# Jordi Visser: The AI stocks that win the next decade

https://www.youtube.com/watch?v=jhXV40oz7OQ

[00:00] Jordi, I want to take us to your paper
[00:02] you wrote recently about how the new age
[00:05] of investing is about your CapEx is my
[00:08] opportunity. That's a play on Jeff
[00:10] Bezos's your margin is my opportunity.
[00:13] What's going on here?
[00:15] [laughter]
[00:15] Um
[00:17] All right. So
[00:20] there's two themes that come out of
[00:21] this. Number one
[00:23] uh really post the great financial
[00:25] crisis uh the alpha that was generated
[00:28] by all investors
[00:30] really from 2009 on
[00:32] was isolated into effectively the Mag 7.
[00:35] Those are the winners and
[00:37] Jeff Bezos famously said, "Your margin
[00:39] is my opportunity." And this was really
[00:42] about the concentration of power being
[00:45] isolated into software companies.
[00:48] Commodities didn't work during the 2010
[00:51] to 2020 uh
[00:53] decade. Uh
[00:55] bonds didn't work. You ended up having
[00:57] negative real yields and or negative
[01:00] yields overall. And then we ended up
[01:02] getting COVID, we got inflation. So
[01:04] bonds basically have not performed. You
[01:07] had a little bit of performance
[01:08] obviously from the early stages of VC,
[01:10] but then that's run into trouble. And
[01:12] then private equity's run into trouble.
[01:14] The the whole point of of
[01:17] the original part was just that we were
[01:20] in a a decade where I don't think people
[01:23] realized how concentrated the alpha was.
[01:25] So now you fast forward.
[01:27] And Jensen Huang comes out earlier this
[01:29] year and basically tells everyone
[01:32] okay, we're at the stage now where we're
[01:33] going to build AI factories. And those
[01:35] AI factories um it's going to take 90
[01:37] trillion dollars to reindustrialize
[01:39] through intelligence. So when Jensen
[01:41] Huang, who arguably not arguably is at
[01:44] the
[01:45] he's the centerpiece that sees
[01:47] everything. Whenever you see any of
[01:48] those circular flows, Jensen Huang is
[01:51] involved no matter what in terms of what
[01:53] he's got cuz he had GPUs and they still
[01:57] today are the most necessary part of the
[01:58] AI build out. So, when he says it's
[02:00] going to be 90 trillion over the next 10
[02:02] to 15 years, and everyone is like
[02:05] focused on the data center side,
[02:08] the winners of this particular 10 to
[02:11] 15-year period are going to be hardware
[02:13] companies. And so, the whole point was
[02:15] at the beginning of the year, the models
[02:16] had reached a point where the agentic
[02:18] world was here. Jensen Huang saying 90
[02:21] trillion, you need to have the agentic
[02:23] world here. The agentic reaching the
[02:25] agentic stage, reaching IQ to a level of
[02:27] say 130, where all of a sudden the
[02:30] coding side that Opus 4.5 brought us
[02:32] into allows us to expand artificial
[02:35] intelligence to a very high level
[02:36] because of the agentic side. We hit
[02:38] recursive self-improvement, then we hit
[02:39] AGI, then eventually we hit
[02:41] superintelligence. All of those get us
[02:43] to humanoids and everything else. So,
[02:45] think of
[02:46] Opus 4.5 as being the gateway to the
[02:48] agentic world, and then Jensen Huang
[02:49] says, "We're here. It's going to be 90
[02:51] trillion dollars to basically
[02:52] manufacture tokens to feed the AI
[02:55] agents."
[02:56] So, a lot of this thesis, from my
[02:59] understanding, is we are moving into the
[03:02] physical AI stage of things.
[03:06] Was there something specific you can
[03:07] point to in the last few months
[03:10] that convinced you that we're moving
[03:12] into that stage right now?
[03:15] Honestly, I started writing about
[03:17] inference in May of last year.
[03:20] It
[03:21] Inference is really the beginning of the
[03:25] agentic AI and the need for all the
[03:26] hardware.
[03:27] And I say May of last year because
[03:29] that's when the companies, particularly
[03:32] the most famous one was when Google
[03:33] openly started talking about how many
[03:35] what the token usage was. When the token
[03:37] usage started to get higher, it means
[03:39] more and more people were using it. It
[03:40] was getting more powerful.
[03:42] Now, when DRAM started to go up, really
[03:45] beginning in September,
[03:48] that's when we started to see the memory
[03:49] stocks truly take off and start to go on
[03:52] their cliff run.
[03:54] And I think at that point Silicon Valley
[03:57] and the people around the world realize
[03:58] we were there. The one thing people have
[03:59] to realize
[04:01] these models that get released,
[04:04] they've had them for a long time. I
[04:07] mean, the speculation is that they're
[04:08] always 6 months ahead of frontier model
[04:10] companies before they release them.
[04:13] And it's probably going to be even a
[04:15] little bit longer now based on what
[04:17] happened with Mythos, meaning once you
[04:19] get to the point that these models are
[04:20] so powerful, they're probably going to
[04:21] be delayed more or the government will
[04:23] get involved, which means they're going
[04:24] to be delayed even more. Maybe they
[04:25] never get released because they don't
[04:27] want competitors to see what they are.
[04:29] Um I think that period last year
[04:32] triggered the beginning, but it was Opus
[04:34] 4.5 that I've I've written about papers.
[04:37] That's when we saw the software stocks
[04:39] really start to get killed. That was the
[04:41] the the point that I think everyone who
[04:43] is using those models and famously Andre
[04:46] Karpathy turned from we won't be dealing
[04:49] with AI agents
[04:51] for a decade.
[04:53] He said it was going to be a gradual
[04:55] thing.
[04:56] This went from gradual to
[04:59] just parabolic. And now Andre Karpathy
[05:01] is actually working for Anthropic. So,
[05:03] you can you can kind of pick the October
[05:05] to November period when of AI agents.
[05:10] I I think that makes sense. Um
[05:12] I want to ask you about this framework
[05:14] that you laid out about the AI economy
[05:16] as a five-layer cake. And uh I have it
[05:20] written here the applications and
[05:22] workflows are at the top followed by the
[05:24] AI models and platforms, then you have
[05:26] data infrastructure, chips, and compute,
[05:28] and then at the bottom you have energy,
[05:30] hardware, and commodities.
[05:32] My understanding is that you want to be
[05:34] investing at the bottom of that cake. Is
[05:36] that the right way to think about it?
[05:38] I'd love to take um credit for creating
[05:41] the five-layer cake, but that's Jensen
[05:42] Huang. So, that's just me writing about
[05:45] what he has said specifically. Um yes,
[05:49] you've over the course of the last, you
[05:52] know, year. And I'll say year just
[05:54] because it's a year ago that Micron
[05:56] really started to break through levels
[05:58] and we started to see
[06:01] the Western Digitals, the SanDisks, and
[06:03] all this stuff just start to go up. And
[06:04] I think that was the recognition that we
[06:06] had reached a point that the five-layer
[06:09] cake, which is really going to be the
[06:10] part that will drive the IQ side. You
[06:13] don't get into the application side
[06:16] until the IQ gets high enough. You don't
[06:17] get into the AI agent side until the IQ
[06:19] gets high enough. So, beginning a year
[06:22] ago, we were, you know, the only pieces
[06:25] that were worth focusing on and even up
[06:27] till now, in in June of this year, you
[06:30] want to be on the bottom three, which is
[06:32] the energy chips, and the infrastructure
[06:35] for the energy side, you know, you can
[06:38] pick anything you want out of it. You
[06:40] can pick Bloom Energy, you can pick some
[06:42] of the battery companies, you can pick
[06:44] some of the natural gas transformer
[06:46] companies, anything that goes into the
[06:47] power. For the chips, it's
[06:48] self-explanatory. It's either
[06:50] semiconductor chips, uh like the
[06:52] Nvidias, the Intels, the Texas
[06:54] Instruments, or you've got the memory
[06:56] side. And then for the infrastructure,
[06:58] think Dell, think HPE, think Pure
[07:00] Storage, think places that are basically
[07:03] going to be part of the AI factories.
[07:06] In all of that
[07:07] uh
[07:08] framework, the Mag 7 are not really
[07:10] included except Nvidia, it sounds like.
[07:13] Is that uh am I wrong to say that?
[07:16] Yeah, the the most of the the the Mag 7,
[07:20] um
[07:20] let's leave Tesla out of it cuz Tesla
[07:22] will benefit when we get into the
[07:24] embodied side, which would be the
[07:25] applications. But let's just say the
[07:27] model side. Um
[07:31] The problem is these are the spenders.
[07:33] So, I've had a
[07:36] Okay, I don't want to say negative,
[07:38] meaning they're going to go down, but
[07:39] these companies are going to go through
[07:40] multiple compression, in my opinion. Uh
[07:43] they become hardware companies. They're
[07:44] spending lots of money. They're they're
[07:46] issuing stock and they're taking out
[07:48] debt.
[07:49] Uh they have to make the revenues in the
[07:50] future.
[07:52] The
[07:53] biggest comparison I can get, not that
[07:55] it's going to end this way,
[07:57] is the fracking boom. Um I've talked
[07:59] about this before where
[08:02] these companies are making a huge bet
[08:04] that they're going to benefit
[08:06] from what is going to be a
[08:07] commoditization of intelligence.
[08:10] And maybe they'll get the money in,
[08:12] maybe they won't. Maybe they'll get a
[08:14] lot of it in, but not enough. That
[08:16] uncertainty and all the issuance and all
[08:18] the things uh the bottlenecks that are
[08:20] coming, I just think they're going to
[08:22] see multiple compression. I'm not sure
[08:24] companies of that size should have a 30
[08:26] PE anyway. Um Microsoft was a
[08:30] single-digit PE in the early or the
[08:33] early 2010s. So, my guess is that this
[08:36] is going to be a multiple compression
[08:37] story. We've definitely seen that with
[08:39] Nvidia already where their PE has
[08:41] declined from the 40s down into the low
[08:43] 20s. And if you look at it a year
[08:45] further, it's in the teens. So, I think
[08:47] all of the hyperscalers and the Mag 7s
[08:49] are going to run through the same thing.
[08:52] And a lot of this is also what you refer
[08:55] to as benchmark arbitrage, right? You
[08:57] have these indexes like the S&P 500 or
[09:00] MSCI World that are essentially weighted
[09:03] to the last decade of out-performers
[09:05] like the Mag 7.
[09:07] But the way that my guess is that uh
[09:11] you're thinking about it as an investor
[09:13] is finding the weightings for the next
[09:15] decade and then trying to build that uh
[09:18] into your framework.
[09:20] The whole game of of out-performing
[09:22] benchmarks is to figure out what the
[09:24] weightings of the benchmarks are going
[09:25] to look like in the future.
[09:27] So, whether you're a mutual fund,
[09:29] whether you're a pension fund who's
[09:30] benchmarked to MSCI, uh mutual fund
[09:33] who's benchmarked to maybe the S&P,
[09:36] maybe a growth index, whatever. But then
[09:39] also RIAs and wealth managers that are
[09:41] managing for people, the whole benchmark
[09:43] arbitrage was to get people to recognize
[09:45] that there's two ways that you're going
[09:48] to end up with the proper weightings.
[09:50] One is by just being long these passive
[09:54] indices and waiting for the weightings
[09:56] to catch up, which in my opinion will
[09:58] will will probably be
[10:03] a disappointing performance
[10:06] uh relative to what you're watching on
[10:08] the sidelines uh as Micron goes up 10
[10:11] times, as Sandisk All of these stocks
[10:13] have gone up three to 10 times. So, my
[10:15] argument has been for people that uh are
[10:18] benchmarked to these passive indices
[10:20] that they should be at least
[10:22] overweighting
[10:23] the names. And I created 100 names. And
[10:27] those 100 names are not all in the S&P
[10:29] 500, they're not all American companies,
[10:31] but they're all driven by the exact same
[10:34] thing, which is the build out in AI.
[10:36] They're in the power side, they're in
[10:37] the chemical side, they're in the uh
[10:39] semiconductor in the packaging side,
[10:41] they're in the infrastructure side.
[10:42] They're all different types of ones that
[10:44] fit in and people are not going to think
[10:46] of chemical companies as being
[10:47] important. But to me, chemical companies
[10:50] are like the perfect example of
[10:51] chemicals are going to be more GDP
[10:52] intensive than oil's going to be. Than
[10:55] oil was for the last 200 years. Oil was
[10:58] a transport thing, it was about
[10:59] globalization. Chemicals is about the
[11:01] build out and the bonding and the
[11:03] necessity for all of this electricity.
[11:05] It's a different world and I think
[11:07] people need to realize that when changes
[11:09] happen in a linear fashion, you don't
[11:11] think about benchmark arbitrage. But
[11:13] when it happens in an exponential
[11:14] fashion, you can end up with an S&P 500
[11:17] that's unchanged. But if you would
[11:18] invested in these 100 companies, they
[11:20] might be up 50% a year. My basket has
[11:22] outperformed the S&P significantly this
[11:24] year and this is the reason why I've
[11:26] taken this approach with kind of
[11:28] deciding not to work for a company and
[11:32] be an AI power user and then to spread
[11:34] the world the the word to people out
[11:37] there. It's really been broken down
[11:38] where it's signal, alpha,
[11:41] and agency. The signal is about making
[11:43] sure they're up-to-date on everything
[11:45] happening with AI. It is a
[11:48] X is filled with doomers. You've got the
[11:51] alpha side, which names, if I believe in
[11:54] AI, what names will make that benchmark
[11:56] arbitrage? And then the agency side is
[11:58] making sure that I'm helping people use
[11:59] the tools cuz if they use the tools,
[12:01] they're more likely to believe in the
[12:03] stuff that is the signal and then
[12:06] likely to be looking for financial
[12:08] empowerment by investing in the alpha.
[12:11] What do you make of
[12:13] investors and let's say veteran fund
[12:15] managers like yourself who have, let's
[12:18] say
[12:18] [snorts]
[12:18] the same years of experience as you, but
[12:21] have fallen on the totally opposite side
[12:23] as far as a world view on AI right now?
[12:27] Uh
[12:28] you know, I I think to each his own on
[12:30] the way that they want to go through
[12:32] this. Um you know, I I believe the world
[12:35] is and predicting the future is
[12:39] uh a world of distributions and
[12:42] the probability of anyone being right to
[12:45] the end is zero. You know, I thought
[12:48] Bitcoin would be five or six times
[12:49] higher than it is today.
[12:51] So, I got that one wrong. I got Micron
[12:53] right.
[12:55] Um there's plenty of things that we all
[12:56] get wrong, we all get right. So, I leave
[12:58] people to their AI views there. I I'm
[13:00] I'm not a preacher. Uh I'm not telling
[13:03] people what they should do.
[13:05] But, I do believe the evidence is you
[13:07] can't make a informed decision if
[13:10] they're you're not using AI. And so, my
[13:11] message to people that are around my
[13:13] age, you know, nearing 60,
[13:16] if you believe that you know whether AI
[13:19] is going to make it and that view is
[13:21] coming solely from your experience of
[13:23] the past of I've seen technology I saw
[13:25] the dot-com bubble I saw
[13:28] I think you're making a huge mistake.
[13:30] If you use it and you're not amazed by
[13:33] it. And when I say use it, you know, I I
[13:36] released something on my paywall this
[13:38] week, which was showing people how to
[13:39] make a knowledge brain.
[13:41] And the response I got from it was
[13:44] overwhelming. I'm actually I'm actually
[13:47] uh pleasantly surprised that I finally
[13:48] put something up. I've put up prompts.
[13:50] I've given people videos.
[13:52] But this was the first time that I gave
[13:54] them a beginning to and go create a
[13:55] knowledge brain. And for the people that
[13:57] did it and more importantly that then
[13:58] had their kids do it,
[14:00] they basically said they immediately
[14:02] understood the difference. Now, I
[14:04] suggested they go we're getting into the
[14:06] fantasy football season.
[14:08] Um
[14:09] like to just verbally say this for your
[14:12] show, for anyone who does fantasy
[14:14] football is going to be in drafts,
[14:15] you're all listening to some podcast or
[14:18] going on some website. Pick your five
[14:21] favorite analysts.
[14:23] And then
[14:24] go
[14:26] to the paywall, get my knowledge brain
[14:27] stuff, and then go connect to all of the
[14:31] times that they give interviews. If they
[14:33] do a weekly podcast, just take the
[14:34] transcript from YouTube and upload it.
[14:37] All of this stuff happens in a matter of
[14:39] 30 minutes. And then you can ask
[14:41] questions about each of the players that
[14:43] they've talked about over each of the
[14:45] interviews from the beginning of the
[14:46] year. They're talking about the drafts.
[14:48] They're talking about the players, where
[14:49] everything stands. It's really hard to
[14:51] keep up on fantasy football. I remember
[14:53] when I was doing it actively. I can't do
[14:55] it anymore cuz I don't have the time,
[14:56] but these are the ways that I think for
[14:58] people who are I don't even want to say
[15:01] bubble people, doomers, whatever you
[15:03] want to call it. If you're not informed
[15:05] and you're not using it, I don't think
[15:06] you have a right I don't I don't think
[15:08] you can possibly have a high probability
[15:10] of being correct if you don't know what
[15:11] it is.
[15:13] Today's episode is sponsored by Public,
[15:16] and they just launched a new product
[15:17] called Agents, and it's honestly one of
[15:19] the most compelling uses of AI in
[15:21] investing I've ever seen. The idea is
[15:24] simple. A lot of investing is
[15:25] repetitive, monitoring markets, moving
[15:27] cash, managing risk, rebalancing
[15:29] positions. It's the kind of stuff
[15:31] investors constantly have to stay on top
[15:33] of. Public's new agents product lets you
[15:35] automate those workflows using plain
[15:37] English. So, instead of manually
[15:39] watching charts all day, you can create
[15:41] an agent by typing something like, "If
[15:43] the VIX hits 25, buy a weekly put option
[15:46] on SPY." Or, "If my cash balance goes
[15:49] above 20,000, move the excess into my
[15:52] direct index."
[15:53] [music]
[15:53] Public's AI then turns that into an
[15:55] automated workflow that you can
[15:56] customize, review, and activate directly
[15:59] in your portfolio. Public is essentially
[16:01] building what they call an agentic
[16:03] brokerage, a platform where AI can help
[16:05] automate the operational side of
[16:07] investing based on rules and strategies
[16:09] you define yourself.
[16:11] [music]
[16:11] If you want to start automating ideas
[16:13] inside your portfolio, visit
[16:14] That's [music] public.com/openingbell.
[16:20] Now, let's get back to the show.
[16:22] You and I are both in the not a bubble
[16:24] camp. So, I to start with that, uh but
[16:28] if you
[16:30] sort of canvas the the views of the
[16:32] bears, is there any
[16:35] bubble arguments that stand out to you
[16:37] that you say, "Wow, that actually makes
[16:39] a lot of sense, or that could be pretty
[16:40] credible?"
[16:43] Um the only And again, I have to fight
[16:45] my own
[16:47] So, we all have
[16:49] bubble PTSD.
[16:52] I've seen plenty of times in my career,
[16:54] having been involved in the markets, and
[16:57] remembering 1987 when I wasn't in the
[16:59] markets, but I remember what it did to
[17:01] my town. I remember the people who
[17:03] worked in the industry and what they
[17:05] went through, and the people that lost
[17:06] their jobs.
[17:08] I've read historically all the books
[17:10] that everyone else has read. It's really
[17:11] hard to consume all of that knowledge,
[17:14] to listen to your grandmother talk about
[17:16] not having food during the Great
[17:17] Depression and and needing to leave the
[17:19] house when she was 16 because they
[17:20] didn't have enough money and not have it
[17:22] impact the way that you view things like
[17:25] everyone in Korea owns the two stocks
[17:27] over there or two stocks are driving
[17:29] like
[17:30] the part about the the bubble that I
[17:32] think is is real is that there are no
[17:35] free lunches in investing.
[17:37] Um, you don't just get to show up and
[17:39] buy call options and make money all the
[17:41] time.
[17:42] Uh, I don't think that's what retail
[17:44] traders do but the reason that I
[17:46] publicly said that I got rid of Micron
[17:48] and trust me it got up to 1,100
[17:50] yesterday. I got out in the 600s.
[17:54] So that means it's almost doubled since
[17:56] I got out which doesn't feel good but I
[17:58] knew that that was the most likely
[18:00] scenario to happen cuz I'm not picking
[18:02] the top of a bubble the same way that
[18:04] I'm not going through it. I don't view
[18:06] Micron as a bubble. I view it as I had
[18:08] better places to put money that over the
[18:09] course of the next 2 years I think I'm
[18:11] going to make more money being invested
[18:12] in these other names. But part of the
[18:14] reason was because what I was reading
[18:16] from Korea, what I was hearing from
[18:18] investors, the pushback that I was
[18:19] getting a year ago and now how all hedge
[18:21] funds were embracing it, it just felt to
[18:23] me like too many people were involved
[18:25] and it was no longer the same risk
[18:27] reward and it was still being bet as the
[18:29] favorite. So I think if there's one part
[18:31] that I would agree with the people on
[18:32] the bubble side
[18:34] is that recently it's gotten a little
[18:36] too easy and there are no free lunches
[18:38] in investing.
[18:42] First of all, I uh I still own Micron
[18:45] and and uh the reason I got into Micron
[18:48] of course was cuz you and I had a
[18:49] conversation in January of last year
[18:52] when Micron was probably 90 bucks and
[18:54] you said that was your favorite stock at
[18:55] the time. And uh I published that to my
[18:58] subscribers. A lot of people have been
[18:59] in Micron that follow me for a long
[19:01] time. Uh so thank you for that first of
[19:04] [laughter] all. Um
[19:05] I I guess when you uh
[19:08] think about your own experience and your
[19:10] years in the market is there anything
[19:13] you can point to specifically that you
[19:14] think gives you an edge in how you're
[19:16] looking at the AI economy and the AI
[19:19] trade that you think is differentiated?
[19:22] I honestly, it's it's using AI. Um
[19:29] Today is a Tuesday.
[19:32] Um
[19:33] So
[19:35] to take you through what I did from
[19:38] 6:00 until 11:00 this morning.
[19:44] I start with a walk through Brooklyn.
[19:47] And this is almost it depends on whether
[19:50] I'm doing workouts in the morning in the
[19:51] very early morning or or later on, but
[19:54] on on most Tuesdays I go for a walk and
[19:57] I'll do usually about a mile and a half
[19:59] sorry, 3 to 4 mile walk.
[20:02] And during that walk I'll listen to
[20:04] podcast.
[20:05] So at 1.75 speed and I'll start
[20:07] consuming information. Now, during the
[20:09] walk something will hit my my brain that
[20:13] I'm listening to. Now, most of these
[20:15] podcasts are they're ones that come out
[20:18] two or three times a week or at least
[20:20] weekly.
[20:21] And on this particular day as I was
[20:24] walking
[20:25] I was able to come up with the
[20:27] inspiration for four papers.
[20:29] Four.
[20:31] Uh one of them involving health.
[20:34] Three of them involving AI. So, what had
[20:37] transpired in the time from that, I will
[20:39] pause the podcast I'm listening to.
[20:42] I will switch over to one of the models.
[20:45] It's usually ChatGPT. Uh and part of the
[20:48] reason it's usually ChatGPT is because
[20:50] the voice thing is just so much easier
[20:52] for me on ChatGPT. I'm not really sure
[20:54] why. Anthropic doesn't have the amount
[20:56] of compute. So, unless my service is
[20:58] com-
[20:59] per- picture perfect Anthropic is really
[21:02] slow and the voice thing is not as good
[21:04] for me as uh either Gemini or ChatGPT
[21:09] Chat GPT is the best. During that period
[21:12] I will have a conversation to make sure
[21:14] that I don't forget the inspiration that
[21:16] I had. That will then immediately go to
[21:18] Google Docs. And then when I get home
[21:22] I've done that four times. I have four
[21:23] papers where I've started kind of the
[21:25] rough draft. Then when I get home, over
[21:27] the course of the next 3 hours, I'm
[21:30] spending time ping-ponging back and
[21:32] forth on the concept of what I came up
[21:34] with. I'm doing research on each of
[21:36] those. The reason I say that using AI
[21:38] gives me an edge right now
[21:41] every company that is part of my model,
[21:44] I spend time on it at some point. I know
[21:47] the names in there. I know the Japanese
[21:49] names. I know the Korean names. I know
[21:50] the European names.
[21:52] I'm having interactive dialogue.
[21:54] As a hedge fund manager and as someone
[21:56] who's a portfolio manager
[21:58] I've never been able to keep up-to-date
[22:00] on everything happening in the manner
[22:02] that's now and more importantly be able
[22:04] to brainstorm that be completely logged
[22:06] both in a context window, but also
[22:08] Google Docs. That's the reason why I
[22:10] don't forget anything.
[22:12] I used to have hundreds of ideas a day
[22:15] that would be lost forever, flushed. I
[22:17] couldn't remember everything. Now if
[22:19] there's something that's really special,
[22:21] it immediately goes to a list. That list
[22:23] goes to a Google Doc. That Google Doc
[22:24] gets updated back into the transcript.
[22:27] And then I go through and if it's
[22:28] something that I want to write a paper
[22:29] with today
[22:30] I spend the time brainstorming with AI
[22:32] over that. So it's the AI usage that's
[22:34] really the advantage.
[22:36] That is an incredible workflow. Um
[22:39] and I I'm using all the models every day
[22:41] as well. Um I can't say I utilize the
[22:44] the voice
[22:45] uh feature as much as you. Um
[22:49] Jordi, I I want to ask you about the the
[22:50] bottlenecks that you've mentioned in
[22:52] some of your papers um moving on from
[22:54] your workflows here.
[22:56] Uh
[22:56] the obvious bottlenecks people point to
[22:58] are power, cooling, optical, memory.
[23:01] Um
[23:02] is it possible that because everyone's
[23:03] talking about investing in these
[23:05] bottlenecks that they're becoming an
[23:06] obvious trade?
[23:08] There's no doubt they've become an
[23:10] obvious trade. Um, the scariest thing
[23:12] about the bottlenecks is this and and I
[23:14] learned uh
[23:16] I learned well, let's just go back to
[23:18] what I said about inference. So, I wrote
[23:20] a paper about inference in May of last
[23:21] year.
[23:23] It took 9 months before that became a
[23:25] major story. I wrote about software
[23:27] being dead.
[23:29] It took a while for that to become a
[23:30] major story. Um, the bottlenecks are a
[23:33] real story and I don't think people
[23:36] recognize that when you're looking at
[23:37] parabolic charts
[23:39] and everyone's like, this looks like a
[23:40] bubble.
[23:42] For something to look like a bubble,
[23:43] there's two things. Because remember,
[23:45] Amazon and the Mag 7
[23:48] if that would have happened in 2 years,
[23:49] it would have been a bubble. But
[23:50] instead, it took 12 years for them to go
[23:52] from a trillion and a half to 20
[23:55] trillion as a group. So, a 20 bagger on
[23:57] like that group of companies. But the
[23:59] fact that it took
[24:01] 12 years doesn't bother people as much.
[24:03] If it would have taken 1 year, then
[24:05] everyone would be freaking out. So, it's
[24:06] about not just the price movement to
[24:09] have a pole or a hockey stick that goes
[24:11] straight up. It has to be the time that
[24:13] happened. It has to be compressed. The
[24:15] problem with the bottlenecks that people
[24:16] are just not paying attention to
[24:19] it only became accepted this AI trade,
[24:22] honestly, in my opinion, in April.
[24:24] That's how scary this whole thing is
[24:26] that everyone now is a genius investing
[24:28] in Micron.
[24:29] When it didn't work for a long time.
[24:32] Like when we had that thing, I remember
[24:34] we had a couple conversations because
[24:37] you even asked me at some point, do you
[24:38] still like it?
[24:40] And I said, yes, I like it just as much.
[24:42] And I remember part of the argument was,
[24:44] this is a story that's going to last for
[24:45] 5 years minimum because we're eventually
[24:48] going to get to humanoids and people are
[24:49] going to realize that you need lots of
[24:51] memory for
[24:53] humans. Like humanoids. So, we have a
[24:56] long way to go. So, I I think as I like
[24:59] kind of think about where we are,
[25:01] I think people are missing the
[25:02] bottlenecks are going to have an impact
[25:04] on the memory names. They're going to
[25:05] have an impact on optical. They're going
[25:06] to have an impact on all the things that
[25:08] I'm positive on before the end of the
[25:10] year because we've only had one earnings
[25:12] quarter so far. That's it. Since it
[25:15] became accepted, it was based on one
[25:18] earning. So, we'll see what happens when
[25:19] July comes in and we're in the
[25:20] summertime,
[25:22] the bar is much higher now.
[25:24] So, the reason that I got out of my
[25:26] Micron is I could see it go to 1200, but
[25:28] then I could see them
[25:30] say that oh, this quarter is not as good
[25:32] because of the bottlenecks that
[25:34] happened. We weren't able to ship as
[25:35] much and it's going to be another three
[25:37] quarters before that happens.
[25:39] I don't know if the stock will be down
[25:40] 30% in a day, but I do know this, the
[25:43] risk of it being down 30% is very high
[25:45] at that point. So, this is not a
[25:47] conversation on whether these names will
[25:49] be good for the next 5 years.
[25:52] I do believe in in all cases this will
[25:55] be a bull market for those names, but I
[25:57] do think the risk is going higher
[25:59] because it's only been one quarter of
[26:01] earnings and now we've reset the bar so
[26:03] high that I'm just worried if there are
[26:05] any bottlenecks that people have
[26:06] underestimated the impact it'll have on
[26:08] the earnings reaction.
[26:10] Well, my constant dilemma in my head is
[26:13] that even if the bottlenecks are real,
[26:15] even if the demand is astronomical, we
[26:18] could still see, let's say quote a
[26:20] bubble in the asset price for a lot of
[26:22] these stocks.
[26:24] And I think Micron, I'm very bullish on
[26:26] Micron, but I think we're soon going to
[26:29] be getting to the point where the asset
[26:30] price of Micron doesn't
[26:32] uh
[26:33] look as good, let's say, or as
[26:35] compelling based on the risk reward of
[26:37] the actual bottleneck or the demand
[26:39] story. Uh do you see that same issue
[26:41] spreading to the rest of the market as
[26:43] far as the disconnect between the asset
[26:45] price and the actual fundamentals?
[26:48] Bottlenecks are a hardware story. So,
[26:51] what I've started to write more about is
[26:52] a rotation in the market that I think is
[26:54] going to happen
[26:56] for 3 to 6 months. Um more on the
[26:58] application side and the places where
[27:01] uh
[27:02] they're going to be used. Um
[27:04] I don't think there's, you know,
[27:06] bottlenecks for companies are different
[27:08] than like say the silver price.
[27:11] Um bottlenecks for silver are not bad.
[27:13] Bottlenecks for copper are not bad if
[27:15] you invest in the commodities because
[27:16] you want there to be bottlenecks cuz you
[27:18] want these to go higher. Um
[27:20] I I'm more interested at this point uh
[27:24] in the commodities that I like, the hard
[27:26] commodities, cuz they've corrected now
[27:28] for a period of 4 to 6 months.
[27:31] Uh for crypto because it's an
[27:32] application side where I think the
[27:34] financial guardrails would go. I don't
[27:35] expect crypto to go parabolic to start.
[27:38] But I do think the harder it is for
[27:40] people to invest in AI
[27:43] where it's a two-sided trade, and this
[27:44] is what I keep saying. What you
[27:45] described is
[27:49] I can make an argument
[27:51] where
[27:53] Micron could peak and be down for 2
[27:56] years. I can make an argument that way.
[27:59] Um people wouldn't like it, but I do
[28:02] have
[28:03] thoughts in my mind in the probability
[28:04] distribution of what people are going to
[28:06] overreact to.
[28:07] Bottlenecks are an issue. Not being able
[28:09] to meet production numbers is an issue.
[28:11] The token price is going down now and
[28:13] nobody seem to caring caring about it.
[28:16] Okay, all well and good. The
[28:17] government's getting more involved with
[28:19] the model companies. Not sure that's a
[28:21] big positive. The Chinese are bringing
[28:23] competition on in DRAM. But most
[28:26] importantly, I think there's an element
[28:28] here of people underestimating the power
[28:30] of the models. So, at some point
[28:33] the models get so powerful
[28:36] that recursive self-improvement comes in
[28:39] and they're able to figure out
[28:40] algorithmic solutions and efficiency
[28:43] solutions that people worried about with
[28:44] deep seek at the same time
[28:48] that they were not paying attention to
[28:53] Micron.
[28:54] So, at some point here, I do believe
[28:57] people will start to realize that if the
[28:58] models keep getting better and better
[28:59] and better without any data center build
[29:02] out. And that's the thing that really
[29:03] worries me the most is that these models
[29:05] have gotten so better that we got to the
[29:07] agentic world
[29:08] not because of
[29:10] Vera Rubin, not because of Blackwell. We
[29:12] got there because of
[29:14] all of the efficiency gains that came
[29:16] from reasoning, from test time compute,
[29:19] from reinforcement learning and
[29:20] reinforcement learning with human
[29:21] feedback. All of those things led to the
[29:23] models getting better on their own. And
[29:25] the Chinese don't have the best chips
[29:27] and yet they're still keeping up. So, I
[29:29] do think there's a chance we could look
[29:31] back and there'll be a model that gets
[29:32] really sick. We don't need as much
[29:34] memory for this model. And everyone's
[29:36] like, well, let's switch over there.
[29:38] Fusion was released by or put together
[29:40] by OpenRouter. That's got people a
[29:43] little freaked out in terms of what this
[29:44] means and the ability to use all these
[29:46] different models. So, I just think when
[29:48] people look back on this, there are
[29:49] risks that show up that I could see the
[29:51] narrative switching and people going,
[29:53] I'd rather be long the application side,
[29:56] which is benefiting from this superhuman
[29:58] intelligence, like drug discovery, like
[30:00] the crypto world, things like that
[30:02] as opposed to the physical hardware
[30:04] side.
[30:05] To me, this sounds like a similar
[30:08] lead up to the software sell-off. It was
[30:10] people were worried that AI was getting
[30:12] so good that a lot of the old winners
[30:15] became losers. So, it's like you had to
[30:17] be bullish on AI to suddenly turn
[30:19] bearish on a certain set of asset
[30:21] prices.
[30:22] Um is that am I thinking about that the
[30:23] right way?
[30:25] Markets rotate.
[30:27] I I Okay. Always. Um
[30:29] narrative switch.
[30:31] Narratives follow price. When price
[30:33] starts getting weaker,
[30:35] everyone goes, where can I make money?
[30:37] Um the one thing is everyone's trying to
[30:39] make money.
[30:40] And that's the part is that no one wants
[30:42] to sit in a trade that isn't making
[30:43] money. So, why aren't people in silver
[30:45] anymore? It's not working. Why aren't
[30:47] people in gold anymore? Everyone talking
[30:48] about gold all last year. Now, they
[30:50] don't want to be in gold. They were in
[30:51] Palantir all last year. Now, nobody
[30:52] wants to be in Palantir.
[30:54] These things have corrected.
[30:56] If you still believe in the story for
[30:58] long-term reasons,
[31:00] none of that's changed for any of that.
[31:01] In fact, I would argue that the silver
[31:03] story has actually gotten far better
[31:05] since silver peaked. And the reason is
[31:09] orbital data centers need more silver
[31:11] than terrestrial data centers.
[31:13] Solid-state batteries need significantly
[31:15] more silver than lithium batteries. And
[31:17] solid-state batteries are becoming a
[31:19] very, very important part of where the
[31:22] the solution could be for all of the
[31:25] power needs that we have. These are
[31:26] things that were not a major story in
[31:30] January. They are now stories. Orbital
[31:33] data centers were not a story. And they
[31:35] are now a story.
[31:37] You're the only guy I know that is
[31:39] talking about silver these days, which
[31:41] is probably a good sign. Um, Joey, let
[31:43] me ask you about Marvell. This is a
[31:45] stock that's had a great run this year.
[31:48] I know you still like it. Could you
[31:49] explain your thesis on that?
[31:51] Yep. Plain and simple, the optical side
[31:54] is in the very early stages. So, Marvell
[31:58] was hated by everyone in uh
[32:01] at the end of March. And I know that
[32:03] because I wrote a paper on March 30th
[32:05] about it.
[32:06] So, here we are. We're not even 3 months
[32:07] from that day. And yes, it has gone from
[32:10] about 90
[32:11] to 300. I had another good one on that.
[32:14] It's very Micron-esque now.
[32:17] I think it's going higher because the
[32:19] optical side and the major benefits from
[32:22] it
[32:23] are really going to come from Vera
[32:24] Rubin, which hasn't even started. So, I
[32:28] view kind of these movements as being
[32:30] two things.
[32:32] Is the stock expensive? It's a little
[32:34] bit expensive, but in the kind of all of
[32:38] the semiconductor
[32:39] and and interconnection areas of
[32:42] optical, it's not that expensive. But
[32:44] most importantly,
[32:46] there is the at some point the earnings
[32:48] just surprise. They haven't had that
[32:50] earning surprise yet. So, Micron has
[32:52] already had the earning surprise.
[32:54] The numbers blew away what people
[32:56] expected a year from now.
[32:58] I have learned with the AI
[33:00] infrastructure build-out that eventually
[33:02] the numbers go
[33:05] insane.
[33:07] Quant strategies go on insane numbers.
[33:10] Human beings go on insane numbers. We
[33:12] have not hit that
[33:14] insanity mode
[33:16] of like a Tesla, you know, supercar
[33:18] where you can start going as fast as you
[33:19] want. We haven't hit that. We hit it in
[33:21] Dell. We hit it in a lot of other names.
[33:23] So, in Marvell, I still see a lot of
[33:25] upside. And as I talked about with the
[33:27] Jensen Huang knowledge brain,
[33:30] Marvell was one of the ideas that came
[33:31] from the Jensen Huang knowledge brain.
[33:33] It was just asking, "Hey, he's talked
[33:35] about this so many times since beginning
[33:38] of the year at CES and then through the
[33:41] Morgan Stanley event, what names would
[33:43] fit what he's describing that Vera Rubin
[33:45] needs?" And it basically said, "Marvell
[33:48] fits it."
[33:49] The next day after I wrote the paper is
[33:52] when Nvidia announced a $2 billion
[33:54] investment into Marvell. And then at
[33:56] Computex 2 weeks ago, what did Jensen
[33:59] Huang say?
[34:00] On stage with I believe his name is Matt
[34:03] Murphy, the head of Marvell, said,
[34:05] "Marvell will be a trillion-dollar
[34:07] company." Well, it's a $275
[34:10] billion company today. So, I'll go with
[34:12] the with Uncle Jensen and say that we're
[34:15] going to continue to go higher.
[34:17] That's a pretty good bet and a good guy
[34:18] to bet on as well. Um,
[34:20] what about Eli Lilly? This is a unusual
[34:23] {quote} AI bet, I think, because it
[34:25] doesn't stand out as something that's a
[34:28] chip or anything related to that space.
[34:32] [snorts]
[34:33] Um
[34:34] this is an easy one for me. Um first of
[34:37] all, I love companies where their
[34:39] revenues are growing 50 plus percent,
[34:41] triple the next highest one within
[34:43] inside the pharmaceutical industry.
[34:46] Uh they have a dominant drug that has
[34:49] demand still to come in their GLP-1s.
[34:53] And they have used that cash to
[34:55] basically
[34:56] build out an AI infrastructure which is
[34:58] unparalleled within side the drug side.
[35:01] So that's Let's just leave Eli Lilly
[35:03] alone with that. Everyone can go read my
[35:05] paper. They can go hear about go look up
[35:08] Tune Lab. They can go look up Lilly Pod.
[35:10] They can go look up the Co-innovation
[35:12] Lab Lab. They're connected to Nvidia.
[35:14] They're connected to Insilico Medicine.
[35:16] They're connected to Isomorphic Labs
[35:18] which is DeepMind. This is a no-brainer
[35:21] in terms of this company betting all in
[35:22] on AI and having the cash to do this.
[35:25] And their number one competitor is a
[35:27] European company
[35:30] that's Danish that instead of using the
[35:32] money to go buy up other companies and
[35:34] to go do what Lilly did, they gave money
[35:36] back to the citizens. And I'm not saying
[35:38] that's the wrong decision to make, but
[35:39] I'm saying as an investor, we've learned
[35:41] in the US shareholder value creating
[35:43] shareholder value is what drives stock
[35:45] prices. So Eli Lilly on that front is
[35:47] there. But from the bigger macro
[35:49] picture, and that's what I am at the end
[35:50] of the day,
[35:51] the amount of money
[35:53] being spent
[35:55] in this country for health care
[35:59] is enormous and getting bitter. 20% of
[36:01] what people spend in this country goes
[36:04] to health care.
[36:05] When people How many people do we know
[36:07] that are
[36:08] billionaires from selling diet books or
[36:10] selling supplements? It's It's a world
[36:13] filled with amount of money that people
[36:15] will buy for anything.
[36:18] I know so many people that their lives
[36:20] have changed from GLP-1s.
[36:22] And more importantly, the data that's
[36:24] being collected from all the people
[36:25] taking them
[36:26] is the most valuable data which goes
[36:28] into AI or Eli Lilly's lab. Um if you go
[36:32] look up tune lab and realize that
[36:34] they're allowing other companies to put
[36:36] their data in there and a lot of those
[36:37] companies
[36:38] are looking are part of the benefits
[36:40] that have come in cancer, that have come
[36:42] from the benefits of kidney disease, of
[36:45] addiction. All of these things that are
[36:47] {quote} {unquote} side effects of the
[36:49] GOP ones that they don't they can't
[36:50] answer why. This is a very powerful
[36:52] story on both a macro basis on a micro
[36:55] basis and where the government is
[36:56] intervening in the
[37:00] the model side
[37:01] the fear has been that they would
[37:02] intervene on the drug side and make the
[37:04] pricing cheaper and cheaper. The one
[37:06] thing they don't want to do is slow down
[37:07] the ability for people to anti-age cuz
[37:10] it is a major need. Meaning we have a
[37:13] problem with Medicare and Medicaid. It
[37:15] is the dominant problem that's out there
[37:17] for entitlements. So the government on
[37:20] both sides needs to find a way to have
[37:21] these companies figure out how to make
[37:24] allow people to live longer. And so I
[37:26] think they're going to give the rope
[37:28] necessary to Eli Lilly, but it really
[37:30] has to do a lot with how much money
[37:32] they're making, how much more money it
[37:34] is than the other places and how they're
[37:35] buying up all of this IP in the biotech
[37:38] side.
[37:39] You know, when I first started hearing
[37:41] you talk about Eli Lilly and then I
[37:43] started doing a little my own homework
[37:45] on it
[37:47] it was like a light turning on. Like it
[37:49] it's it's so clear the way you lay out
[37:51] lay it out. I became a shareholder a few
[37:53] weeks ago because of you just like
[37:55] Micron. Um
[37:57] you mentioned something just now about
[37:59] the US government taking interest in
[38:02] taking stakes in these AI companies.
[38:05] Um
[38:07] is that the looming bear case you think
[38:10] for the entire AI ecosystem?
[38:15] Uh so there's no doubt that in the same
[38:17] way that you have the AI doomers and the
[38:19] AI bubble listas as I call them that I'm
[38:22] on the flip side that everyone calls me
[38:24] a
[38:25] perma bull on AI.
[38:27] And that's not the case.
[38:29] I have said publicly, which I still
[38:31] believe, that by the time we get to
[38:32] 2030, I do think AI will be
[38:35] likely to be disrupting all kinds of
[38:37] businesses and the stuff that we saw
[38:39] from software companies will spread to
[38:41] public companies. Now, I want to make
[38:42] sure people hear this. I'm a believer in
[38:44] tokenization. I'm a believer that small
[38:47] AI native companies have a huge
[38:48] advantage and I'd rather invest in a
[38:51] thousand small AI native companies than
[38:54] one mega super company within side the
[38:57] S&P 500. And I'll just give you an
[39:00] example. Give me a bunch of AI native
[39:02] companies competing with MasterCard and
[39:04] Visa. All day long. I do not believe in
[39:06] MasterCard and Visa being able to
[39:08] survive through the AI side. And I could
[39:09] be dead wrong,
[39:11] but I think that is a perfect example of
[39:13] the merging of AI and crypto, payments
[39:15] and all the middle men that have all but
[39:17] I think AI and crypto combined will
[39:19] destroy these things. So, I'm not a uber
[39:21] positive person when you go out, but I
[39:24] also don't make bets or investments
[39:26] based on 5 years from now. Where I think
[39:29] the near-term is very risky is on the
[39:30] bottlenecks, but it's also on what you
[39:32] just talked about. Um
[39:35] Leopold
[39:37] of the situational awareness fame,
[39:40] whose hedge fund gets all this, you
[39:42] know, you know, whatever his whatever he
[39:44] went from some small number to some
[39:46] multi-billions of dollars,
[39:49] whatever the truth is and how well he
[39:50] has done.
[39:53] He's been on the infrastructure side. If
[39:55] you go back to situational awareness,
[39:57] his forecasts on where we would be have
[39:59] been eerily good.
[40:02] What people should worry about is that
[40:04] he said we'd be reaching AGI around
[40:06] 2027. That was the whole point of this.
[40:08] 2027, early 2027 into 2028.
[40:13] He also mentioned that at that point the
[40:15] government would get more involved. Now,
[40:18] in the last month,
[40:20] the government has gotten significantly
[40:22] more involved. And if you take it back
[40:24] to the beginning of the war in Iran, you
[40:26] have to remember what happened with
[40:27] Anthropic the day before
[40:30] and the government um it's a very very
[40:33] dangerous thing to me for public
[40:35] companies to have the government
[40:37] involved.
[40:39] And even though if it's done in the way
[40:41] that
[40:43] a sovereign wealth fund would do where
[40:44] the money goes in and you hope that
[40:46] they're going to be passive investors.
[40:48] What he talked about, which I agree
[40:50] with, is these models get so strong and
[40:52] powerful that they're basically
[40:54] dangerous.
[40:56] And we have to shut off the rest of the
[40:57] world from having them. So, they become
[40:59] part
[41:01] software, part nuclear bomb, and they're
[41:03] that dangerous. And at that point, I
[41:05] don't know what happens. I don't know
[41:07] when you cross that line what happens.
[41:09] That has always been the fear is that if
[41:11] a technology or something gets so
[41:13] strong,
[41:14] the government starts to put regulations
[41:16] around it, starts to own it, it starts
[41:17] to take it. I think that's an
[41:19] uncertainty for the market that becomes
[41:22] very risky. I think for global
[41:24] investors, it becomes very dangerous.
[41:26] What has happened with Fable 5 is
[41:28] something everyone should really be
[41:29] paying attention to. I don't think we've
[41:30] seen the complete fallout yet. But
[41:33] sovereign AI, having your own artificial
[41:36] intelligence,
[41:38] what would you rather be focused on now?
[41:40] A model that if you built your entire
[41:43] system on a US model,
[41:45] but then the government says you can't
[41:46] use it. And a more powerful model's
[41:49] being built by the Chinese, which you
[41:50] can just download onto your computer and
[41:52] use it the way that you want.
[41:54] This is the competition from open
[41:55] source. So, I do view the government
[41:58] getting involved and all the things that
[42:00] I mentioned before about the negatives,
[42:02] I think this is in a point in time where
[42:04] people are not thinking clearly about
[42:06] this and they're just focusing on we
[42:07] just beat numbers and we're going to
[42:08] need memory for the rest of time. You
[42:10] can have multiple compress and have all
[42:13] this positive news happening.
[42:16] I'm going to I I'm willing to guess that
[42:20] things will look very different sometime
[42:22] over the next 3 to 6 months for the
[42:24] market and for all of these parabolic
[42:26] moves than it does today.
[42:28] I think there are so many
[42:30] social fabric issues that are going to
[42:33] come about with AI and part of this is
[42:35] the government involvement competing
[42:37] with other countries, banning other
[42:39] countries from certain models.
[42:41] And all these things, you know, it's so
[42:42] easy to focus on just the market
[42:44] reaction and the view as an investor,
[42:46] but there are so many other
[42:47] ramifications that
[42:50] you know, trying to keep up with
[42:52] Jordi, when you think about how you as
[42:55] an individual have changed as an
[42:57] investor, as a professional from when
[42:59] you left, let's say big firms on Wall
[43:01] Street being independent,
[43:04] what do you point to?
[43:06] Uh it's very simple for me. I I never
[43:08] had the So,
[43:11] I am like a retail person now. I'm home.
[43:14] Um
[43:17] I don't have a job,
[43:19] but I'm making
[43:21] money.
[43:22] And I'm making money by helping other
[43:24] people make money.
[43:26] Now, that's what all hedge fund people
[43:28] would say they do is, well, I take
[43:30] people's money and I turn it to more
[43:32] money, which
[43:34] I I I'll leave it alone and let the
[43:35] hedge fund people describe that to
[43:36] people. Um
[43:38] I'm not sure that's their ultimate goal
[43:40] at the end of the day and I don't think
[43:41] the distribution of wealth is there.
[43:43] In the case of for me, being home and
[43:46] being able to watch AI, use AI, think
[43:49] about AI, listen to the podcasts on AI,
[43:53] I don't think you can keep up with this
[43:55] stuff because of how fast it's moving if
[43:57] you don't have that time. So, I made a
[43:59] bet, which as of right now has proven to
[44:02] be a
[44:03] not only true, but I I recommend it,
[44:06] especially to people in college. And
[44:07] then my my son
[44:09] had just finished up his sophomore year.
[44:11] He's a very very
[44:13] um
[44:14] smart kid.
[44:17] And I got him
[44:20] to take a a Python boot camp
[44:23] when his senior in between his senior
[44:25] year and his his freshman year of
[44:26] college.
[44:27] And he uses the models all the time. I
[44:29] gave him the best model. I've given him
[44:31] things like the knowledge brain. I've
[44:33] given him so many tasks to do. I had him
[44:36] basically do an investment thesis for me
[44:38] on I can't remember which company it
[44:40] was. But using my approach, which again
[44:43] was on my subscriber paywall. And I
[44:44] wanted to make sure that a kid could do
[44:46] it. But more importantly, the
[44:48] empowerment that it gave him, the the
[44:50] internship that he's doing during the
[44:52] summer time where he chose not to do one
[44:54] in the financial industry, and he chose
[44:56] to do one outside of it. I I think those
[44:59] things are relate directly into my
[45:02] empowerment that I feel as an
[45:04] entrepreneur. My ability to pivot when I
[45:07] want to pivot.
[45:09] I didn't like working um for for large
[45:13] companies. I quit Morgan Stanley at the
[45:16] peak of my career
[45:18] because my job as a young person was
[45:21] already firing people as the major
[45:22] thing. It was it was dealing with
[45:25] management above me. It was dealing with
[45:26] people. I just didn't enjoy it.
[45:28] And at some point there's a balance
[45:30] between how much money you personally
[45:33] care about to live your life, what life
[45:35] you want to leave, and the garbage that
[45:37] comes associated with work. And so as as
[45:41] someone in AI, it's very easy to balance
[45:44] out what brings you joy
[45:47] and making money. For me, I love
[45:48] learning. I'm an insatiable learner. I'm
[45:50] sure people have gotten to know me
[45:52] because I've done enough podcasts now
[45:53] that
[45:55] my parents gave me a good brain. They
[45:57] gave me a brain that allows me to
[45:58] consume lots of information and kind of
[46:00] be able to speak in a way that it comes
[46:02] flying out. There's plenty of things I
[46:03] suck at. There's plenty of things that
[46:05] I'm that I wish I could be better at,
[46:07] but I'm using this skill to enjoy what
[46:10] comes with AI, which is an ability to do
[46:12] rabbit hole learning and to go learn
[46:14] topics on so many different things.
[46:17] I think that has been the edge that I've
[46:19] had is the time that I have, how I've
[46:21] used the AI to fill that time, and how
[46:23] there's never a time when I
[46:26] don't go through the Okay, what is the
[46:28] signal?
[46:29] What is the alpha? What agency? I highly
[46:32] recommend everybody kind of shape their
[46:34] lives with this rule of three. For me,
[46:36] to my subscribers, it's can I bring them
[46:38] signal, things that
[46:41] read through the noise. Just break
[46:43] through the noise, the whole signal or
[46:45] the noise thing with Nate Silver. The
[46:46] alpha. What are the names you should
[46:48] focus on as opposed to all of this
[46:50] stuff. For every Micron, there's a home
[46:52] builder that's unchanged. There's a Ford
[46:54] that's unchanged for 30 years. There's a
[46:56] lot of alpha that does nothing. So, in
[46:59] the benchmark arbitrage, hey guys, here
[47:01] are the 100 names. These are the ones
[47:02] I'd focus on. Use your trading
[47:04] techniques. The river is flowing with
[47:05] you. This is like jumping in a salmon
[47:08] stream. You're going to make money if
[47:09] you just focus on these 100 names. And
[47:11] then there's the agency side. If I teach
[47:13] you how to use it in a way, you will
[47:15] have the same thing that I have, which
[47:17] is the ability to build things and to go
[47:19] through things that you can build a
[47:21] business around. You can make your boss
[47:23] appreciate you. It'll get you more of a
[47:24] raise. If I've met plenty of people at
[47:27] these events where the people in their
[47:28] 30s are like, you know what the best
[47:29] thing is? I've automated a lot of my
[47:31] workflow, which means the days that I'm
[47:33] working at home, they don't even know
[47:35] what I'm doing, and I'm spending my time
[47:36] trading.
[47:38] Um
[47:39] if you had to give some advice to
[47:42] college students graduating into the
[47:44] current world, where would you start?
[47:47] Um number one, the doomers on the job
[47:50] side, it's never going to happen.
[47:52] Um
[47:53] so, as time has gone on, uh I've never
[47:56] been part of the Doom crowd on the job
[47:58] situation. But now I can say
[48:02] with certainty
[48:04] based on the data
[48:05] that everything's been blown out of
[48:07] proportion and that by the time we
[48:10] it won't happen
[48:12] like that.
[48:14] Yes, there's a depressing part of job
[48:17] where your the corporate ladder has been
[48:19] smooshed
[48:20] and every day there'll be more of
[48:21] replacements of jobs, but the thought of
[48:24] being out of work to go back to what my
[48:25] grandmother taught me about the Great
[48:26] Depression. Nobody's going to be out of
[48:28] work.
[48:29] Um which means best case scenario we're
[48:31] creating a little bit of job. So I'm not
[48:33] in any way that.
[48:35] For college kids, learn AI.
[48:38] Get the skill. It literally I I'm
[48:40] telling you think of it as Iron Man.
[48:43] You either have a superpower or you
[48:45] don't. And if you're going to be walking
[48:46] the earth without a superpower and
[48:49] you're going to listen to your
[48:50] professor, don't use AI. Your parent,
[48:52] don't use AI.
[48:55] Literally stick your middle finger up at
[48:56] all of them and just start using the
[48:58] tool. Start with the free version and
[49:01] then find a way to use this. And if you
[49:03] can, build something every single week.
[49:06] Start with that. And then once you build
[49:08] something every week, your career
[49:09] curiosity is unleashed and I think the
[49:12] strength as a person
[49:14] in fact, this is not a guess. As a kid
[49:16] who was very insecure
[49:18] and still I you never lose that
[49:20] insecurity. The insecurity that I had,
[49:23] the acne that I had, the stuff that just
[49:25] made me kind of this shy, skinny
[49:28] kid who really didn't know where things
[49:31] were going to go.
[49:33] As time went on, you start recognizing
[49:35] and looking back when you start
[49:37] something starts good happening, you
[49:39] doubt it. I didn't deserve that.
[49:41] But then as they start you're going
[49:43] through it, you're like, wait a second.
[49:45] The facts are the facts. Like I keep
[49:47] moving up the ladder no matter where I
[49:48] am.
[49:49] If you use AI
[49:51] every time that you go forward, it will
[49:54] get rid of that insecurity. You will
[49:56] start to feel like you went to the best
[49:57] college. I didn't go to the best
[49:59] college. I started at Morgan Stanley at
[50:01] the bottom. Eventually, I was managing a
[50:03] lot of people that went to the best
[50:04] colleges. Do you think they liked me? Of
[50:06] course not. And if I were them, I
[50:08] wouldn't either because they had put in
[50:09] all this time working super hard, and
[50:11] this guy barely paid attention in
[50:12] school, and now he's managing me. But
[50:14] the reality is at some point if you put
[50:16] the work in, as we just learned with the
[50:17] New York Knicks, you can beat the more
[50:19] talented people. You can actually get
[50:22] through this as a team, and I think AI
[50:24] is the best teammate you can have to get
[50:26] you through the hurdle, and I advise all
[50:28] college kids, all young people
[50:31] use AI every single day, and never never
[50:34] stop, and find out whatever you're
[50:35] passionate about, fantasy football,
[50:37] anything else, and start figuring out
[50:38] how you can incorporate it into what you
[50:39] enjoy.
[50:41] That is fantastic advice. Certainly
[50:42] advice that I follow every week. Um
[50:44] Jordi, where can people find your work
[50:46] online?
[50:47] [snorts]
[50:47] Uh they can go to ai.22vresearch.com.
[50:48] They can go watch my YouTube
[50:53] every Sunday. They can go to my
[50:56] substacks. Yes, folks, there are two
[50:58] substacks. Um and I will say that uh
[51:02] I'm working on something very um
[51:04] important to me since I do believe that
[51:06] crypto is going to go through
[51:09] uh
[51:10] a a new phase as the financial
[51:12] guardrails are necessary for the world
[51:14] to become more agentic. Uh I am rolling
[51:16] out another YouTube that'll probably
[51:18] start just after the summertime. I'm
[51:21] spending my summer in Maine figuring out
[51:23] how to be able to launch this. This will
[51:26] be to translate in the same way I do
[51:28] with AI, with Signal Alpha and Agency
[51:31] back into crypto. So, I know there's a
[51:33] lot of people that have never been
[51:35] involved with crypto in the same way
[51:36] they're not involved with AI. This is
[51:38] another financial empowerment side. I do
[51:40] think crypto combined with AI, if you
[51:42] know those two things, and that was my
[51:43] goal when I decided not to stay in the
[51:45] work world working for someone. No
[51:48] companies that I met, and I don't care
[51:49] if it's BlackRock or any of them,
[51:52] had a plan for how to deal with the the
[51:54] intersection of AI and crypto, and
[51:56] that's why my stuff is AI macro nexus
[51:58] when I write. It's all based on the fact
[52:01] that the world is changing rapidly, and
[52:02] if you don't understand these two
[52:04] things, you're not going to financially
[52:05] be able to deal with what's happening.
[52:07] Everyone should definitely go subscribe
[52:09] to Jordy everywhere. Uh thank you so
[52:12] much for your time, Jordy. We'll do it
[52:13] again.
[52:13] Thanks, Phil.