put them in one
[21:16] table and you ask them a question. what
[21:19] the Google shakeup reveals about compute
[21:21] talent incentives and multi-state that's
[21:23] the way that I like to go through these
[21:24] things and then I go through through the
[21:26] disruption side the syn synthesis for
[21:30] all large companies the durable assets
[21:32] of scale are capital physical
[21:33] infrastructure distribution and
[21:35] regulatory position the rational endgame
[21:38] which Google is stumbling into or
[21:39] engineering is the large company as
[21:41] capital allocator and infrastructure
[21:43] landlord to an ecosystem of
[21:44] missiondriven vehicles is partly owns
[21:47] companies that fight this trying to
[21:48] retain everything internally will lose
[21:50] the talent and keep the costs.
[21:54] I subscribe to that entire view.
[21:58] SpaceX semi analysis put this out. Um
[22:02] Elon shocked the world again on SpaceX
[22:03] first earning call saying 6 to8 gawatts
[22:06] of incremental data centers in 2027
[22:08] alone, possibly well above 10 gigs.
[22:12] Sounds impossible, but we believe it. We
[22:14] see SpaceX on track for about 10
[22:16] gigawatts by year end 2027.
[22:19] That's why SpaceX has value to the
[22:22] company. And again, you should view the
[22:24] other side as a call option. But Elon
[22:26] Musk has already proved that he can
[22:28] build data centers faster than anyone.
[22:31] And the gigawatts are worth a lot of
[22:33] money as we're seeing in terms of the
[22:35] compute prices.
[22:38] Um 80% of its output earmarked for
[22:41] space-based data centers. So again, he's
[22:44] doing a lot of stuff for terraab, which
[22:47] is what I brought this up for. Terraab
[22:50] is just confirmation of what's
[22:52] happening. Just look at this thing that
[22:54] he showed in terms of what the scale of
[22:56] it will look like. I mean, you're
[22:58] talking about him building something.
[23:00] Elon Musk knows where the compute side
[23:03] is and he's basically telling you I need
[23:05] to build this which looks like a Star
[23:09] Trek or a Star Wars situation and a
[23:12] massive amount of land. He needs this
[23:15] for the chips and in particular memory.
[23:19] SpaceX goes exclusive with Nvidia
[23:21] putting Ver Rubin GPUs in orbit. And
[23:23] again, this was part of the reason why
[23:26] when I talk about Nvidia and go through
[23:28] it, he said, "We think the Ver Rubin
[23:30] architecture is the best architecture.
[23:31] We're exclusive to Nvidia." Yet, he's
[23:34] building terra. So again, as you guys
[23:37] are listening to this and I'm picking
[23:39] out people that are just bearish on
[23:40] this, I want to remind you. So for raas
[23:44] this is the one pager that I'm now
[23:46] sending to people to go through on why
[23:50] my research is gear is geared towards
[23:53] people to understand what's happening in
[23:55] AI which to me cannot be covered by the
[23:57] street you certainly cannot get research
[24:00] that is put together
[24:02] in a way like this know what has changed
[24:06] and whether it matters to give all this
[24:09] information there's a hundred slides put
[24:11] together in this video today. I'm taking
[24:13] you on a journey to make you learn
[24:15] through data-driven stuff. You can go
[24:17] upload this anywhere you want. You can
[24:19] have your own agents bring it in. You
[24:21] can go check everything I'm saying. You
[24:22] can take the transcript. I do this every
[24:25] week. I come with a hundred plus slides
[24:27] every week of new information. I listen
[24:30] to podcast relentlessly. And this way
[24:33] you can go in and see what's changed on
[24:35] a real-time basis. You can't get that
[24:36] from the street. Know where it matters
[24:39] and when it is turning. I not only give
[24:41] you the names, but I spent the time, as
[24:43] I'm going to show you soon, every week
[24:46] doing this, telling you what I think the
[24:47] riskreward is in the market. And if you
[24:49] don't believe that, I'm going to show
[24:51] you how you can do it on your own, and
[24:52] give you a prompt to do it on your own.
[24:54] The agency stuff is important. I'm
[24:56] giving you prompts, which I'm giving you
[24:57] one this week, workflows, Excel files
[25:00] with all of the information on this,
[25:02] tutorials on how you can put it to work
[25:04] immediately, and hopefully things like
[25:06] the Jensen Senuong knowledge brain,
[25:08] which not only help you, but your kids,
[25:10] which I've heard from them. But next
[25:13] year,
[25:15] starting in September, the next
[25:17] constraint is not physical, it's
[25:19] financial, the guard rails, consumer
[25:22] agents, the settlement, it's stable
[25:24] coins. the tokenization. This is going
[25:27] to impact all of your portfolios,
[25:30] the bridge between traditional
[25:31] portfolios and unchanged assets. Your
[25:32] client's holdings are what moves onto
[25:34] these rails. You obviously have to learn
[25:36] all this stuff before it happens. Reach
[25:39] out to Mark.
[25:41] Mark will guide you. Now,
[25:45] I have said repeatedly in this that the
[25:48] economic structure is changing, the
[25:50] market structure is changing and the
[25:52] earning structure is changing.
[25:55] So earnings history, economics history,
[25:57] market history, all of this stuff is
[25:59] becoming useless. This is what this
[26:00] whole thing is focused on. There is a
[26:02] regime change that's happening. So if
[26:04] you're listening to economists saying
[26:05] the labor market is too strong, we have
[26:07] to raise rates. Inflation is too high,
[26:09] we have to raise rates. We are seeing a
[26:11] disconnect here that has never happened
[26:12] before. On the V side, the market
[26:15] structure side with AI agents, we are
[26:16] seeing tech momentum V up here while the
[26:19] NDX V barely budges.
[26:23] The earnings are growing rapidly.
[26:25] Companies are able to grow earnings
[26:27] because time is changing. This portfolio
[26:30] is meant to show all of that. And again,
[26:33] you're dealing with something that has
[26:35] gone up massively,
[26:38] had a consolidation, and has now bounced
[26:40] back and made back 50%. I've given you
[26:43] this in a 25 name, in a 10 name, in a
[26:47] 100 name, and then Morgan Stanley gives
[26:49] you different ways to play it.
[26:53] In terms of the reason that the street
[26:55] can't cover it, the research firms are
[26:57] all built vertically. They're built with
[27:00] silos. There's a reason why I left
[27:02] Morgan Stanley in 2001. And it wasn't
[27:04] just because of 911. It was because 911
[27:07] just recognized that life is too short.
[27:09] And if my job as a 30-year-old who is
[27:12] now firing people for a living is what
[27:14] my life is going to be, I need to move
[27:16] on. They can't cover everything
[27:19] together. It's not possible. AI is not a
[27:24] sector. It goes across industrials,
[27:26] materials, chemicals, energy, power,
[27:29] semiconductors.
[27:31] Marll is not the same semiconductor as
[27:33] Infinion, as Invidian, as Intel. They're
[27:37] all different, guys, and you know that.
[27:39] So, my job is to show you the structure
[27:42] that's changed. The silos can't see
[27:44] across them. the speed we're moving at
[27:48] at least five times the speed of what
[27:50] the world used to look like which makes
[27:52] academics
[27:54] any back test anything on earnings every
[27:56] time we see earnings like this it's a
[27:57] bubble okay but we're not hiring anyone
[28:00] and there's usually hiring that's going
[28:01] on before a bubble we're overhiring at
[28:03] the bubbles
[28:05] synthesis averages destroy the signal
[28:08] too many people slow decision-m the
[28:11] friction that goes in big companies is
[28:12] the thing that is going to kill them
[28:14] relative to AI native companies. I am an
[28:17] A9 native company. Everything you're
[28:18] seeing right now is created in AI and it
[28:21] takes me a small amount of time to
[28:23] create it. This is just the easiest way
[28:25] to bring it to you. The organizational
[28:28] structure that makes a firm excellent at
[28:30] sector coverage is the structure that
[28:32] makes it slow at a cross- sector regime
[28:34] change. Those are the same property
[28:36] viewed from two directions. So if you're
[28:38] paying for research,
[28:40] I'm giving you something what at a
[28:42] minimum is different than what you're
[28:44] getting from other people. It doesn't
[28:46] mean it's better, but it's different.
[28:47] And I guarantee you it's coming with new
[28:50] stuff every week and new work every
[28:51] week. This is the proof of this is the
[28:54] NASDAQ returns year-to date 20%.
[28:58] The 100 name portfolio
[29:01] is up 46% 74 for the 25 name and 94 for
[29:06] the 10 names. And the 10 names are good
[29:08] 10 names. They're highly correlated. In
[29:09] fact, if you want the 10 names, I showed
[29:12] this on the week the subscriber webinar
[29:14] this week. Equal weighted average return
[29:16] 103. So I showed you 94. Um median
[29:20] return is 65. The best performer is up
[29:23] 249%. The worst one is up 16 and a half.
[29:27] That's Nvidia. It's outperforming the
[29:28] S&P by 13%. Eight of the 10 are up more
[29:30] than 40%. The average return is about
[29:33] 7.9 times the S&P's year-to- date
[29:36] return. The median is five times. From a
[29:38] valuation perspective with where we are
[29:40] right now, the average forward PEG ratio
[29:42] is 0.8. The median is 081. This is for
[29:45] the 10 names despite being up 103% year
[29:49] to date. This is all happening because
[29:51] earnings growth is growing rapidly. The
[29:54] median next year forecast for earnings
[29:55] growth 40%. Seven of the 10 names have
[29:58] expected earnings growth of over 30. Why
[30:00] would you go hunting for names?
[30:04] This is going to continue because the
[30:06] demand for compute is insatiable. the
[30:09] multiples can compress. And when we go
[30:12] through this, again, my job in doing
[30:15] this as someone who thinks in riskreward
[30:17] is to turn more negative when it seems
[30:21] like things are due for a consolidation,
[30:24] navigating the AI midcycle slowdown on
[30:27] June 8th.
[30:29] So, here's the prompt, guys. Take my
[30:32] transcripts, as many as you want. I did
[30:34] everything from May on to just highlight
[30:36] to you guys something. and you just take
[30:39] the prompt. It'll be uploaded this
[30:40] weekend. Its job is analyze a
[30:42] chronological series of weekly market
[30:44] commentary transcripts to determine how
[30:45] the speaker's long-term market
[30:47] conviction, short-term outlook, and
[30:48] willingness to take investment risks are
[30:50] changing over time. This is an unbiased
[30:52] thing. You guys can do it on your own.
[30:53] It's a prompt that goes through, and I'm
[30:54] going to show you the results. You can
[30:56] do this not with mine, but I'm sure
[30:57] there's other podcasts you listen to on
[30:59] a weekly basis. Go measure their tone.
[31:03] All of this is supposed to do is to
[31:05] trans. So what it does is on a
[31:07] structural basis I'm staying bullish the
[31:10] entire time. On a tactical basis it's
[31:13] shifting. So I believe computer is
[31:16] insatiable. I want to be in this trade
[31:17] for the next three to five years. But I
[31:19] want to be tactical and make sure my
[31:21] risk is lower when the tactical level is
[31:24] low and the riskreward is low. Selective
[31:27] risk reduction in May. So this is the
[31:29] risk posture is saying this is the cycle
[31:30] stage. I said we're in parabolic
[31:33] exhaustion in memory. Selective risk
[31:35] reduction. You guys can go through
[31:37] defensive selective risk neutral
[31:39] consolidation.
[31:42] Here's just some of the stages. This
[31:44] what I said on May 17th. I've been
[31:45] scaling out of things. I'm effectively
[31:47] as far as I'm concerned out of
[31:48] everything. Definitely out of all
[31:50] micron. So, if people call me up and say
[31:53] they bought in at May 17th or they did
[31:56] whatever, I'm trying to give you guys a
[31:59] riskreward on what's happening and
[32:02] trying to go capital preservation or
[32:04] capital rotation. Rotation with a
[32:06] genuine reduction in portfolio beta. I
[32:08] was talking about the need for
[32:10] deleveraging. The d-risking goes from
[32:13] partial to near total in the crowded
[32:15] leg. So again, if you don't believe what
[32:18] I'm saying, here's what this will do for
[32:20] you. So this is taking all of those
[32:23] numbers and I overlaid my thematic
[32:25] portfolio. So for those of you
[32:27] particularly the asset managers who I
[32:29] speak to regularly.
[32:31] This is a way for you to incept and see
[32:33] how I'm doing. So I start going into a
[32:36] more cautious. This is a yellow light.
[32:38] This doesn't mean bail out of
[32:39] everything. This means I'm getting more
[32:41] cautious. I'm reducing my micron. The
[32:44] red side is when it's like stop putting
[32:46] anything to work right now. I'm getting
[32:49] more worried. I'm seeing signs. This is
[32:51] what happened in those meetings. Every
[32:52] one of these dots is one of the videos.
[32:55] It's the transcripts. Then I start
[32:58] becoming back to neutral. And then the
[33:01] first time that I start saying that I'm
[33:02] buying stuff, which is when Micron broke
[33:04] below 900 is in here. And now you start
[33:08] seeing the green dots where I'm starting
[33:09] to turn more positive. So back in here,
[33:12] what I was saying is we need a
[33:13] consolidation. I was showing you
[33:15] consolidations, price and time. Even
[33:18] here, I was still saying we will never
[33:21] be able to pick a bottom. So, those of
[33:23] you who want me to try and pick this or
[33:26] this, nobody can do that and certainly
[33:29] not me. And I don't try to do that. What
[33:31] I want to do is reduce risk in here and
[33:34] then add back when I think there's been
[33:36] enough panic and enough consolidation.
[33:38] So I'm looking for a combination of and
[33:41] this is what makes these green there is
[33:43] I you start hearing me say I think we're
[33:45] closer to capitulation. I'm seeing the
[33:47] VIX go higher. I'm seeing this go on.
[33:48] I'm hearing about this. Then situational
[33:50] awareness happens and then all of a
[33:52] sudden I give a very positive sign on
[33:55] things and say I believe that's it. The
[33:57] riskreward is now shifted. You can use
[34:00] that create your own. And during that
[34:02] entire consolidation, I was creating the
[34:05] consolidation playbook to try and put
[34:07] people in a position. Two, three weeks
[34:10] ago, the week right before situational
[34:12] awareness, I put something out said
[34:14] combing through the rubble to give you
[34:16] guys the ability to go find stuff. So,
[34:18] here's what's happened. We've now
[34:21] already made back over 50% of the fall.
[34:24] This was a situational awareness day.
[34:26] So, again, my goal is not to pick you
[34:29] tops and bottoms. It's actually to help
[34:32] you navigate through to stay with the
[34:33] structural trend to minimize losses and
[34:36] maximize gains. Plain and simple. That's
[34:38] the name of the game. You maximize
[34:39] gains. You try to find things with a
[34:41] riskreward in your favor where you feel
[34:43] like things have been cleansed. And
[34:44] that's why that 50-day rate of change
[34:46] and some of the tools that I showed in
[34:47] the subscriber webinar, which I'm
[34:49] working on, which are algorithms to time
[34:51] the risk side. So, I'm taking what I
[34:53] just showed you and I'm incorporating
[34:55] all of that stuff into an algorithm to
[34:57] hopefully give you a score to show you
[34:59] where I think the risk is. For those of
[35:00] you who've never gone, I have my website
[35:02] now. I just wanted to show you this for
[35:04] one reason. Okay, it's visserlabs.com.
[35:06] Great. Here's my logo. That's a linear
[35:08] move. That's an exponential move.
[35:10] Academic over here, entrepreneurs over
[35:13] here. The main thing about this, when
[35:15] you go to my website, I want you to go
[35:16] there. Took me one hour to build. Took
[35:19] me about a day to work with network
[35:22] solutions to get it up where I secured
[35:25] the domain of visserlabs.com.
[35:28] But it took me literally about an hour
[35:31] to build the website. So I'm telling you
[35:34] this because when Network Solutions did
[35:36] a scoring on it, it got a hundred out of
[35:38] a hundred.
[35:40] If you use AI and you use it properly,
[35:44] all I did was put prompts in to build
[35:46] the site. It's not perfect. It still
[35:48] needs some changes, but it works, guys.
[35:51] You can build stuff like that. And
[35:52] that's gets back. My job here is to help
[35:54] you guys prepare for the future, not
[35:57] just an in AI to navigate this, to deal
[35:59] with the the parabas and the bubbles and
[36:02] everything that happens, but it's also
[36:03] to make you smarter and how to use it
[36:06] and to prepare you for this. The next
[36:08] constraint is financial crypto rails for
[36:10] AI agents. I'm telling you, mark my
[36:13] words. You guys want to reach out in a
[36:14] year when I'm wrong, it's going to be
[36:16] because of that. So, one company that
[36:18] reported Figure.
[36:21] I talked about Figure a lot with Anthony
[36:23] Pompiano this week. I highly recommend
[36:25] watching the video this week. I think we
[36:27] got into some interesting topics, but
[36:28] more importantly, my brain was kind of
[36:30] firing today on a lot of different
[36:32] things involving time, which is where my
[36:34] focus is. I'm spending a lot more time
[36:36] with the crypto community. I'm speaking
[36:38] at more um places. I'll be speaking at
[36:41] the Pomp event. I'll be speaking at a
[36:43] Bitcoin Policy Institute event in DC. A
[36:46] lot of things are going on. I probably
[36:47] did four had four podcasts released this
[36:49] week that had me speaking on it. We're
[36:52] at the stage now, guys, where I'll be
[36:54] doing more crypto stuff and you're going
[36:55] to be seeing more stuff. So, go to the
[36:57] site, go to the web page, but figure is
[37:01] a perfect example of how this is
[37:02] starting already. Figure released
[37:05] earnings consumer loan marketplace
[37:07] volume was up 4 was 4.3 billion, up 132%
[37:11] year-over-year. The tokenized onchange
[37:14] marketplace figure connect is now
[37:16] expected to approach 70% of volume. We
[37:19] are building the modern capital
[37:20] marketplace amidst the paradigm shift in
[37:22] the capital markets towards tokenization
[37:23] and standardization. This is happening
[37:26] now guys. This is why I built this. This
[37:29] is my ecosystem
[37:31] correlated with Bitcoin.
[37:34] This is the year-to-day performance. So,
[37:36] it's normalized. If I didn't normalize
[37:37] it, this would be a direct overlay.
[37:39] except you see the ecosystem is trading
[37:41] better than here. I've added some more
[37:43] names in here. I'm working with a bunch
[37:44] of crypto people. If anyone wants to
[37:46] reach out and they want to be involved
[37:47] in this process, I'm enjoying doing
[37:49] this. I need I have 10 verticals. I have
[37:52] about four names to five names in each.
[37:54] There's about 45 names in the index.
[37:57] Bitcoin is a representation of the S&P
[37:59] 500 of the future. I will continue to
[38:00] say that to anyone who wants to listen.
[38:02] Figure is part of this.
[38:06] Humans will be rounding error on the
[38:07] internet. Bot traffic will exceed human
[38:09] traffic by a factor of 1,000 within five
[38:11] years. Elon Musk on cloudfire's forecast
[38:15] AI agentic internet traffic will
[38:17] obviously vastly exceed human usage.
[38:19] Guys, go make money on that.
[38:21] Transactions are going to happen. Stable
[38:24] coin card spending volume surge another
[38:26] 16%. 16%.
[38:29] You want to do an ARR on that?
[38:35] US bank regulator opens national
[38:37] charters to Bitcoin and crypto firms.
[38:40] The SEC is about to cook while Congress
[38:41] is stuck arguing over the Clarity Act.
[38:43] Just like AI, nothing stops crypto,
[38:47] guys. It will continue. Even if it's
[38:49] not, the Clarity Act's not done. So, if
[38:51] you're waiting for the Clarity Act or
[38:52] it's thinking there's going to be a sell
[38:53] the news moment on this, I'm telling
[38:56] you, a year from now, we are going to
[38:57] have a volume explosion on this stuff.
[38:59] We don't know where all markets are
[39:01] going to be, but I do know that volumes
[39:02] will be expanding. financial guardrails
[39:05] will be connected. I wrote this and
[39:06] released it both on Substack this week's
[39:08] and my subscriber website. I released it
[39:11] because it's something I believe is
[39:13] really important. This is around the yen
[39:16] intervention.
[39:18] It is a connection of everything
[39:20] happening between AI mac uh the macro
[39:23] side in terms of the debt, the deficit,
[39:26] Japan's debt, the US debt. This goes
[39:28] through all of it. It is meant to
[39:30] realize that in my opinion, we will look
[39:32] back on the week of situational
[39:33] awareness where we had the dollar yen
[39:35] intervention and we had Kevin Worsh
[39:37] basically losing credibility according
[39:39] to people
[39:41] and Lewis Gau put this piece out this
[39:43] week like me. He released it to
[39:45] everyone. He does not do that. I don't
[39:48] do that for my subscribers stuff, but
[39:50] that's how important um the week was for
[39:52] me. And then he gets into something
[39:54] equally as important on the exact same
[39:56] theme, which is why I wanted to show you
[39:57] guys. But this is the most important
[40:00] thing. And he said, "AI and stable coins
[40:02] could reshape rather than simply destroy
[40:04] dollar dominance." Gavka raises the
[40:05] provocative idea of a compute dollar
[40:08] replacing the petro dollar. That word
[40:10] compute again. Insatiable demand for
[40:13] compute, an insatiable race for both
[40:16] compute and owning the crypto landscape.
[40:20] AI is becoming more than a technology
[40:22] story. computers evolving into an
[40:23] industrial resource, a geopolitical
[40:25] asset and potentially a pillar of the
[40:27] monetary system. This is on the other
[40:30] side of the debt world where we just
[40:32] continue to do whatever is necessary to
[40:34] keep the debt going and keep spending.
[40:36] Bessence, whatever it takes, yenval mass
[40:38] limited firepower. This is the reason
[40:40] it's so important because they whether
[40:43] it's yield curve control, whether it's
[40:45] raising the cap on FEMA and them getting
[40:48] invol, it doesn't really matter. Scott
[40:50] is up against it. He didn't create this
[40:53] problem, but they're trying to race out
[40:55] of it. And the problem is they're not
[40:56] getting the receipts and they're
[40:57] spending their way out. So, the debt and
[40:59] the deficit are just not getting better
[41:02] fast enough, even if they are getting
[41:04] better at all. Wall Street sees a bond
[41:06] market angst and we're seeing that the
[41:09] bond market just won't rally. We had a
[41:11] week where the inflation data surprised
[41:13] on the downside. Retail sales missed. We
[41:16] had a week before where the jobs number
[41:18] missed. And yet long-term yields are
[41:21] just not coming down. They're not going
[41:22] up either, but they're not coming down.
[41:25] And the yen is also not moved much since
[41:28] the intervention. And that's despite the
[41:30] fact that now we're down to 31% for
[41:33] September for Wor
[41:36] and 80% for the BOJ. These numbers were
[41:39] about 5050 each before the payroll
[41:42] numbers. Now we're at here and yet we're
[41:45] still not seeing any change. Now, I
[41:46] happen to believe this is a mistake by
[41:48] the markets and you should be buying the
[41:49] the basement trade. You should be buying
[41:51] gold, buying silver, buying Bitcoin. But
[41:53] regardless of that until we start seeing
[41:55] yields come down, it's just not
[41:57] happening. So, the Worsh attacks
[42:00] continue. And he's going to be speaking
[42:02] at Jackson Hole. This came out from Bill
[42:05] Dudley. Worsh implied that the PCE index
[42:09] could be jettisoned as the Fed's primary
[42:11] inflation metric as soon as January. Who
[42:14] knows come after January he might say
[42:16] about a strategy he said I suspect the
[42:18] task force might have something that
[42:20] the reason I wanted you guys to read
[42:22] this this is academic this is exactly
[42:25] the problem this is what Kevin Walsh
[42:27] that's what that whole paper I wrote is
[42:29] that the academic fed
[42:31] does not exist you cannot focus on the
[42:33] academic fed this be bank of America
[42:37] after the CPI is still calling for three
[42:39] right hikes this year
[42:42] holding firm
[42:44] At least Canro understands. Okay, so
[42:47] here's what the Fed is saying. So this
[42:50] is Fed speak chart overlaid with a
[42:53] leading indicator which is the labor
[42:55] plus inflation surprise index which has
[42:58] now come down. We'll see if this goes.
[43:00] But this is the reason why at this point
[43:03] we haven't seen rates come down. So the
[43:04] question is is Walsh going to give any
[43:06] guidance or do anything that helps?
[43:09] two-year rates have come down just
[43:11] because the probabilities have come
[43:12] down. They did break the 50-day moving
[43:15] average after being above it for one of
[43:17] the longest times in the last decade.
[43:19] We'll see if it goes. But this is why I
[43:22] wrote this. The academic Fed, a secular
[43:25] regime shift, it's far more powerful. It
[43:27] usually reflects dramatic changes in
[43:28] monetary policy, fiscal policy, and
[43:30] innovation. I believe we were in the
[43:31] early stages of one. That is what the AI
[43:33] trade is within that paper. Despite or
[43:37] what your favorite Oh, this is inside my
[43:39] uh other paper. Despite what your
[43:40] favorite economist may tell you while
[43:42] looking through a historical lens and
[43:43] assuming the old relationships still
[43:45] hold, something has changed. AI is
[43:47] already disrupting the labor market and
[43:49] the rise of AI agents is only beginning.
[43:52] If you look at the labor market
[43:53] contextually, it is very weak. That gets
[43:55] back to the chart I showed you with the
[43:57] S&P earnings relative to year-over-year
[44:00] labor.
[44:02] So again,
[44:03] research firms are b built vertically.
[44:06] Academics seldom change their view.
[44:09] That's what entrepreneurs do. That's
[44:11] what traders do. Hopefully that's the
[44:14] way you view the work that I do. I like
[44:17] to be bearish equally as bullish if I
[44:20] think it's going to save me money. You
[44:22] can't get away from this divergence. So
[44:24] if you're an academic, highlight this
[44:26] and tell me why this isn't a fair way to
[44:28] look at things. The labor market is very
[44:30] weak. while we are growing earnings at
[44:32] 32%. We've never seen anything like this
[44:34] before and we're not coming out of a
[44:37] recession. Here's more in the labor
[44:39] market. I showed this last week. I just
[44:41] want to re-emphasize it. I don't see
[44:43] anyone doing this. This is the sixmonth
[44:45] rate of change on the aggregate weekly
[44:46] payrolls, which takes into account
[44:48] weekly hours, the wage growth, plus a
[44:51] number of job hires. A much broader and
[44:53] better way of looking at the payroll
[44:54] market. And we're at the weakest
[44:56] six-month rate of change since 12, 2012,
[45:00] 14 years ago. Here is the job creation
[45:03] right now for non-farm payrolls. X
[45:06] healthcare and leisure. I don't like
[45:09] doing X things, but I do want to show it
[45:11] as these things are both related to old
[45:13] people that the boomers who have money.
[45:16] So health care and travel, that's great
[45:19] except for one thing. Those are the jobs
[45:20] that can't be replaced by AI.
[45:23] So you have a very difficult situation.
[45:25] Here's the non-farm payrolls again
[45:27] year-over-year. So we're sitting down
[45:28] here because
[45:30] of AI agents.
[45:34] This is the weekly ADP numbers. So we
[45:37] had a jump here. This is when the Fed
[45:40] speak changed. This is when you saw the
[45:43] December rate hikes be built in. Well,
[45:45] now the non-farm the ADP numbers just
[45:47] keep weakening every week.
[45:50] We're not adjusting anything. Here's the
[45:52] labor force participation rate, which
[45:54] just continues to go down. If we're
[45:56] losing labor supply,
[45:59] that's 10ear rates overlaid with it. Why
[46:01] aren't wages going up? I saw someone
[46:04] write something this week. Look at this.
[46:05] An uptick. Are you kidding me? So, the
[46:08] Atlanta the Atlanta Fed wage thing, this
[46:11] is a hookup. The hourly earnings we just
[46:14] saw from the uh non-farm payrolls
[46:17] dropped to 3.2%.
[46:19] The Atlanta Fed is at the same level it
[46:21] was before 2020. So is the hourly
[46:24] earnings. So is the quits rate.
[46:27] Give me something that shows. We know
[46:29] that if we're not getting wage growth
[46:30] that is significant to offset the weekly
[46:32] hours not increasing and the fact that
[46:34] we're losing labor force participation
[46:36] rate and we're not hiring any people. I
[46:39] don't see why the Fed should be raising
[46:40] rates especially not three times. I
[46:42] don't even understand what that is. Um
[46:44] if their mandate is labor and inflation.
[46:46] So let's go through the second part of
[46:47] it. the inflation data. So, we're going
[46:50] to follow Bill Dudley on this PCE thing.
[46:52] I think Kevin Worsh has it right. Here's
[46:55] PCE all on the same scale, guys. PCE
[46:57] core and here are the other core
[46:59] numbers. Plus,
[47:02] this doesn't include true inflation. So,
[47:04] I left them out of it just because it's
[47:06] way down here below two. all of these
[47:09] other gauges, whether it's the sticky,
[47:12] the Atlanta sticky CPI core, which is at
[47:15] 2.5, the CPI core, which is 2.6, the
[47:19] Dallas trim mean, which is 2.2. They
[47:22] just made a change in the way they're
[47:23] calculating it, 2.4, whatever. Cleveland
[47:25] 2.7, only the PCE core, that one point
[47:29] is up there. Sounds like it's not a good
[47:32] one to use. Here are the inflation swaps
[47:34] relative to core CPI, lower. Why aren't
[47:37] these being talked about as mattering? I
[47:40] took all of those ones I just showed you
[47:42] on the prior screen, not the swaps, the
[47:45] one before this, and I put them as one
[47:48] index. That's what this number is. That
[47:50] was taking those four and averaging them
[47:52] together. That's 2.5. This is PCE core.
[47:54] This is it over time. There's two things
[47:55] to note. Number one, during the prior
[47:59] from 1990
[48:01] to 2020, the white line stayed above the
[48:05] orange line all the time.
[48:09] So if you normalize this and again
[48:11] contextually this number should be down
[48:14] here based on history, something has
[48:17] changed in this calculation.
[48:19] Kevin, you want me to work on the Fed,
[48:21] give me a call. I'm happy to help you
[48:23] guys with some of this stuff to make
[48:24] sure we're not making decisions based on
[48:27] uh people on Wall Street academics
[48:29] telling you what you should do. Real
[48:31] rates, the white line here. Real
[48:33] two-year yields are up here. Here's the
[48:35] debasement trade. This is gold, silver,
[48:36] and Bitcoin. And the dollar, the dollar
[48:40] has already responded. It was down again
[48:41] today. This is before today. So, it's
[48:43] already down here at new lows. Gold has
[48:46] moved. Silver has moved. Bitcoin has not
[48:48] moved yet. That's the green line here.
[48:49] It's still sitting there. Eventually, it
[48:52] will. I don't know when. We'll see.
[48:55] Academics ask, "When does the evidence
[48:57] tell us has happened under comparable
[49:00] conditions?" Markets ask, "What's going
[49:02] to happen before everyone else realizes
[49:03] it?" This is the money-making side.
[49:05] Entrepreneurs ask, "What becomes
[49:06] possible now that wasn't possible
[49:08] before?" That is the AI native world.
[49:10] That is the world we speak about on
[49:12] Visser Labs. Now, finally, China added
[49:15] 20 tons of gold in July, its largest
[49:17] monthly purchase since October 2023.
[49:20] China's central bank adds gold to Hong
[49:23] Kong as well. And finally, the Bank of
[49:26] Korea resumes gold investment after 13
[49:28] years. Korea is back in the game. AI is
[49:31] becoming more than a technology story.
[49:32] Computers evolving into an industrial
[49:34] resource, a geopolitical asset, and
[49:35] potentially a pillar of the monetary
[49:37] system. Thank you, Lewis Gav. Thank you
[49:39] to all of you guys. I will see you maybe
[49:43] from New York next week, maybe still
[49:45] here. I hope it's still here, but I
[49:47] don't know.