Jordi Visser: The AI stocks that win the next decade

Summary

---

Full Signal with Phil Rosen

Episode Summary: "Jordi Visser: The AI stocks that win the next decade"

Release Date: June 17, 2026 | Guest: Jordi Visser

---

Episode Overview

In this episode, Phil Rosen sits down with veteran investor Jordi Visser to unpack the next era of AI investing. Visser shares his frameworks for understanding the coming “physical” AI buildout, why the next decade’s market winners will be fundamentally different from the last, and how investors can position themselves for massive structural shifts. The conversation dives deep into the transition from software to hardware, portfolio construction in an exponential environment, bottlenecks, government risk, and personal philosophies on keeping up in a fast-changing world.

---

Key Discussion Points & Insights

1. From Software Led Winners to Physical AI — The CapEx Opportunity

• Jordi’s thesis: The previous post-GFC decade’s alpha was concentrated in “MAG7” software firms (Microsoft, Apple, Google, etc.), but the next era is shaped by AI “factories” and physical compute infrastructure.
• Jordi:

  “Jeff Bezos famously said, ‘your margin is my opportunity.’ And this was really about the concentration of power being isolated into software companies. ...Now, Jensen Huang comes out this year and basically tells everyone, okay, we're at the stage now where we're going to build AI factories ...it's going to take $90 trillion to reindustrialize through intelligence.” [01:08]

• The focus is shifting from software margins to who can capitalize on the capex needed for AI’s physical buildout—hardware, chips, energy, and commodities.

2. The Agentic AI Shift and Opus 4.5 as the Gateway

• The rapid advancement to the “agentic” stage of AI, marked by models with higher IQ and self-improvement, was signaled by the release of Opus 4.5.
• Jordi:

  “Think of Opus 4.5 as being the gateway to the agentic world. And then Jensen Huang says, we’re here, it’s going to be $90 trillion to basically manufacture tokens to feed the AI agents.” [02:41]

• This stage spurred a rotation in equity markets, notably from software to hardware names.

3. The Five-Layer Cake Framework for the AI Economy

(Framework attributed to Jensen Huang)

1. Applications/Workflows
2. AI Models/Platforms
3. Data Infrastructure
4. Chips & Compute
5. Energy, Hardware, Commodities

• Investment focus: The opportunity is greatest at the bottom three layers, particularly energy, chips, and infrastructure.
• Jordi:

  “You want to be on the bottom three, which is the energy chips and the infrastructure... For the chips... it's either the semiconductor chips like the Nvidias, the Intels... For the infrastructure, think Dell, HPE, Pure Storage...” [06:01]

4. What Happens to the MAG7?

• With the exception of Nvidia, most mega techs become the big spenders (not winners), facing “multiple compression” as margins and growth slow.
• Jordi:

  "These companies are going to go through multiple compression... They've become hardware companies, they're spending lots of money, they're issuing stock and taking on debt... it's a multiple compression story. We've definitely seen that with Nvidia already.” [07:31]

• Comparison: The upcoming AI buildout is likened to the fracking boom—big capex bets, some pay off, but many do not.

5. Benchmark Arbitrage — Positioning for the Next Decade

• Explains why traditional indices are backward-looking (weighted toward yesterday’s superstars) and why overweighting future AI infrastructure names is key to outperformance.
• Jordi:

  “The whole game of outperforming benchmarks is to figure out what the weightings of the benchmarks are going to look like in the future.” [09:20]

• He’s constructed a “basket” of 100 global stocks across energy, chemicals, hardware, and infrastructure—not just US-listed giants.

6. The Power of "Using AI" as an Investor Edge

• Process: Jordi describes his own workflow—using AI models like ChatGPT and Google Docs to capture, develop, and act on fast-moving ideas.
• Jordi:

  “As a hedge fund manager... I have never been able to keep up to date on everything happening in the manner that’s now... Now if there's something that's really special, it immediately goes to a list, that list goes to a Google Doc... That's the reason why I don't forget anything.” [21:16]

---

Notable Quotes & Memorable Moments

• On Capex as Opportunity:

  “Your capex is my opportunity.” [00:00]
  (Jordi paraphrasing Bezos, tailoring it to the AI buildout)

• On Market Rotation:

  “Markets rotate always. Narratives switch, narratives follow price. When price starts getting weaker, everyone goes, ‘where can I make money?’” [30:25]

• On Benchmark Arbitrage:

  “If you invested in these 100 companies...they might be up 50% a year. My basket has outperformed the S&P significantly this year.” [11:09]

• On the Value of Using AI:

  "...if you're not informed and you're not using it, I don't think you have a right... I don't think you can possibly have a high probability of being correct if you don't know what it is." [14:30]

• Career & Life Advice:

  “I advise all college kids, all young people, use AI every single day and never, never stop... Find out whatever you're passionate about, fantasy football, anything else, and start figuring out how you can incorporate it into what you enjoy.” [50:30]

---

Deep Dives & Timestamps

[05:10] – The Five-Layer Cake in Detail

• Phil introduces the structured AI economy model
• Jordi emphasizes investing in the "bottom three" (energy, chips, infrastructure)

[07:17] – Fate of the MAG7

• Jordi warns on multiple compression and a shift away from growth-centric valuations

[09:20] – Benchmark Arbitrage in Practice

• Outperformance means identifying tomorrow’s leadership, not yesterday’s

[12:29] – Generational Divide & AI Literacy

• On the necessity for seasoned managers to actually use AI to make informed investment decisions

[19:22] – Jordi’s Workflow: How AI Transforms His Investing

• Describes leveraging daily walks, podcasts, and direct engagement with AI tools to spark and capture investable ideas

[23:08] – Bottlenecks: Obvious Yet Unavoidable?

• Discusses whether everyone crowding into bottleneck trades makes them overvalued, and the real risks of execution misses

[26:49] – Fundamentals vs Asset Prices

• Ongoing tension as hardware bottleneck stories can still disconnect from fundamentals and become bubbles

[31:51] – Marvell (MRVL) Thesis

• Optical’s infancy, huge upside ahead. Recent $2B Nvidia investment and CEO commentary reinforce the bull case.
• Jordi:

  “Marvell will be a trillion dollar company. Well, it’s a $275 billion company today. So I’ll go with Uncle Jensen and say we’re going to continue to go higher.” [33:36]

[34:34] – Eli Lilly as an "AI Bet"

• Drug company is all-in on AI for drug discovery, data aggregation (GLP1s), and owns critical infrastructure.
• Jordi:

  “This is a no brainer in terms of this company betting all in on AI and having the cash to do this.” [35:05]

[38:15] – Government Involvement and AI “Sovereignty”

• Risks to private valuations as sovereigns get involved
• Worries about models becoming too powerful, leading to regulations or outright nationalization

[47:39] – Practical Advice for Young Graduates

• Learn AI—ignore naysayers; become a “superpowered” operator in any field
• Use AI for productivity, confidence, and career acceleration

---

Broader Themes & Takeaways

• The AI buildout is physical: Real gains will come from capex-intensive sectors—semis, power, infrastructure, and enabling commodities.
• Indicies lag paradigm shifts: Winning stocks of the past aren’t the best guide—be proactive about future weighting.
• AI literacy = table stakes: Investors of any age/experience should use AI, not just read headlines, to keep an edge.
• Expect volatility: Even where the long-term story is positive, bottlenecks, government, and model shifts mean stocks will be volatile—risk-reward can flip quickly.
• New market leaders: Companies capitalizing on data, hardware, and energy needs will shape the next decade, not just “big tech.”
• Don’t bet against human adaptability: The doom around jobs and AI is overstated—curiosity and learning will drive new opportunities.

---

Where to Find More from Jordi Visser

• AI.22vresearch.com
• Weekly YouTube show (Sundays)
• Substacks (Signal Alpha, Agency, and upcoming AI-Macro-Nexus/crypto edition)

---

Final Words

Visser’s advice is practical and urgent: Engage directly with AI, look beyond yesterday’s winners, follow exponential rather than linear change, and seek empowerment through curiosity and adaptability. His optimism is anchored by a clear-eyed view of risk, especially as government and geopolitical tensions mount.

Phil:

“That is fantastic advice, certainly advice that I follow every week.” [50:41]

---

Transcript

[Phil]

Jordi, I want to take us to your paper. You wrote recently about how the new age of investing is about your capex is my opportunity. That's a play on Jeff Bezos's your margin is my opportunity. What's going on here? All right, so

[Jordi]

there's two themes that come out of this. Number one, really post the great financial crisis, the Alpha that was generated by all investors really from 2009 on, was isolated into effectively the MAG7. Those are the winners. And Jeff Bezos famously said, your margin is my opportunity. And this was really about the concentration of power being isolated into software companies. Commodities didn't work during the 2010-2020 decade. Bonds didn't work. You ended up having negative real yields or negative yields over. And then we ended up getting Covid. We got inflation. So bonds basically have not performed. You had a little bit of performance, obviously from the early stages of vc, but then that's run into trouble and then private equities run into trouble. The whole point of the original part was just that we were in a decade where I don't think people realized how concentrated the Alpha was. So now you fast forward and Jensen Yuan comes out earlier this year and basically tells everyone, okay, we're at the stage now where we're going to build AI factories in those AR factories. It's going to take $90 trillion to re industrialize through intelligence. So when Jensen Yuan, who arguably, not arguably is at the. He's the centerpiece that sees everything. Whenever you see any of those circular flows, Jensen Yuan is involved no matter what, in terms of what he's got, because he had GPUs. And they still today are the most necessary part of the AI buildout. So when he says it's going to be 90 trillion over the next 10 to 15 years, and everyone is focused on the data center side, the winners of this particular 10 to 15 year period are going to be hardware companies. And so the whole point was at the beginning of the year, the models had reached a point where the agentic world was here. Jensen yuang saying, 90 trillion. You need to have the agentic world here. The agentic reaching the agentic stage, reaching IQ to a level of say, 130, where all of a sudden the coding side that Opus 4.5 brought us into allows us to expand artificial intelligence to a very high level. Because of the agentic side, we hit recursive self improvement, then we hit AGI, then eventually we hit super intelligence. All of those get us to humanoids and everything else. So think of Opus 4.5 as being the gateway to the gentic world. And then Jensen Yuan says, we're here, it's going to be $90 trillion to basically manufacture tokens to feed the AI agents.

[Phil]

So a lot of this thesis from my understanding is we are moving into the physical AI stage of things. Was there something specific you can point to in the last few months that convinced you that we're moving into that stage right now?

[Jordi]

Honestly, I started writing about inference in May of last year. Inference is really the beginning of the agentic AI and the need for all the hardware. And I say May of last year because that's when the companies, particularly the most famous one was when Google openly started talking about what the token usage was. When the token usage started to get higher, it means more and more people were using it. It was getting more powerful. Now when DRAM started, started to go up really beginning in September, that's when we started to see the memory stocks truly take off and start to go on their cliff run. And I think at that point Silicon Valley and the people around the world realized we were there. The one thing people have to realize, these models that get released, they've had them for a long time. I mean the speculation is that they're always six months ahead the frontier model companies before they release them. And it's probably going to be even a little bit longer now based on what happened with Mythos. Meaning once you get to the point that these models are so powerful, they're probably going to be delayed more or the government will get involved, which means they're going to be delayed even more. Maybe they never get released because they don't want competitors to see what they are. I think that period last year triggered the beginning. But it was Opus 4.5 that I've written about papers. That's when we saw the software stocks really start to get killed. That was the point that I think everyone who is using those models and famously Andrej Karpathy turned from we won't be dealing with AI agents for a decade. He said it was going to be a gradual thing. This went from gradual to just parabolic. And now Andrej Karpathy is actually working for Anthropic. So you can kind of pick the October to November period when 4/4.4.5 got released, that that was the beginning of AI agents.

[Phil]

I think that makes sense. I want to ask you about this framework that you laid out about the AI economy as a five layer cake. And I have it written here. The applications and workflows are at the top followed by the AI models and platforms. Then you have data infrastructure, chips and compute and then at the bottom you have energy, hardware and commodities. My understanding is that you want to be investing at the bottom of that cake. Is that the right way to think about it?

[Jordi]

I'd love to take credit for creating the five layer cake, but that's Jensen Yuang, so that's just me writing about what he has said. Specifically. Yes, you've over the course of the last year, and I'll say year just because it's a year ago that Micron really started to breakthrough levels and we started to see the Western digitals, the sandisk and all this stuff just start to go up. And I think that was the recognition that we had reached a point that the five layer cake, which is really going to be the part that all drives the IQ side. You don't get into the application side until the IQ gets high enough. You don't get into the AI agent side until the IQ gets high enough. So beginning a year ago we were the only pieces that were worth focusing on. And even up till now in June of this year, you want to be on the bottom three, which is the energy chips and the infrastructure. For the energy side you can pick anything you want out of it. You can pick Bloom Energy, you can pick some of the battery companies, you can pick some of the natural gas transformer companies, anything that goes into the power. For the chips it's self explanatory, it's either the semiconductor chips like the Nvidias, the Intels, the Texas Instruments or you've the memory side and then for the infrastructure, think Dell, think hpe, think Pure Storage, think places that are basically going to be part of the AI factories.

[Phil]

In all of that framework the Mag 7 are not really included except Nvidia it sounds like. Am I wrong to say that, yeah,

[Jordi]

Most of the Mag 7, let's leave Tesla out of it because Tesla will benefit when we get into the embodied side which would be the applications, but let's just say the model side. The problem is these are the spenders. So I've had a. Okay, I don't want to say negative meaning they're going to go down, but these companies are going to go through multiple compression. In my opinion. They've become hardware companies, they're spending lots of money, they're, they're issuing stock and they're taking out debt. They have to make the revenues in the future. The biggest comparison I can get, not that it's going to end this way is the fracking boom. I've talked about this before where these companies are making a huge bet that they're going to benefit from what is going to be a commoditization of intelligence. And maybe they'll get the money in, maybe they won't, maybe they'll get a lot of it in, but not enough. That uncertainty and all the issuance and all the things, the bottlenecks that are coming, I just think they're going to see multiple compression. I'm not sure companies of that size should have a 30pe anyway. Microsoft was a single digit PE in the early 2010s. So my guess is that this is going to be a multiple compression story. We've definitely seen that with Nvidia already where they're PE has declined from the 40s down into the low 20s. And if you look at a year further, it's in the teens. So I think all of the Hyperscalers and the Mag 7s are going to run through the same thing.

[Podcast Sponsor/Host]

And a lot of this is also

[Phil]

what you refer to as benchmark arbitrage. Right? You have these indexes like the S&P 500 or MSCI world that are essentially weighted to the last decade of outperformers like the Mag 7. But the way that my guess is that you're thinking about it as an investor is finding the weightings for the next decade and then trying to build that into your framework.

[Jordi]

The whole game of outperforming benchmarks is to figure out what the weightings of the benchmarks are going to look like in the future. So whether you're a mutual fund, whether you're a pension fund who's benchmarked to msci, a mutual fund who's benchmarked to maybe the S and P, maybe a growth index, whatever, but then also RIAs and wealth managers that are managing for people. The whole benchmark arbitrage was to get people to recognize that there's two ways that you're going to end up with the proper weightings. One is by just being long these passive indices and waiting for the weightings to catch up, which in my opinion will probably be a disappointing performance relative to what you're watching on the sidelines. As Micron goes up 10 times, as Sandit, all of these stocks have gone up three to 10 times. So my argument has been for people that are benchmarked to these passive indices, that they should be at least overweighting the names. And I created 100 names. And those hundred names are not all in the S&P 500, they're not all American companies, but they're all driven by the exact same thing, which is the build out in AI. They're in the power side, they're in the chemical side, they're in the semiconductor, in the packaging side, they're in the infrastructure side. They're all different types of ones that fit in. And people are not going to think of chemical companies as being important. But to me, chemical companies are like the perfect example of chemicals are going to be more GDP intensive than oil's going to be than oil was for the last 200 years. Oil was a transport thing. It was about globalization. Chemicals is about the buildout and the bonding and the necessity for all of this electricity. It's a different world. And I think people need to realize that when changes happen in a linear fashion, you don't think about benchmark arbitrage, but when it happens in an exponential fashion, you can end up with AN S&P 500 that's unchanged. But if you had invested in these a hundred companies, they might be up 50% a year. My basket has outperformed the S and P significantly this year. And this is the reason why I've taken this approach with kind of deciding not to work for a company and be an AI power user and then to spread the word to people out there. It's really been broken down where it's Cigna, Alpha and agency, the signal is about making sure they're up to date on everything happening. With AI, it is a X is filled with doomers. You've got the alpha side, which names, if I believe in AI, what names will make that benchmark arbitrage. And then the agency side is making sure that I'm helping people. People use the tools because if they use the tools, they're more likely to believe in the stuff that is the signal. And then they're more likely to be looking for financial empowerment by investing in the alpha.

[Phil]

What do you make of investors and let's say veteran fund managers like yourself who have the same years of experience as you but have fallen on the totally opposite side. As far as a worldview on AI right now,

[Jordi]

I think to each his own on the way that they want to go through this. I believe the world and predicting the future is a world of distributions and the probability of anyone being right to the end is zero. I thought Bitcoin would be five or six times higher than it is today. So I got that one wrong. I got Micron right. There's plenty of things that we all get wrong, we all get right. So I leave people to their AI views there. I'm not a preacher. I'm not telling people what they should do. But I do believe the evidence is you can't make a informed decision if you're not using AI. And so my message to people that are around my age, you know, nearing 60, if you believe that, you know whether AI is going to make it. And that view is coming solely from your experience of the past of I've seen technology, I saw the dot com bubble. I think you're making a huge mistake if you use it and you're not amazed by it. And when I say use it. I released something on my paywall this week which was showing people how to make a knowledge brain. And the response I got from it was overwhelming. I'm actually pleasantly surprised that I finally put something up. I've put up prompts, I've given people videos, but this was the first time that I gave them a beginning to end. Go create a knowledge brain. And for the people that did it, and more importantly that then had their kids do it, they basically said they immediately understood the difference. Now I suggested they go, we're getting in the fantasy football season. Like to just verbally say this for your show. For anyone who does fantasy football is going to be in drafts. You're all listening to some podcast or going on some website. Pick your five favorite analysts and then go to the paywall. Get my knowledge brain stuff and then go connect to all of the times that they give interviews. If they do a weekly podcast, just take the transcript from YouTube and upload it. All of this stuff happens in a matter of 30 minutes. And then you can ask questions about each of the players that they've talked about over each of the interviews from the beginning of the year. They're talking about the drafts, they're talking about the players, where everything stands. It's really hard to keep up on fantasy. I remember when I was doing it actively, I can't do it anymore because I don't have the time. But these are the ways that I think for people who are. I don't even want to say bubble people, doomers, whatever you want to call it, if you're not informed and you're not using it, I don't think you have a right. I don't. I don't think you can possibly have a high probability of being correct if you don't know what it is.

[Podcast Sponsor/Host]

Today's episode is sponsored by public and they just launched a new product called Agents. And it's honestly one of the most compelling uses of AI in investing I've ever seen. The idea is simple. A lot of investing is repetitive. Monitoring markets, moving cash, managing risk, rebalancing positions. It's the kind of stuff investors constantly have to stay on top of. Public's new agents product lets you automate those workflows using plain English. So instead of manually watching charts all day, you can create an agent by typing something like, if the Vix hits 25, buy a weekly put option on spy. Or if my cash balance goes above 20,000, move the excess into my direct index. Public's AI then turns that into an automated workflow that you can customize, review and activate directly in your portfolio. Public is essentially building what they call an agentic brokerage, a platform where AI can help automate the operational side of investing. Based on rules and strategies you define yourself. If you want to start automating ideas inside your portfolio, visit public.com openingbell that's public.com openingbell now let's get back to the show.

[Phil]

You and I are both in the not a bubble camp, so I to start with that. But if you sort of canvas the views of the bears, is there any bubble arguments that stand out to you that you say, wow, that actually makes a lot of sense, or that could be pretty credible.

[Jordi]

And again, I have to fight my own. So we all have bubble ptsd. I've seen plenty of times in my career having been involved in the markets and remembering 1987 when I wasn't in the markets, but I remember what it did to my town. I remember the people who worked in the industry and what they went through and the people that lost their jobs. I've read historically all the books that everyone else has read. It's really hard to consume all of that knowledge to listen to your grandmother talk about not having food during the Great Depression and needing to leave the house when she was 16 because they didn't have enough money and not have it impact the way that you view things. Like everyone in Korea owns the two stocks over there or two stocks are driving. Like the part about the bubble that I think is real is that there are no free lunches in investing. You don't just get to show up and buy call options and make money all the time. I don't think that's what retail traders do. But the reason that I publicly said that I got rid of Micron and trust me, it got up to 1100 yesterday, I got out in the 600s. So that means it's almost doubled since I got out, which doesn't feel good. But I knew that that was the most likely scenario to happen because I'm not picking the top of a bubble the same way that I'm not going through it. I don't view Micron as a bubble. I view it as I had better places to put money that over the course of the next two years, I think I'm going to make more money being invested in these other name. But part of the reason was because what I was reading from Korea, what I was hearing from investors, the pushback that I was getting a year ago and now, how all hedge funds were embracing it, it just felt to me like too many people were involved and it was no longer the same risk reward and it was still being bet as the favorite. So I think if there's one part that I would agree with the people on the bubble side is that recently it's gotten a little too easy and there are no free lunches in investing.

[Phil]

First of all, I still own Micron. And the reason I got into Micron, of course, was because you and I had a conversation in January of last year when Micron was probably 90 bucks and you said that was your favorite stock at the time. And I published that to my subscribers. A lot of people have been in Micron that follow me for a long time. So thank you for that. First of all, I guess when you think about your own experience and your years in the market, is there anything you can point to specifically that you think gives you an edge in how you're looking at the AI economy and the AI trade that you think is differentiated?

[Jordi]

Honestly, it's using AI. Today is a Tuesday. So to take you through what I did from 6 o' clock until 11 this morning, I start with a walk through Brooklyn. And this is almost. It depends on whether I'm doing workouts in the very early morning or later on. But on most Tuesdays, I go for a walk and I'll do usually about a mile and a half, sorry, three to four mile walk. And during that walk I'll listen to podcasts. So at 1.75 speed and I'll start consuming information. Now, during the walk, something will hit my brain that I'm listening to. Now, most of these podcasts are. They're ones that come out two or three times a week or at least weekly. And on this particular day as I was walking, I was able to come up with the inspiration for four papers.

[Podcast Sponsor/Host]

Four.

[Jordi]

One of them involving health, three of them involving AI. So what had transpired in the time from that? I will pause the podcast I'm listening to. I will switch over to one of the models. It's usually chatgpt. And part of the reason it's usually chatgpt is because the voice thing is just so much easier for me on ChatGPT. I'm not really sure why Anthropic doesn't have the amount of compute. So unless my service is picture perfect, Anthropic is really slow and the voice thing is not as good for me as either Gemini or ChatGPT. ChatGPT is the best. During that period I will have a conversation to make sure that I don't forget the inspiration that I had that will then immediately go to Google Docs. And then when I get home, I've done that four times. I have four papers where I've started kind of the rough draft. Then when I get home I Over the course of the next three hours I'm spending time ping ponging back and forth on the concept of what I came up with. I'm doing research on each of those. The reason I say that using AI gives me an edge. Right now, every company that is part of my model I spend time on at some point. I know the names in there. I know the Japanese names, I know the Korean names. I know the European names. I'm having interactive dialogue. As a hedge fund manager and as someone who's a portfolio manager, I have never been able to keep up to date on everything happening in the manner that's now and more importantly be able to brainstorm that be completely logged both in a context window but also Google Docs. That's the reason why I don't forget anything. I used to have hundreds of ideas a day that would be lost forever, flushed. I couldn't remember everything. Now if there's something that's really special, it immediately goes to a list, that list goes to a Google Doc, that Google Doc gets updated back into the transcript and then I go through and if it's something that I want to write a page paper with today, I spend the time brainstorming with AI over that. So it's the AI usage that's really the advantage.

[Phil]

That is an incredible workflow and I'm using all the models every day as well. I can't say I utilize the voice feature as much as you. Jordi, I want to ask you about the bottlenecks that you've mentioned in some of your papers. Moving on from your workflows here, the obvious bottlenecks people point to are power cooling, optical memory. Is it possible that because everyone's talking about investing in these bottlenecks that they're becoming an obvious trade?

[Jordi]

There's no doubt they've become an obvious trade. The scariest thing about the bottlenecks is this. And I learn. Well, let's just go back to what I said about inference. So I wrote a paper about inference in May of last year. It took nine months before that became a major story. I wrote about software being dead. It took a while for that to become a major story. The bottlenecks are a real story. And I don't think people recognize that when you're looking at parabolic charts and everyone's like, this looks like a bubble. For something to look like a bubble, there's two things, because remember Amazon and the Mag 7. If that would have happened in two years, it would have been a bubble. But Instead it took 12 years for them to go from a trillion and a half to 20 trillion as a group. So a 20 bagger on that group of companies. But the fact that it took 12 years doesn't bother people as much. If it would have taken one year, then everyone would be freaking out. So it's about not just the price movement to have a pole or a hockey stick that goes straight up, it has to be the time that happened. It has to be compressed. The problem with the bottlenecks that people are just not paying attention to, it only became accepted to say I trade. Honestly, in my opinion, in April, that's how scary this whole thing is, that everyone now is a genius. Investing in Micron when it had worked for a long time, when we had that thing, I remember we had a couple conversations because you even asked me at some point, do you still like it? And I said, yes, I like it just much. And I remember part of the argument was, this is a story that's going to last for five years minimum, because we're eventually going to get to Humanoids and people are going to realize that you need lots of memory for humans like Humanoids. So we have a long way to go. I think as I think about where we are, I think people are missing the bottlenecks are going to have an impact on the memory names, they're going to have an impact on optical, they're going to have an impact on all the things that I'm positive on before the end of the year because we've only had one earnings quarter so far. That's it. Since it became accepted, it was based on one earnings so we'll see what happens when July comes in and we're in the summertime. The bar is much higher now. So the reason that I got out of my Micron is I could see it go to 1200, but then I could see them say that, oh, this quarter is not as good because of the bottlenecks that happened. We weren't able to ship as much, and it's going to be another 3/4 before that happens. I don't know if the stock will be down 30% in a day, but I do know this. The risk of it being down 30% is very high at that point. So this is not a conversation on whether these names will be good for the next five years. I do believe in all cases this will be a bull market for those names. But I do think the risk is going higher because it's only been 1/4 of earnings and now we've reset the bar so high that I'm just worried if there are any bottlenecks that people have underestimated the impact it'll have on the earnings reaction.

[Phil]

Well, my constant dilemma in my head is that even if the bottlenecks are real, even if the demand is astronomical, we could still see, let's say, quote, a bubble in the asset price for a lot of these stocks. And I think Micron. I'm very bullish on Micron, but I think we're soon going to be getting to the point where the asset price of Micron doesn't look as good, let's say, or as compelling based on the risk reward of the actual bottleneck or the demand story. Do you see that same issue spreading to the rest of the market? As far as the disconnect between the asset price and the actual fundamentals, bottlenecks

[Jordi]

are a hardware story. So what I've started to write more about is a rotation in the market that I think is going to happen for three to six months more on the application side in the places where they're going to be used. I don't think there's Bottlenecks for companies are different than, say, the silver price. Bottlenecks for silver are not bad. Bottlenecks for copper are not bad. If you invest in the commodities because you want there to be bottlenecks because you want these to go higher. I'm more interested at this point in the commodities that I like, the hard commodities because they've corrected now for a period of four to six months. For crypto, because it's an application side where I think the financial guardrails will Go. I don't expect crypto to go parabolic to start, but I do think the harder it is for people to invest in AI where it's a two sided trade. And this is what I keep saying, what you described is I can make an argument where Micron could peak and be down for two years. I can make an argument that way. People wouldn't like it. But I do have thoughts in my mind in the probability distribution of what people are going to overreact to. Bottlenecks are an issue, not being able to meet production numbers is an issue. The token price is going down now and nobody's seemingly caring about it. Okay, all well and good. The government's getting more involved with the model companies. Not sure that's a big positive. The Chinese are bringing competition on in dram. But most importantly, I think there's an element here of people underestimating the power of the models. So at some point the models get so powerful that recursive self improvement comes in and they're able to figure out algorithmic solutions and efficiency solutions that people worried about with deep seek at the same time that they were not paying attention to Micron. So at some point here I do believe people will start to realize that if the models keep getting better and better and better without any data center build out. And that's the thing that really worries me the most is that these models have gotten so better that we got to the agentic world not because of Vera Rubin, not because of Blackwell. We got there because of all of the efficiency gains that came from reasoning from test time, compute from reinforcement learning and reinforcement learning with human feedback. All of those things led to the models getting better on their own. And the Chinese don't have the best chips and yet they're still keeping up. So I do think there's a chance we could look back and there'll be a model that gets released. Like we don't need as much memory for this model and everyone's like, well let's switch over there. Fusion was released by or put together by OpenRouter. That's got people a little freaked out in terms of what this means and the ability to use all these different models. So I just think when people look back on the this there are risks that show up that I could see the narrative switching and people going, I'd rather be long the application side which is benefiting from this superhuman intelligence like drug discovery, like the crypto world, things like that, as opposed to the physical hardware side.

[Phil]

To me this sounds like a similar Lead up to the software sell off, it was people were worried that AI was getting so good that a lot of the old winners became losers. So it's like you had to be bullish on AI to suddenly turn bearish on a certain set of asset prices. Am I thinking about that the right way?

[Jordi]

Markets rotate always. Narratives switch, narratives follow price. When price starts getting weaker, everyone goes, where can I make money? The one thing is everyone's trying to make money. And that's the part is that no one wants to sit in a trade that isn't making money. So why aren't people in silver anymore? It's not working. Why aren't people in gold anymore? Everyone talking about gold all last year, now they don't want to be in gold. They were in Palantir all last year, now nobody wants to be in Palantir. These things have corrected. If you still believe in the story for long term reasons, none of that's changed for any of that. In fact, I would argue that the silver story has actually gotten far better since silver peaked. And the reason is orbital data centers need more silver than terrestrial data centers. Solid state batteries need significantly more silver than lithium batteries. And solid state batteries are becoming a very, very important part of where the, the solution could be for all of the power needs that we have. These are things that were not a major story in January. They are now stories. Orbital data centers were not a story and they are now a story.

[Phil]

You're the only guy I know that is talking about silver these days, which is probably a good sign. Jory, let me ask you about Marvell. This is a stock that's had a great run this year. I know you still like it. Could you explain your thesis on that?

[Jordi]

Yeah, plain and simple. The optical side is in the very early stages. So Marvell was hated by everyone in at the end of March. And I know that because I wrote a paper on March 30 about it. So here we are, we're not even three months from that day. And yes, it has gone from about 90 to 300. I had another good one on that. It's very micronesque now. I think it's going higher because the optical side and the major benefits from it are really going to come from Vera Rubin, which hasn't even started. So I view kind of these movements as being two things. Is the stock expensive? It's a little bit expensive, but in the kind of all of the semiconductor and interconnection areas of optical, it's not that expensive. But most importantly, there Is the at some point the earnings just surprise? They haven't had that earnings surprise yet. So Micron has already had the earnings surprise. The numbers blew away what people expected a year from now. I have learned with the AI infrastructure buildout that eventually the numbers go insane. Quant strategies go on insane numbers. Human beings go on insane numbers. We have not hit that insanity mode of like a Tesla supercar where you can start going as fast as you want. We haven't hit that. We hit it in Dell, we hit it in a lot of other names. So in Marvell I still see a lot of upside. And as I talked about with the Jensen Yuan knowledge brain, Marvell was one of the ideas that came from the Jensen Yuan knowledge brain. It was just asking, hey, he's talked about this so many times since beginning of the year at CES and then through the Morgan Stanley event. What names would fit what he's describing that Vera Rubin needs. And it basically said Marvell fits it. The next day after I wrote the paper is when Nvidia announced a $2 billion investment into Marvell. And then at Computex two weeks ago, what did Jensen Yuang say on stage with, I believe his name is Matt Murphy, the head of Marvell said Marvell will be a trillion dollar company. Well, it's a $275 billion company today. So I'll go with Uncle Jensen and say that we're going to continue to go higher.

[Phil]

That's a pretty good bet and a good guy to bet on as well. What about Eli Lilly? This is an unusual quote, AI bet I think because it doesn't stand out as something that's a chip or anything related to that space.

[Jordi]

This is an easy one for me. First of all, I love companies where their revenues are growing 50 plus percent triple. The next highest one with inside the pharmaceutical industry they have a dominant drug that has demand still to come in their GLP1s and they have used that cash to basically build out an AI infrastructure which is unparalleled with inside the drug side. So let's just leave Eli Lilly alone with that. Everyone can go read my paper. They can go hear about go look up tunelab, they can go look up Lillypod, they can go look up the co innovation lab. They're connected to Nvidia, they're collected to in Silico medicine, they're connected to Isomorphic Labs which is DeepMind. This is a no brainer in terms of this company betting all in on AI and having the cash to do this. And their number one Competitor is a European company that's Danish that instead of using the money to go buy up other companies and to go do what Lilly did, they gave money back to the citizens. And I'm not saying that's the wrong decision to make, but I'm saying as an investor, we've learned in the US shareholder value, creating shareholder value is what drives stock prices. So Eli Lilly on that front is there. But from the bigger macro picture, and that's what I am at the end of the day, the amount of money being spent in this country for healthcare is enormous. In getting bitter, 20% of what people spend in this country goes to healthcare. How many people do we know that are billionaires from selling diet books or selling supplements? It's a world filled with amount of money that people will buy for anything. I know so many people that their lives have changed from GLP1s and more importantly, the data that's being collected from all the people taking them is the most valuable data which goes into Eli Lilly's lab. If you go look up Toon Lab and realize that they're allowing other companies to put their data in there. And a lot of those companies are part of the benefits that have come in, cancer that have come from the benefits of kidney disease, of addiction. All of these things that are quote unquote side effects of the GLP1s that they can't answer why. This is a very powerful story on both a macro basis, on a micro basis and where the government is intervening in the model side. The fear has been that they would intervene on the drug side and make the pricing cheaper and cheaper. The one thing they don't want to do is slow down the ability for people to anti age because it is a major need. Meaning we have a problem with Medicare and Medicaid. It is the dominant problem that's out there for entitlement. So the government on both sides needs to find a way to have these companies figure out how to make allow people to live longer. And so I think they're going to give the rope necessary to Eli Lilly. But it really has to do a lot with how much money they're making, how much more money it is in the other places and how they're buying up all of this IP in the biotech side.

[Phil]

You know, when I first started hearing you talk about Eli Lilly and then I started doing a little my own homework on was like a light turning on. It's so clear the way you lay it out. I became a shareholder a few weeks ago because of you Just like Micron, you mentioned something just now about the US government taking interest in taking stakes in these AI companies. Is that the looming bear case you think for the entire AI ecosystem?

[Jordi]

So there's no doubt that in the same way that you have the AI doomers and the AI Bubba as I call them, that I'm on the flip side that everyone calls me a permeable on AI and that's not the case. I have said publicly, which I still believe that by the time we get to 2030, I do think AI will be likely to be disrupting all kinds of businesses. And the stuff that we saw from software companies will spread to public companies. Now I want to make sure people hear this. I'm a believer in tokenization. I'm a believer that small AI native companies have a huge advantage and I'd rather invest in 1000 small AI native companies than one mega super company with inside the S&P 500. And I'll just give you an example. Give me a bunch of AI native companies competing with MasterCard and Visa all day long. I do not believe in MasterCard and Visa being able to survive through the AI side and I could be dead wrong. But I think that is the perfect example of the merging of AI and crypto payments and all the middlemen that evolve. I think AI and crypto combined will destroy these things. So I'm not a uber positive person when you go out. But I also don't make bets or investments based on five years from now. Where I think the near term is very risky is on the bottlenecks. But it's also on what you just talked about. Leopold of the situational awareness fame, whose hedge fund gets all this whatever he went from some small number to some multi billions of dollars whatever the truth is and how well he has done, he's been on the infrastructure side. If you go back to situational awareness, his forecasts on where we would be have been eerily good. What people should worry about is that he said we'd be reaching AGI around 2027. That was the whole point of this 2027, early 2027 into 2028. He also mentioned that at that point the government would get more involved. Now in the last month the government has gotten significantly more involved. And if you take it back to the beginning of the war in Iran, you have to remember when what happened with Anthropic the day before and the government, it's a very, very dangerous thing to me for public companies to have the government involved. And even though if it's done in the way that a sovereign wealth fund would do, where the money goes in and you hope that they're going to be passive investors. What he talked about, which I agree with, is these models get so strong and powerful that. But they're basically dangerous, and we have to shut off the rest of the world from having them. So they become part software, part nuclear bomb, and they're that dangerous. And at that point, I don't know what happens. I don't know when you cross that line, what happens. That has always been the fear is that if a technology or something gets so strong, the government starts to put regulations around it, it starts to own it, it starts to take it. I think that's an uncertainty for the market, that becomes very risky. I think for global, it becomes very dangerous. What has happened with Fable 5 is something everyone should really be paying attention to. I don't think we've seen the complete fallout yet. But sovereign AI, having your own artificial intelligence, what would you rather be focused on? Now, a model that if you built your entire system on a US model, but then the government says you can't use it, and a more powerful model is being built by the Chinese, which you can just download onto your computer and use it the way that you want. This is the competition from open source. So I do view the government getting involved and all the things that I mentioned before about the negatives, I think this is at a point in time where people are not thinking clearly about this and they're just focusing on. We just beat numbers and we're going to need memory for the rest of time. You can have multiple compress and have all this positive news happening. I'm willing to guess that things will look very different sometime over the next three to six months for the market and for all of these parabolic moves than it does today.

[Phil]

I think there are so many social fabric issues that are going to come about with AI, and part of this is the government involvement competing with other countries, banning other countries from certain models and all these things. You know, it's so easy to focus on just the market reaction and the view as an investor, but there are so many other ramifications that trying to keep up with. Jordi. When you think about how you as an individual have changed, as an investor, as a professional, from when you left, let's say, big firms on Wall Streets being independent, what do you point to?

[Jordi]

It's very simple for me. I never had the. So I am like a retail person now. I'm home. I don't have a job, but I'm making money, and I'm making money by helping other people make money. Now, that's what all hedge fund people would say they do is, well, I take people's money and I turn it into more money, which I'll leave it alone and let the hedge fund people describe that to people. I'm not sure that's their ultimate goal at the end of the day. And I don't think the distribution of wealth health is there in the case of, for me, being home and being able to watch AI, use AI, think about AI. Listen to the podcasts on AI. I don't think you can keep up with this stuff because of how fast it's moving if you don't have that time. So I made a bet, which as of right now, has proven to be not only true, but I highly recommend it, especially to people in college. And my son had just finished up his sophomore year. He's a very, very smart kid, and I got him to take a python boot camp in between his senior year and his freshman year of college. And he uses the models all the time. I gave him the best model. I've given him things like the knowledge brain. I've given him so many tasks to do. I had him basically do an investment thesis for me on I can't remember which company it was, but using my approach, which again was on my subscriber paywall, and I wanted to make sure that a kid could do it. But more importantly, the empowerment that it gave him, the internship that he's doing during the summertime, where he chose not to do one in the financial industry and he chose to do one outside of it. I think those things relate directly into my empowerment that I feel as an entrepreneur, my ability to pivot when I want to pivot. I didn't like working for large companies. I quit Morgan Stanley at the peak of my career because my job as a young person was already firing people as the major thing. It was dealing with management above me, it was dealing with people, and I just didn't enjoy it. And at some point, there's a balance between how much money you personally care about to live your life, what life you want to leave, and the garbage that comes associated with work. And so as someone in AI, it's very easy to balance out what brings you joy, and making money. For me, I love learning. I'm an insatiable learner. I'm sure people have gotten to know me because I've done enough podcasts now that my parents Gave me a good brain. They gave me a brain that allows me to consume lots of information and kind of be able to speak in a way that it comes flying out. There's plenty of things I suck at. There's plenty of things that I wish I could be better at. But I'm using this skill to enjoy what comes with AI, which is an ability to do rabbit hole learning and to go learn topics on so many different things. I think that has been the edge that I've had is the time that I have how I've used AI to fill that time and how there's never a time when I don't go through the okay, what is the signal? What is the alpha? What is the agency? I highly recommend everybody kind of shape their lives with this rule of three for me. To my subscribers, it's can I bring them signals? Things that read through the noise just break through the noise. The whole signal of the noise thing. With Nate Silver, the alpha, what are the names you should focus on as opposed to all of this stuff? For every micron, there's a home builder that's unchanged. There's a Ford that's unchanged for 30 years. There's a lot of alpha that does nothing. So in the benchmark arbitrage. Hey guys, here are the hundred names. These are the ones I'd focus on. Use your trading techniques. The river is flowing with you. This is like jumping in a salmon stream. You're gonna make money if you just focus on these hundred names. And then there's the agency side. If I teach you how to use it in a way, you will have the same thing that I have, which is the ability to build things and to go through things that you can build a business around. You can make your boss appreciate you. It'll get you more of a raise. If I've met plenty of people at these events where the people in their 30s are like, you know what the best thing is? I've automated a lot of my workflow, which means the days that I'm working at home, they don't even know what I'm doing and I'm spending my time trading.

[Phil]

If you had to give some advice to college students graduating into the current world, where would you start?

[Jordi]

Number one, the doomers on the job side. It's never going to happen. So as time has gone on, I've never been part of the doom crowd on the job situation. But now I can say with certainty, based on the data, that everything's been blown out of proportion. And that by the Time we. It won't happen like that. Yes, there's a depressing part of job where the corporate ladder has been smushed and every day there'll be more replacements of jobs. But the thought of being out of work, to go back to what my grandmother taught me about the Great Depression, nobody's going to be out of work. Which means, best case scenario, we're creating a little bit of job. So I'm not in any way that for college kids, learn A.I. get the skill. It literally, I'm telling you, think of it as Ironman. You either have a superpower or you don't. And if you're going to be walking the earth without a superpower and you're going to listen to your professor, don't use AI, your parent, don't use AI. Literally stick your middle finger up at all of them and just start using the tool. Start with the free version and then find a way to use this. And if you can build something every single week, start with that. And then once you build something every week, your curiosity is unleashed and I think the strength as a person, in fact, this is not a guess. As a kid who was very insecure and still you never lose that insecurity. The insecurity that I had, the acne that I had, this stuff that just made me kind of this shy, skinny kid who really didn't know where things were going to go. As time went on, you start recognizing and looking back. When you start, something starts good happening, you doubt it. I didn't deserve that. But then as they start you going through it, you're like, wait a second. The facts are the facts. I keep moving up the ladder no matter where I am. If you use AI every time that you go forward, it will get rid of that insecurity. You will start to feel like you went to the best college. I didn't go to the best college. I started at Morgan Stanley at the bottom. Eventually I was managing a lot of people that went to the best colleges. Do you think they liked me? Of course not. And if I were them, I wouldn't either because they had put in all this time working super hard, and this guy barely paid attention in school, and now he's managing me. But the reality is, at some point, if you put the work in, as we just learned with the New York Knicks, you can beat the more talented people. You can actually get through this as a team. And I think AI is the best teammate you can have to get you through the hurdle. And I advise all college kids all Young people use AI every single day and never, never stop and find out whatever you're passionate about, fantasy football, anything else, and start figuring out how you can incorporate it into what you enjoy.

[Phil]

That is fantastic advice, certainly advice that I follow every week. Jordi, where can people find your work online?

[Jordi]

They can go to AI.22vresearch.com they can go watch my YouTube every Sunday. They can go to my sub stacks. Yes, folks, there are two substacks and I will say that I'm working on something very important to me since I do believe that crypto is going to go through a new phase as the financial guardrails are necessary for the world to become more gentic. And I am rolling out another YouTube that'll probably start just after the summertime. I'm spending my summer in Maine figuring out how to be able to launch this. This will be to translate in the same way I do with AI with Signal Alpha and Agency back into crypto. So I know there's a lot of people that have never been involved with crypto in the same way. They're not involved with AI. This is another financial empowerment side. I do think crypto combined with AI, if you know those two things. And that was my goal when I decided not to stay in the work world. Working for someone. No companies that I met, and I don't care if it's BlackRock or any of them, had a plan for how to deal with the intersection of AI and crypto. And that's why my stuff is AI Macro Nexus. When I write, it's all based on the fact that the world is changing rapidly and if you don't understand these two things, you're not going to financially be able to deal with what's happening.

[Phil]

Everyone should definitely go. Subscribe to Jordi everywhere. Thank you so much for your time, Jordi. We'll do it again.

[Jordi]

Thanks, Phil.