战争期间管理一家10亿美元规模的对冲基金——（SuspendedCap）

Yo, yo, yo. Hey, man. How’s it going? What’s up, dude? Welcome to the stream, man. It’s great to meet you.

Yeah, thanks for having me. Appreciate it.

I’ll go easy on the glaze in the beginning. We could do that at the end, but I’ve become a fan, man. We’re a crypto stream, so we primarily cover crypto, and every single day, just due to proximity, we’ve been really deep on all the AI stuff. Obviously, you venture into things like Citrini, Doom article and AI SaaS, and now we’re geopolitical war speculators as well.

That’s right. Through this transition, I’ve been saying, “Dude, one of your tweets makes it into our stream doc every single day.” You’ve had a lot of face time—less screen time—on the stream recently. Do you want to start with a quick intro: who you are, what you do, as much as you want to share, and then we can get into some of the fun stuff?

Yeah. I’ll keep it anonymous, as I usually do, but I am a portfolio manager. I’m from Canada. I used to work exclusively in U.S. equities, and then kind of took over the new mandates last summer. I manage just over $1 billion. It’s a global fund, so I can really own anything.

The core of the fund is still U.S. equities because that’s my bread and butter. But it’s a good mandate because I can really buy anything that I want. At the fund level, I’m running it more like an institutional-type fund, with the considerations that go into that, but I’m also very active personally within the constraints that I’m given.

My core expertise, which I kind of came up with as an analyst, was semiconductors, and I started in 2018. I would have never guessed when I started in 2018 where things would be today. But it’s kind of worked out that where my core expertise is is where all the excitement is and where everybody’s focusing in the market.

It’s worked out well and been a good contributor to performance over the last few years, especially. So, yeah, that’s a little bit about me. I don’t know how many people on the stream follow this stuff, but I’m always bitching about Canada. I want to get down to the U.S., so maybe one day, but we’ll see.

Yeah, I was going to say I can hear the Canada, and you said “one B” really casually right there, which is awesome.

I think I kind of sound like a Minnesotan. For U.S. listeners, anyway.

Yeah, anytime I’ve been down in Minneapolis, I’m like, “Oh, you just sound like me.”

Nick Shirley mode.

Yeah, exactly. Well, dude, it’s great to have you here. The timing of this is fascinating, with so many things going on. I hate to start with a heavy one, but I kind of want to get into it. You made this tweet that I won’t read back to you in full, but it starts with, “I’m not much of a geopolitical guy, but this is ultimately, I think, pretty historical stuff from Trump,” and you ended the long tweet with, “Every time a bomb goes off isn’t World War III. This might be Trump’s legacy achievement.”

If you’re not a geopolitical guy, I really am not either, but we’ve done our best to monitor things over the last 3 days here since this weekend on stream. I’m most interested in figuring out how the market is going to react: what’s going to do well, what’s going to get hit, and how are we tracking the situation as it escalates?

Do you want to start with your general framework for how you’re thinking about being positioned through where we are right now with geopolitical tension, and what your general read is on it?

Maybe it’s worth explaining how I view this administration generally. I think there are 2 things that the administration is worried about more than anything. One is the deficit and how high interest rates are, and the other is China—its ascendancy and its technological competition, which is a new thing for the U.S.

I don’t think they’ve ever had to worry about their place technologically. Now I think China has been fairly ascendant, even though they’re kind of just ripping off everything that the U.S. is doing. But that aside, I think those are the 2 things they’re worried about.

The reason that you go to Venezuela, and the reason why you’ve got to be careful about what you’re doing in Iran, is because if you start to take oil supply offline, you will start to get inflation and higher input costs. That limits their ability to reduce interest rates, which causes problems around the deficit.

They made sure that they secured Venezuela, they made sure that OPEC is pumping, and they made sure that they’re pumping. They did a pretty good job. They got barrels down to $55, and now they have a little bit of breathing room to do what they need to do in Iran and stabilize that region.

On the other side of this, I think the goal is that you actually get—they’re not blowing up the oil facilities in Iran for a reason.

If they really wanted to hurt Iran, not only the regime but economically, they would just blow up all those facilities. But that would feed into their inflation goals, kind of going the other way.

I think what they want is a stable Middle East that’s pumping shitloads of oil, and they can keep barrels with a lid on them. That kind of helps them with their interest-rate goals. So that’s how I view it.

I don’t think they want this to spiral out of control. I think this also explains why Iran is lashing out at the Gulf partners: because they want to cause disruption and mayhem. That’s the best way to get the U.S. off their back, by pressuring barrels as much as they can.

I don’t think they’re going to be all that successful, but I think that’s how it’s all connected. I don’t think the U.S. politically wants to take this on. They don’t want a protracted war. They know that there’s no bipartisan support for another Middle East war.

So I think they’ve got a limited amount of time. I think they’re ultimately going to be successful. I don’t know if they’re going to get full-on regime change and have democratic votes next year in Iran or anything like that. You probably end up with another theocratic leader, but one who is just glad to be alive and is fairly compliant.

At the end of the day, I try to stay out of this. Obviously, the Israel stuff is very heated and divisive, and I try to stay out of that. But I think what you can say about it is that the reason they couldn’t do anything to Iran in the past is because Hamas and Hezbollah were so active and disruptive.

They’ve kind of taken the knees out of both of those groups, so this was the time to do it. They had to take this window. If you waited another year or 2, you’d let those groups recover and retrench, and then you’d have much bigger issues around ever taking action against them.

And then you’re always talking to Iran and saying, “Well, you guys are going to stop with the nukes, right?” They’re like, “Yeah, yeah, yeah, yeah.” How many times are you going to hear that until it’s like, “Okay, enough of this. We’ve got to do something about it”?

My view is, I don’t think it’s a problem unless barrels get above $90.

Okay, so once we’re above $90, what does that look like? Barrels get above $90, and what happens then?

I think what happens is interest rates start to tick up, and then people are going to start to price out the 1 or 2 cuts that are kind of baked in for the year.

At the end of the day, what’s supportive of equities is forward support, or at least bullets in the chamber for support. If you get a weakening economy where consumer and jobs data are fairly soft, it’s actually supportive of equities because it’s like, “Okay, they actually have headroom to continue to be supportive.”

If you take the ability to have monetary support, if you need it, off the table, then you’ve got problems for equities. So I think that’s the mechanism through which this pressures equities: input costs go up too high, and it takes the ability to respond with monetary policy off the table.

I’m a rookie situation monitor on some of the geopolitical stuff, but from my novice watching here, it feels like every day the timeline gets crazier. It’s 1 day, then it’s 5, then it’s 3, then it’s as long as we want. Then Pete Hegseth this morning is saying 8 weeks, 10 weeks, and we’re talking about boots on the ground.

It feels like the timeline starts to expand significantly. I don’t have a ton of history monitoring the Trump administration in a similar situation, but what is spiraling out, like you mentioned at the beginning? What does the U.S. not want to happen? Do you think it already has, or what would that look like if we were to get there?

My sense is that this has already been a little more difficult than they planned for.

Feels like it, right? I think what caught them off guard was the retaliation against the Gulf states, because on paper that looks very irrational.

I actually think it’s a good way to be disruptive, and then you get indirect pressure through those Gulf states for the U.S. to back off. I feel like Iran is thinking, “You know what? If we can just hang on and survive another couple of weeks, they’ll get pressured to stop. Then we can keep our regime, get a new leader, and start to negotiate with them.” I think that’s their view.

Yeah, yeah, yeah, yeah, yeah.

It’s very difficult in an election year with the midterms. If they start killing tens of thousands of people in Tehran, they’re going to get wiped out in the election. They know that’s not popular, especially when what happened in Palestine is already very divisive.

So I think Tehran is just trying to hold on. The sense I get is that the U.S. is surprised by what Iran had in its pocket and by the fact that it lashed out at people in the Gulf. The longer you hold on, the more of a major problem you create.

Thread Guy By the way, I’m done with the “monitor the situation” term. I think it’s getting cringe. “Narrative violation” is my new favorite. What is your strategy or stack for following everything that’s happening? Are you just terminally on Twitter?

Yeah, I’m just on Twitter a lot. I wouldn’t be in any specialized Discords or anything like that. I made a tweet a few weeks ago about a client emailing me, like, “Hey, have you seen this CNN thing?” I’m like, “Brother, I’ve seen it. Don’t worry. I’m on my Bloomberg all day. I see everything that comes across the wire.”

I don’t know. I love this stuff, so I’m usually pretty tapped in. One of the benefits of being in a small financial market is that I don’t participate in the game of telephone quite as much. I’m usually trying to come up with my own views and figure out what makes sense to me.

It sounds crazy, but I think it might coincide with AI as well. If you’re not on Twitter as an investor, things are just moving so much quicker now. Things get priced in really quickly, and I think you need to be there in real time.

I fully agree with you. Things do get priced in alarmingly quickly right now. Outside of some oil moves, which I’m relatively new to and following as well, there hasn’t been much of a stock reaction. Things are green today.

A lot of our best crypto traders got really bearish and started hammering puts on SPY. They’re driving themselves crazy right now because this thing just won’t go down. I guess the market doesn’t seem to think this is going to spiral out of control.

How are you positioned right now? I saw you buying video calls, and I saw you posting about Apple. How has your positioning changed as tensions have escalated?

I didn’t do any sort of major repositioning. I have a quite high cash balance in general. That’s been my way of protecting the downside and making sure I have capital to deploy if I see dislocations.

I had huge semiconductor positions over the last couple of years, and those have pretty well all come off. I’ve got stuff specifically in the Nvidia supply chain. I’ve got a lot of software that I think is going to work out pretty well. I’ve got Amazon, Apple, Disney, and a smattering of other things in there right now.

I’m a firm believer that there are times to make money and times to survive. This feels more like a time to survive. I’ve been pretty happy with the way I’ve traded. I haven’t made a lot of returns in the last month, but I’ve definitely survived pretty well relative to some other people.

Software has had trapdoors everywhere. The Cosbys down a percent yesterday. There have been all sorts of blowups, and I’ve been pretty happy to avoid almost all of them.

When you say you keep a lot of cash, what percentage is cash?

45%.

Okay.

Yeah, pretty high. I run a lot of options, too, so you can run out pretty quickly. If you have a whole portfolio of LEAPS and options, you’re obviously levered to the gills.

One of the things I like about options is that I can take capped-downside bets and get a lot of the same upside exposure while leaving more cash on the sidelines. That’s how I use options in my accounts.

I hold a couple of European stocks, which I usually don’t venture outside the US for in my personal account, but I found a few good opportunities there.

Okay. Speaking of venturing outside US stocks, Korean stocks have been a major topic over the past couple of months, but specifically over the last two days. We’ve had back-to-back circuit breakers, and things are in shambles.

I haven’t traded it at all. We’ve talked about it a little bit on stream, and someone on the team is huge in Korean stocks. What’s your read on what’s happening over there and the investment opportunities in that market?

I really appreciate how degenerate the Koreans are. They are one of a kind. I work with an Asian guy, and he says, “Brother, I know Asians. They’re probably taking loans from their parents right now to reload these things.”

Wait, what does that have to do with it? Is this just a known Korean thing?

My understanding is that a lot of Asian cultures like gambling. They have no problem with risk. I’ve been tweeting about memory stocks a little bit, and whenever I tweet bearish stuff or say, “Hey, this is how cycles work,” people take that as, “You should short it,” or, “You’re wrong.”

I’m like, “I can also just not play here because I think the risk-reward in both directions might be really shitty.”

Yeah, yeah, yeah, yeah. You can just do nothing.

At the end of the day, these cycles end because these are commodity bets. When prices have gone so parabolic, the idea that the companies aren’t going to respond to that price signal is crazy, especially when all the advantages in that space accrue to you because you’re the lowest-cost producer.

There is more co-development going on now, and higher thresholds for how these chips need to perform. Maybe there’s a situation where only Hynix or only Samsung or only Micron is the one that can create the chip that is able to perform better than the other two, and it ceases to become a commodity.

One of them could drive much higher margins than the others because Nvidia can deliver better performance on the back of those chips and pass it through their racks. But that’s not what we’re seeing, right? Gross margins are rising at the same rate for all these companies. It’s the same dynamic.

You see rumors like, “Hynix is going to get this,” and, “Samsung is going to get this,” and, “Micron is going to get cut out.” Then, “No, we’re not actually cut out.” The way I read it, Jensen is just playing them all against one another. They’re bringing on all this capacity because they want to contract it out.

I think there’s probably a 20% chance this is just a supercycle that will end the same way in a few years, so maybe it goes on for a little while. I’d give it another 10% to 20% chance that this isn’t a commodity because the co-development with Nvidia and the biggest customers becomes much deeper.

Then there’s probably a 60% chance this blows up in the next year, because that’s how long cycles usually last. I posted a picture of the multiples that Hynix is trading at yesterday, and it trades at around 4 times earnings or something like that.

Huh.

That’s honestly probably the most bearish thing I could think of.

I hear you. The market seems to be sniffing out that there may be limits to how much these companies can earn in the out years. The multiples have compressed to the max, and the earnings revisions for 2027 and 2028 are probably too high.

You mentioned Jensen, and I touched on this briefly. I think I brought it up as a stream topic yesterday—that you’re long Nvidia. Somehow it made the list, but it did make the list. You’re long Nvidia. What’s going on there?

I haven’t owned Nvidia for a long time. What changed my mind is that there’s really no way to catch them because they’re doing so much optimization and verticalization within those racks.

It’s really no different from the way the iPhone works. The Android phone ecosystem has to buy parts from everybody and try to cobble them together.

And then Apple makes a lot of the chips within the phone and can do all these optimizations to eke out more performance because they are verticalized within the phone and own different parts of it. That's kind of what Nvidia is trying to do. AMD has to go out and get its networking and do each piece, and they don't own the networking, so they have to buy it. They have to buy ZT Systems and integrate that, and they just don't have the level of expertise that Nvidia does.

Jensen bought Mellanox years ago, and people were like, “Why are they getting into networking?” It turns out that Jensen saw something 5 years ahead that nobody else did. When he bought Groq right before the holidays, maybe one day we'll find out what the story was there. Clearly, somebody else was getting its eyes on Groq, or they saw something or felt like they could crack something that provided a different architecture.

I think you'll hear some details about that at GTC, about big leaps in inference performance. I think they'll be able to further expand their lead against the other players. That's why I'm long now, and the stock's gotten fairly cheap, I think.

But you always have to frame multiples, right? An 18-times multiple, or whatever it's trading at, isn't necessarily an indication that it's cheap. It's the market saying, “Ah, you're probably not going to do the level of earnings that people think,” right? It may be—

But what does cheap mean exactly? This has historically been a cyclical business at times. This is obviously a very different environment and end market than in the past. You've got to keep that in mind, too.

I feel pretty good about the risk-reward. I feel like GTC is going to be a banger compared to the last couple of years because I think there are big switches in architecture that need to come. There are going to be some fireworks. I think there's going to be lots of stock movement through that event.

I want to talk to you about the AI SaaS Citrini article—all of this, which I think is out of our topic sphere. It's the one that we've covered the most on the stream. We probably dedicated a week to Citrini.

I don't even know why I want to start with the question, but what I will say on it is that when we read the Citrini article, or when we did the rebuttals, you had this really good tweet. I'm going to misquote it—it was so good—but it was like, maybe AI isn't taking everyone's job, but the world unanimously woke up to the fact that a $200,000 liberal arts degree from NYU to be an email writer was never [expletive] worth mid-six figures, you know? Some of these P/E ratios are maybe just compressing, rather than AI destroying everything.

I also think your takes on this are valid because I read your tweets. A lot of times I bring these AI bears or AI bulls on the stream, and they're blathering and covering a lot of topics, but they're not using the tools. I'm like, all right, it's kind of hard to take your take at face value if you're not using the [expletive].

I see you tweeting about Claude all the time, so I feel like you have some broad knowledge on it. I guess we'll just start broadly, and then we can go more niche on the SaaS apocalypse and AI taking everyone's job. Jack Dorsey just fired 4,000 people for Claude. Where are you on the SaaS apocalypse via AI, broadly?

Yeah, so I think a couple of things. 1. You do have to acknowledge that one of the moats in software was the scarcity of software engineers, and we've completely revolutionized how software is going to get written and how quickly it can be written going forward. I think that's definitely real and valid to try to price into the stocks. It makes sense that the multiples have come down, and there are some terminal risks that definitely need to be considered.

2. I think the starting point for a lot of these stocks is very relevant. These valuations have been ridiculous for years and years. They probably deserve to get cut in half. Maybe if it wasn't this, it would have been something else.

A lot of these companies are really bad abusers of share comp, and basically there were good SaaS companies and bad ones. Some of the bad ones were supposed to be software—build it once and sell it for zero marginal cost—but they literally couldn't bring revenue in the door without a super-heavy sales and marketing motion. There was no operating leverage in some of these shitty businesses. Maybe those ones didn't need to exist in the first place; they deserve to get hammered.

With all that being said, I think software, to the degree that all these things have crashed and it's correlated and there's so much value coming out, is crazy. The dream of “we're just going to have tokens doing all this [expletive] in the background” is such an inefficient use of compute. You'd be recreating the same thing day after day, and people are like, “You won't need software; you'll just ask your Claude to make you a dashboard.”

You might want it to make the same dashboard every day, and that's called a UI. That's what a UI is: a coding process that is repeatable day to day. You probably don't want to waste your compute and tokens recreating the same loops every single day. That's what UI and deterministic software are for.

Second, software controls a lot of relationships, data, and processes. It takes on a lot of liability and checks off a lot of regulatory boxes. At the very least, if the blue-sky dream for AI takes over in a few years, these are key partners for the labs to distribute their intelligence as quickly as possible.

They don't want to go into each business, each vertical, and each application and try to figure out what you need, how it can work, or whether you've seen it do something. It's way easier for them to sell their tokens and models through the existing software vendors.

Change is also so slow at the corporate level. You can tell people have never worked at a big business. I've worked in a few, though not for very long periods of time, and it is [expletive] impossible to get anything done at these places.

I tweeted about how long we've had an RFP going and the amount of resources and time we've thrown at it.

It's like a request for proposal. We're shopping for an application.

Right. Just the amount of time we've put into choosing one—and I'm talking consultants, dozens of people, countless meetings—and we haven't made a decision, let alone started the implementation.

We're in a regulated industry as an asset manager, so there's all this risk and disruption. What if this gets messed up? What if these numbers are wrong? It is so slow, and we're a relatively small organization. If you scale that up to a Fortune 100 company, it is not happening.

There's a reason Anthropic literally has engineers in Goldman trying to figure out what they can do. They're trying to understand what the low-hanging fruit is and where they can actually skirt around people. But this is not easy stuff. Lots of these enterprises have dozens and dozens of SaaS applications that fit together. It's way more difficult than people think.

How do you filter through the difference between “this SaaS company has a moat and AI can't disrupt this government contract” versus “AI could disrupt this business and this industry, but it's just going to take a [expletive] while because they're massive and they're a mess”? How do you decipher those two?

I try to imagine what Claude can do with what I've seen it do. I think Cowork and Code are incredible as they are today, so even if I scale up the complexity, what can I see LLMs actually outputting in the future that would disrupt somebody?

Adobe is actually [expletive] because—

Yeah, of course.

Of course it is. It's cheap now, which I get. Maybe it still works out; who knows? But the reason they're actually [expletive] is that maybe the bottom 40% of users were only using the most basic [expletive].

It's filling in PDFs. It's making a poster for a nightclub. You can do that on Canva. You can do that with Nano Banana. They're actually screwed. TurboTax might actually be screwed because that output can be recreated by an LLM.

Whereas SAP is super complex. It has petabytes of data. The whole context and operation of a giant enterprise is contained within that ERP, and you have thousands of people working in the administrative function who, at the very minimum, would have to work alongside AI within that modality.

Like, that's not coming out. So, what I'm trying to think about is whether AI can theoretically produce the output of whatever these SaaS companies are making, and who knows if that will keep me safe going forward.

Yeah, so what do you think about your industry, finance, and these asset managers and traders broadly? What impact will AI have on that? Do you think your job is safe?

So, within my team, I have a small team.

Okay.

I think we can use Cowork to codify the base layer of analysis. I would say 80% of the job, or maybe a little bit less—maybe 70%—is distilling all these different data sources, reading the 10-Ks and earnings calls, talking to management, and synthesizing it into: What does this business do? Why is it trading where it is? What are the catalysts? That's like 70%.

Got it.

Of an investment decision. It probably takes up 70% of your time, but it's probably only delivering 10% of the value of your decision.

Got it. Right? So, the way that I view it is: let me get through this 70%, because the value isn't there anyway. Now we can look at more ideas and spend more time making the judgment, which is the last 30% of the time. We can get more shots on goal at the actual value-added process, which is discussing, debating, and poking holes in it.

Yes. Feeling out what the market believes versus what we believe is way more important. That's the part that I can automate. I don't know. It's going to be interesting.

My hunch is that when you drop the barriers to information, the information edge gets competed away. I think one of the reasons the market trades so silly is because everybody is a lot more knowledgeable about the businesses. So, the set of considerations that move stocks becomes flows and narratives and sentiment and factors and all that kind of stuff. It becomes all the squishier stuff, because all the hard fundamental stuff is a lot more well-known and competed away.

This is one of my favorite evolutions: everything is narrative and vibes. We're crypto traders.

Yeah, basically everything is some level of slop. It's some level of garbage, but you have to determine if it's wrapped in gold or if it's wrapped in pig shit. How you do that is very difficult, and it feels like the world is increasingly moving in this direction.

I think the most obvious representation of that was the Citrini article itself. I read some of your posts about it. I saw you calling a couple of people out who also have paid groups, which is kind of funny. I think that one guy—I forget his name, K or whatever—that was good. I liked that.

Yeah, I was giving him shit for that. I was just saying, the guy just launched a Substack and he's getting on another Substack. It's ridiculous. Just focus on your own shit, man.

Citrini was an independent small creator, and the guy's done incredibly well. I know him a little bit—not personally, but I had one of the ideas in his “26 Trades for 2026.” It was yours?

This year?

So, I know him. He attributed it to me and all that kind of stuff, so we're cool. I have a lot of respect for him. He's a very smart guy.

But around the whole narrative thing, I think the fundamentals are getting competed away a little bit. That's why it feels like it's about narratives a little bit more.

I also think that, the way that Druckenmiller talks about it, this guy is probably the greatest of all time. He might be. It's kind of like when you hire somebody to make you a logo for $30,000, and they do it in 5 minutes. You say, “What the hell? It only took you 5 minutes.” And they say, “Yeah, but I know what I'm looking at.” You're not paying for the 5 minutes; you're paying for the pattern recognition.

I think what a lot of the time feels like narrative or vibe investing is actually just a strong ability to filter out what doesn't matter. I guarantee you, when Druckenmiller saw it, he was like, “Okay, this is going to be huge. This is going to be big.” He knew we were at the bottom of an S-curve because he's a capital-cycles guy. He was probably like, “Who's the leader in this? Oh, Nvidia. Why?” He probably had his analyst, or whoever was around him, explain it to him: “They have this software advantage, Jensen's a visionary,” and blah, blah, blah. He was probably like, “Okay, jam it. Rip it.” And that was it.

People were like, “That's hasty. You shouldn't be teaching people like that.” And I'm like, “Yeah, but that comes from decades of pattern recognition about when to put the chips in.” It's way harder than he makes it look. When you're a young, new investor, you shouldn't be investing on vibes like that, but I guarantee you the ability to make a decision like that is the distillation of decades of pattern recognition.

He probably undersells it. I've been in the industry for 8 years, and I was talking to a group of students. There's a real value in being a generalist, because you'll learn how an apparel company with a hit product has a P&L that looks the same as a trucking company in an upcycle. But if you specialize too quickly in one area, you don't start to make all those connections. I'm pretty sure that's what Druckenmiller is doing: he's just really good at pattern recognition over the years.

Of course he is. He's the GOAT. That was a sick take. Which of the 26 trade ideas was yours?

Synopsys.

Okay, what's the thesis?

They are EDA software. They basically make the software that helps tape out chips at all the big designers.

There are a few different tailwinds for Synopsys. When you're a Synopsys shop, like Intel, and you have your whole engineering function and R&D on the Synopsys stack, you're customizing right on top of that stack. There are a few different steps to tape out a chip, and the engineers don't know how to use the other software suite. They only know how to use one step.

Obviously, this is very complicated. It's a very expensive workflow, and getting to market is a huge part of the value. If you said, “Hey, we're switching to Cadence,” people would quit because the engineers don't know how to use the other stack. You literally cannot rip this stuff out.

The big value add is that they can simulate the chip once it's designed and tell you if it's actually going to work, instead of sending it off to TSMC and saying, “Yeah, it should be fine.” Then you have to wait 3 months for it to fab, and they're like, “Oh, shit, it doesn't work. We have to go back, respin it, and try again.” Synopsys can physically simulate the chip and tell you how it's going to work. There are some pretty high-end engineering problems there.

It's a niche bottleneck that you just cannot get rid of.

Yes. They also have an IP business. That was part of the reason it drew down. They were making IP for Intel basically for free because they wanted to lock people into the ecosystem. But because Intel had no customers, they said, “Hey, you have to pay us to port this IP to 18A.”

The guy running Intel now used to run Cadence, Synopsys's biggest competitor. He told Synopsys to fuck off because they didn't have any customers, so they yanked the contract. Synopsys doesn't really have anything they can do about that. That blew a hole in their IP business.

SNPS, right?

They also bought a big industrial software company called Ansys. It's a physics-simulation software suite for physical objects, structural engineering, RF simulation, and all that kind of stuff.

Obviously, that's becoming a lot more important now because we're not just talking about chips; we're talking about whole-system simulation. One of the big things is that the racks are getting denser and denser with power. In exchange for better efficiency and performance, you're going to get a lot more heat and a lot more power demand in the coming years. Ansys is well placed within that ecosystem to help simulate and make sure that all works before you actually build it out.

I think Synopsys used to trade at around 50 times earnings. It's a little below 30 now. It's not a cheap stock, but you don't have the same seat-compression risk or other seat-compression issues as some other software.

There’s no way the LLMs are outputting this. It’s deeply embedded and impossible to switch out, and complexity is good for these guys. Because we’re having so much difficulty scaling transistors down further, this is why there’s more advanced packaging going on. This is why there are different architecture innovations: everybody’s trying to figure out how we can keep the scaling going.

Synopsys and Cadence are huge enablers. They’re the only reason that companies can continue to do this stuff, so I think they’re very well placed. NVIDIA put $2 billion into them as an investment a few months ago, when the stock was a little bit lower. They’ve gotten generous.

That was such a sick thesis. That was sick. Okay, wait, answer this for me. I’m going to mess your numbers up, but you said that 70% of your time is spent digesting earnings reports and technical slop, basically, for 10% input into trade decisions. Where does the other 90% of trade decisions come from?

I think everybody goes about it a different way. I get a lot of value from watching the tape. I know that sounds crazy. In terms of really famous investors, Paul Tudor Jones is one of these guys who just stares at the tape.

Like staring at 15-minute candlesticks all day?

Not candlesticks. I’m usually watching a big universe of stocks and stuff, and I’m not doing this all day. I’m not quite that insane. I have a theory that the value I get out of it is pattern recognition over time. I like seeing what groups are trading together, where the dispersions are, what’s working on a relative basis, what’s falling apart—all that kind of stuff.

I probably take more signal from price action than the average person does. Everybody comes up with their own way to make money, and that works for me. I have a lot of respect for price action and what it’s telling me, and I try to pick my spots where I want to disagree rather than always fighting the tape. I love the tape. I love it.

I think when you first join the industry, everything that’s up and to the right is, “Oh, it’s too expensive. It’s going to collapse.” Everything that’s down in the gutter is, “Oh, I see it differently. This is going to rip.”

Druckenmiller talks about this, too. Being contrarian is overrated. The market is going to be right 85% of the time, and then you’re going to pick the other 10% to 15% of the time where you’re like, “No, no, no. This has gone too far. It’s gone too extreme. They have this thing wrong.”

When you first join the industry, you always want to be a hero. You want to be Burry, basically, all the time.

Dude, I have this theory that The Big Short has done irreversible damage to the upcoming generation. If everybody watched The Wolf of Wall Street instead, markets would be in a better spot.

Tell me this as a follow-up: what are your best and worst trades ever? It could be while you were running your own book or while you were running money.

My best trades ever were AEM and LRCX last year. They were 80% of my PA at one point: big, chunky January 2027 leaps at $100 strikes on LRCX and $250 strikes on AEM.

The last bit of AMAT that I sold was up 1,200% from my cost or something like that. I sold my LRCX ones up 3,400%. I don’t even—it actually makes me sick how much money I left on the table.

I was up almost 500% last year. It was life-changing.

Wow, congrats. But you left so much money on the table.

It actually makes me sick to my stomach. That was definitely my best trade.

How much did you leave on the table?

At least I could have been up 2x.

Oh, 500%. Oh, man.

Yeah, it’s gross. It’s gross.

AMAT is $375 today, so just the intrinsic value, with a year to go, is $12,500 per contract, plus the time value. I don’t even know what they’re trading at. They’re probably trading at $15,000 or $16,000 a piece, and I was paying around $1,000 per contract for these things. I put 80% of my account into it.

LRCX is even worse. I think I was paying around $500 per contract for the $100 strikes, and they’re worth $15,000 a piece on intrinsic value alone today. It kind of makes me sick.

It was a great trade. I was in the trenches in 2022 with a lot less capital. It was a really rough year. I was young, dumb, and took too much risk, so I’m a lot better at being prudent and controlling risk. Obviously, if you’re able to go up 500%, you’re definitely able to go down 50%, 60%, or 70%. I don’t take that lightly.

I had a lot of conviction, but it’s always easier in hindsight to say, “What if?” That was probably my best trade. I followed those stocks for years and made a lot of money for the fund I worked for with those names over the years. I thought, “If I get those prices, I’m going hard.”

Around the Liberation Day stuff, I got my prices. It worked out well, so that’s probably my best trade.

As for bad trades, I don’t know. I didn’t get sucked into a ton of stuff during COVID, but I got sucked into Peloton and lost some money.

Jeez. I haven’t heard that name in a while.

Yeah, stupid. I learned a lot of lessons. I had a pretty bad China trade and lost quite a bit of money on KWEB, expecting a recovery that never came. Then, of course, as soon as that trade rolled off, that’s when China started to work. That’s usually how it goes.

Yeah, yeah, yeah. Right now, I’ve eaten quite a bit of shit on Builder FirstSource, expecting a housing recovery that’s just not materializing in the U.S. Even with mortgage rates coming down quite a bit, and even when we were below 4% on the 10-year, these stocks weren’t reacting. I’ve had a few bad trades for sure.

I’m still pretty bullish on certain aspects of where the NVIDIA architecture is going right now. Everybody is focused on optics now, with Lumentum, Coherent, Ciena, Applied Optronics, and all those guys. They’re moving to co-packaged optics.

The other part of the trade that people aren’t as keyed in to yet is the switch to an 800 V architecture within these racks. Everything is usually 48 V, but it has to go to 800 V because the amount of current you would need to pipe in at 48 V is 18,000 amps. You would need copper busbars that are a foot thick and weigh a ton. It’s not viable at all.

In lieu of copper, you need a lot more high-power-chip content going forward. In the Kyber racks that they’re going to be showing off the reference design for at GTC, you’re probably going to have 3.5 to 4 times the amount of power-chip content in there.

What’s the trade there?

I have one. It’s a German company called Aixtron.

Aixtron.

They make the machines that deposit material onto wafers. For silicon carbide and gallium power chip wafers, you have to grow these crystals on the wafer, and Aixtron deposits the film to grow those crystals. They have a near monopoly on it.

The other big silicon-carbide and gallium-nitride markets in the past have been EVs, which have been terrible for years, so a lot of these stocks have been down in the dumps. Aixtron reported some pretty rough results, but they said they were getting a lot of orders, so the stock has started to work.

The other theme that’s really important is that the reason chips get more efficient—at the single-die level—is that the transistors keep getting smaller and closer together microscopically, so the data doesn’t have to move as far. Now that they’ve exhausted that, what NVIDIA has to do to eke out every last bit of performance is bring the literal components of the server rack closer and closer together.

That’s why HBM is stacked right on top of the GPU: it’s a shorter distance for the data to move. There are going to have to be all these innovations in packaging to bring everything closer and closer together.

There’s a company called BE Semiconductor, or BESI. I think it’s a Dutch company. They’re European, and they’re a leader in hybrid bonding. That’s going to be even more in focus post-GTC. You’re going to need a lot of packaging innovations to smoosh all these chips together so they can be as close as physically possible.

That’s going to be a scaling vector for NVIDIA going forward, allowing it to eke out every last bit of performance.

How are these stocks? Okay, we have some more research to do. What is your thought on the AI bubble broadly? How do you think the market’s going to play out for the rest of this year and then maybe through next year? Are we topped, or do we still have a lot of gas in the chamber?

I don’t think we’ve topped. I tweeted about this too in regard to software: I actually don’t see any scenario where the AI bubble or the AI build-out can continue without software working.

The way that AI is going to pay for itself is through labor replacement or productivity. There is no way for AI to distribute the intelligence that it needs quickly enough without the software players.

For this to work, I actually think the AI blue-sky scenario is that, yes, you had some dispersion as people chewed through this new way of building software, but going forward they might become a lot more correlated. I think it’s really just going to be a mismatch in how quickly people can integrate this stuff and see benefits.

To me, the shortest route is labor replacement. I think that’s the only way that you would actually see this kind of build-out get paid for, and I think that’s going to happen via software. So that’s probably my big contrarian take right now: software actually has to accelerate for this all to work.

Do you think the Block layoffs were because of AI, or did they just overhire?

They overhired.

You do?

Yeah. I have no doubt that tech organizations are going to be way quicker to implement this. The whole thing I’m talking about—where people say they’re never going to change anything because it’s so corporate and bureaucratic—applies more to Caterpillar than it does to Square, for sure.

I think they overbuilt, and now they can use these tools to get lean. That’s going to be interesting. If Square is charging, let’s say, 2.5% per transaction on its payments, what if they start running that organization at a 50% or 60% lower cost base and can come out and do it at 2% flat? That becomes competitive pressure. I do think there’s a chance of that.

The AI thing is the real deal. I don’t think there’s ever going to be enough compute for a decade, but the capital markets may call time on somebody in the ecosystem—whether it’s Oracle, Microsoft, Meta, who knows. The market is going to say, “You better stop spending because you guys are burning capital, and we’re not going to underwrite it anymore.”

That’s probably how it ends—not, “AI’s kind of shitty and people don’t actually need it.” It’s going to be a capital-cycle thing rather than a demand-driven thing. That is a very different complexion for the semiconductor market than in the past.

We will never actually have enough compute. I know that might sound like a big take, but we’ve never in history had enough compute. Every time we find a new way to compute—

Yeah. Especially now, when capability is more directly tied to compute than at any other time in the past, I don’t see why that would suddenly change. Everybody’s praying for a DeepSeek moment now. When it happened, it was, “Oh, maybe we don’t need as much compute.” Now people are like, “Oh God, we need a DeepSeek moment.”

We’ll get one tomorrow when it launches, right? Or whatever. People are just going to use 2 times the amount of tokens if they can be twice as efficient. It’s going to get twice the usage.

I’m not bearish, but NVIDIA is supposed to grow 25% next year. If the capex guys are doing 6 to 10% and say, “Actually, we’re going to do 15% next year,” these stocks are going to get cracked hard.

[Laughter]

Not for the delta between 25% and 50%—they’re going to get cracked by 30% or 40%. That’s what you have to keep in mind.

Semiconductors are the best end market for alpha because you combine the demand curves of technology with capital intensity, and that’s why you get such whippy cycles. It’s a great hunting ground.

Dude, this was sick. SuspendedCap, you’re one of the best. I had to tell him, “Guys, I’ve never heard this guy talk, but I really don’t think he’s a chud. I really think he’ll be worth staying around.”

Any final remarks or anything you wanted to say that I didn’t ask? Anything you want to shill—something personal, a ticker, whatever? You got some shills in.

Yeah, I talked the book a little bit. I’m pretty bullish. I guess the only other one we didn’t talk about is Apple. I know it’s not a great stock, but I think that Apple silicon program has differentiated them for years. I think it’s going to matter more now than it ever has because they’re actually producing hardware where you can run edge AI, and I think that’s going to be really important in the next year or so. That would be the only other one.

We’ll leave it there. All done. What do you think of the new Mac they just announced?

I think they’re going to blow away numbers and estimates, at least for the Mac division. I kind of want to buy one. I have 3 Macs, but I kind of want to buy the new one. The thing is fresh.

I’m sure I have some shitty Alibaba coin model that runs my OpenClaw on my 3-year-old Mac mini, so I can only imagine what you’ll be able to do with a fully juiced-up one right now.

It’s interesting. I think within 1 or 2 years, that’s probably also a bear case for NVIDIA and the cloud guys: some people will just run this stuff locally over time. Apple is like—if you’re going to own NVIDIA, owning Apple over the next couple of years is actually a good natural hedge, I think.

I actually like that take.

Yeah, I really like that take. Sick, dude. It was a pleasure, man. I appreciate you coming on again. Hopefully we’ll do it again in the future, and good luck with everything.

I appreciate it. All right, peace, dude.