00:00 In this market, long-term uncertainty is so high that anything with short-term certainty is trading at a huge premium. Whenever you get heavy retail involvement in any trade, that usually in my experience has meant that the clock is ticking for the end of that trade. There's been maybe a disconnect between the financial analysis of this and the religious analysis that maybe comes from the West Coast.
00:22 There's going to be a lot of pain, I think, if this ever rolls over. And I just don't think it's yet. Welcome to Other People's Money. I'm Max and I'm joined today by Jeff Keller, founder and portfolio manager of Capeite Partners, a technology sector focused hedge fund that he's been running since 2021. Jeff, thank you so much for joining me today.
00:43 >> Hey, thanks for having me, Max. Big fan of yours. So glad to be here. >> I'm a fan of yours. I've been really enjoying following along with Capeite and everything that you've been doing. There were a lot of tech focused hedge funds that were birthed in 2021. A lot of them rode the tech sector up and then they rode it down in 2022.
01:01 And so just seeing how well you navigated 2022 and the undulations that we've had in the tech sector has been something that's been a pleasure to watch and observe from the outside. That's why I wanted to bring you on today because it does feel like since the start of Q3, we have had a bit of shift.
01:22 I actually ran the numbers as of yesterday's close. So since the start of Q3 software and services are up 20%. Technology hardware and equipment are basically flat but if you go down to semiconductors and semiconductor equipment we're down 7 and a half% to start the quarter. Clearly a massive divergence from what we saw to start 2026.
01:46 So what are you seeing and maybe a little bit about your background to give us the perspective that you're coming at from the tech sector. >> My background is a little more operational. So I spent some time actually working at Salesforce and MongoDB. So a bit of a software background and I think that maybe orients me a little more towards the big themes, a longer term horizon.
02:05 Looking for extreme dislocations of markets. I think you had in the beginning of this year an extreme dislocation actually where the war in Iran was distracting everyone from what was just massively inflecting adoption of AI. And that's been a long-running story, but you really had the takeoff moment in Q1.
02:22 And obviously Q2 was, I don't know that it was the blowoff top for this trend, but you had 100% gain in socks. You had retail get heavily involved. And so it's not surprising that that trend has chopped around a bit. And obviously, we had situational awareness and deleveraging in Korea. So I think some of this volatility is certainly obvious in hindsight.
02:41 It's to be expected. I think we're still moving upwards on an S-curve with regards to AI adoption. And the bulls would say, hey, Apple services is still growing double digits today. Google search is still growing double digits today. Meta is growing in the 20s. And so you could have 10 to 15 years of double-digit growth for some of these AI labs.
03:00 And that'll lift all boats. So I think the volatility is to be expected. I think software was thrown out with the bathwater, obviously, and so some of that bounce makes sense. But if you zoom out, I don't think that late June was the end of this mega trend. I think this is just kind of normal volatility.
03:16 And so within software obviously there was the big disruption narrative and it's hard to say whether that narrative has been disproven. If you take like the analog of newspapers and how much the market was able to suss out the pain the newspapers were going to go through before it actually showed up into earnings.
03:35 So if you are in the disruption of software camp, it's really easy to point to examples like that and say, "Yeah, software looks strong now. We really haven't seen the disruption, but the results have been extremely strong." And so when you look at this rally, is it the market saying that software actually isn't as much at risk as maybe we felt at the beginning of the year, or is it more technical in nature? Because software shorts were paired with a lot of these AI longs.
04:03 As much as the big deleveraging event has happened, degrossing is still happening across the whole hedge fund complex. >> It's been entertaining to see socks and IGV basically have a negative one correlation. That does suggest that there's clearly a technical element to this that some bad news for AI leads into it and ServiceNow and Salesforce stocks go up several percent.
04:25 So that's an odd dynamic. I think you split software into application software on the one hand and infra and cyber on the other. I think the application software obviously is still under question. I think a lot of what we've seen so far is just a recalibration from that sector trading at a premium to the rest of the market, being double digit growth for a decade now being maybe low double digit or high single digit in some cases.
04:50 And once you get into that five to 15% growth area, it's really hard to trade on revenue multiples and non-GAAP EPS and it's really hard to dream the dream. So I think with the Salesforce, Workday, obviously we've seen an acquisition potentially of Workday, but I think the initial crash was kind of coming back down to earth and it wasn't actually pricing that AI was going to disrupt the entire business.
05:16 It was just these are low to moderate growers with a little bit left tail risk. So I think that was somewhat rational. They've bounced a little, makes sense, but you've really seen the bounce in cyber and infra. And I think part of that is anything usage based, anything with the ability to accelerate.
05:33 I think in this market, long-term uncertainty is so high that anything with short-term certainty is trading at a huge premium. And so people know that cyber numbers are going to be good because people are rushing to build up their defenses with these new AI threats. People know that infra can accelerate because you see more software adoption in general.
05:51 You see at least some portion flow into existing tools. I don't want to say I'm bearish on that yet, but those multiples have come a long way. You're paying sort of 25, 30 times revenue for large companies. I think that the success rate on those is paying up there is low.
06:09 Obviously, the next couple quarters are going to be good. But cyber is also not immune from disruption threats. It's a new paradigm. You talk to folks at some of the AI labs and they think cyber is very much something that they will disrupt as well. So I think that's TBD.
06:25 I think it's a lot of momentum and they're going to beat for a little bit has driven them up. But I think that's a lofty territory that they're in more so than AI, frankly. I'd argue that the real AI winners are CrowdStrike and some like that rather than maybe some of the semis. >> So within the infrastructure, what are you talking about? Can you give me some examples of infrastructure and why those names haven't been under the same pressure as application? >> Yeah, you look at a company like
06:53 MongoDB where I used to work so I'm biased but AI is clearly leading to a proliferation of software. People are creating either vibe-coded apps or they're just able to develop more efficiently within large organizations. And every software application has a database line underneath it.
07:12 And so when you have an explosion of your end market, that's going to lead to an explosion in database software. And there's a question of how are these databases being selected? It's no longer human. It's now various large language models. So it's not clear how the mix shift works but I think the view is a rising tide may lift that boat.
07:34 That's an example. Obviously you've seen some pick up with like CDNs or the Fastlys and basically anything that can deliver computing resources that's driven on more compute flows through to their numbers. I think that they've seen some uplift and so things that are usage based certainly in the short term you can see a pretty quick spike.
07:57 I think the long-term still more under question but I think infra there's a higher chance of beating than there is in a Salesforce or Wix or something that's a little bit less likely to really blow up numbers in the short term. >> It sounds like you're not concerned so much with market share, right? Like a lot of times with these more mature sectors, it's all about like who's winning within the sector.
08:20 And do you think that within tech it's less about those types of bets and more about where is the pie growing? Like is this a relative betting market where you're really trying to pick winners between these competitors or are you just trying to see like where is the wind at their backs? Ultimately the work that I think one has to do is bottoms up and company driven and who the winners are.
08:46 But I think in technology a lot more money gets made being in the correct major theme than it does picking within those themes. If you were 10 years ago aware that cloud software was taking share and there was a big migration to the cloud, certainly there were certain names that were better to buy, but for a while there it was more important to be in the sector.
09:05 I think that's what we've seen with AI and you start to get dispersion, but early on I think it's much more thematic. It's much more sector driven and over time as the open-ended growth story slows down, then it starts to be about more dispersion. I think we actually have a little bit of that with AI where it's still a pretty open-ended growth story, but you've had huge — the numbers now are sort of GDP scale and so I think you're starting to see a lot more dispersion where it does matter who has more sustainability
09:33 rather than just this rising tide in the short term is going to lift everyone. >> Well, let's talk about that dispersion a little bit. I'm sure obviously you have this software background, but if you're a tech focused investor, you probably have some exposure to some of these AI themes.
09:49 Post shakeout with what you have seen from the market where do you think investors should be within the big AI trades? I also ran the numbers on just the S&P 11 sectors and technology came in third. The only things that were worse were industrials and utilities which were really thrown into the AI trade.
10:14 So that has been the laggard coming out of the bounce in the market. So where do you think if people still want to be exposed to AI, they should be looking? >> I guess I'll start maybe with a negative. I think one risk is to always extrapolate a very short-term dynamic. I think you're seeing a little bit of that with the pricing of compute, certainly maybe with neoclouds.
10:34 Obviously, Elon is entering that space. So I do think that there's a little bit of a risk that this very short-term extremely high price of compute — Elon might move us quicker to the glut. So I think that extrapolating that is a risk. The other thing is it's funny that of all these various, whether they're bottlenecks or semiconductor companies, they trade on the same factor and they trade on basically hyperscaler capex and the ARR of the labs. Like they need
11:02 the same underlying drivers and they trade at hugely different multiples and I think Gavin Baker's made this point which makes sense. You've got a lot of behind the meter, you've got some services companies that are trading at pretty high multiples on outyear numbers and then frankly you have things like Nvidia and Micron and the well-known names that are really not extrapolating.
11:24 In fact, they're telling you we're at peak. And so maybe it's a little contrarian, but I think it's likely that the low multiple names across the AI trade in general are going to do better, even if they're at different cycles of earnings growth. But it's all driven by the same factor. There's little micro cycles within each.
11:40 But I think that a lot of the semis are not extrapolating as much as some of the kind of more creative whether it's behind the meter, some services, highly cyclical businesses, kind of hidden AI winners that people have piled into. Those I think are pricing in a rosier future than actually a lot of semis.
11:57 >> And valuation is so different as you said across this — how are you thinking about what is or isn't cheap? How far out are you looking that you feel confident on earnings? Are you valuing these things off of 27, 28 and then between software, hardware, right, you're just talking about very different multiples.
12:20 How are you thinking about valuation across tech right now? >> It's a good question. It's funny to think about how many people, myself included, we'll talk about four or five year projections and you think back four or five years ago and just completely different things we were talking about five years ago is NFTs and SPACs and the metaverse and AI was not on the radar.
12:39 I think there's a big risk with valuation. I'm contradicting myself a little bit earlier, but where a lot of these are highly cyclical businesses, they're obviously trading at maybe it's not peak earnings, but massively inflated earnings. You have a new profile investor, you have a lot more growth investors and tech investors.
12:58 And so multiple can be dangerous. Even though I think that the multiple — there's extreme dispersion and there's some that are very cheap. Just relying on multiple can be very dangerous because there's high cyclicality here that a lot of tech investors are not used to.
13:11 But I think if you have good visibility into three years, beyond that is just hard to forecast in technology. Again, you look back to five years ago, very hard to predict what we're doing now. And so I think if you've got three-year visibility into stocks and valuation, you can do well.
13:27 I think there's a very high premium on one to two quarters. I think that's a different game that I try to play, but it's generally a two to three year through cycle earnings multiple with decent downside protection. >> Okay. But as you said earlier, you think there is a premium on that certainty right now in the coming quarters.
13:48 So as a portfolio manager, how are you bridging the fact that you like to take a bit more of a long-term view with knowing that the market is really rewarding short-term certainty and short-term growth. >> That's the beauty of the long short model is you can bet on both sides of the book. You can try to isolate some of these factors.
14:06 I think that maybe it's anecdotal, but I think that the time horizons of people that I talk to is probably shorter than I can remember. Again, I haven't been doing this for 20 years, but I've never heard more conversations about what's the catalyst path and who's going to beat sellside consensus numbers.
14:24 And whenever you see people that are that short term, I think you can take a look a little bit longer. I think it's never as easy as simply this company's going to beat earnings for the next quarter, even though it works like that sometimes. I think there's got to be a longer thesis.
14:38 So shorting is a dangerous game. You've got to have awareness of how much people can extrapolate, how juicy some of these near-term imbalances can get. But I think if you have a year or two of horizon, then you can look at these and say, "Okay, where are people extrapolating and where can I sit and wait?" And I think that's a pretty good opportunity.
14:58 >> So you're talking about that earnings feedback loop, right? One of the other interesting factors from earlier this year that felt different to me was you had a feedback mechanism with a lot of these partnerships deals. There were catalysts really outside of earnings. One semiconductor company would make a deal with Nvidia and the stock would rip 10, 20%.
15:23 We're still seeing those big sorts of deals, right? Somebody would announce a huge bond offering to fund capex or some other sort of buildout and the presumed receivers of that capital would rally. We're still seeing those sorts of announcements, but we're not seeing that feedback loop in terms of the market's reaction.
15:49 What is that telling you that this feedback loop has really dampened its effect? There's certainly a lot more sobriety in the trade. I think there's a lot of 1990s parallels. Obviously, I was not investing at the time, but it's funny to even compare it to 2021. The euphoria was multiples higher in 2021 than it is now, which is interesting because I think we have a much more bankable long-term technology trend.
16:12 So I think the sobriety in a way is encouraging. To your point about the lower reactions to some of these announcements, I think that there's a lot of announcements where the economics are unclear, especially think about with data centers. This company is leasing from that company. We don't know who's on the hook for extra costs.
16:31 We don't know what happens if there's a delay. And so we've gone through this phase where any association with a year ago's OpenAI, this first half of this year, any association with Anthropic was gold. And I think there's definitely a lot more scrutiny to say, are you actually going to be generating sustainable long-term profits from this? I think you've seen that with the crypto miners converting to data centers.
16:55 Euphoria has totally come off there because I think there's concerns. >> But now Bitcoin's ripping and they're ripping along with it, too. So they get the best of both worlds. >> Yeah, exactly. They might have to shut down the data centers and go back to Bitcoin mining.
17:09 No, I think you're right that there's more sobriety. I think it's another point where it actually makes me a little more bullish. You'd rather be in an environment where you have to sit and earn the returns than one where you get stocks moving 20, 30% on press releases.
17:24 So I think it's better to have a longer slower cycle especially as a manager of investments because it's much easier to stay with things that compound over time than it is with stocks that go up 20, 50% on press releases. >> I think it's also giving an interesting picture into some of the newer holders of these names, right? If you go type in a ticker on X, your favorite tech ticker that just had one of these big announcements and the stock hasn't moved, you're going to see a lot of commentary from retail
17:56 sort of like cheering on the deal, right? Cheering on that deal and wondering why am I not getting the 20% pop that I got when this type of deal was announced three to six months ago. And so I'm just wondering how you're thinking about the holder bases and the way the new participants who have been coming into this trade.
18:19 >> One of the things you want to be cautious of is whenever you get heavy retail involvement in any trade that usually in my experience has meant that the clock is ticking for the end of that trade. Now I'm not saying that I think the AI trade is over but you saw it with gold and silver in the beginning of this year.
18:34 You've obviously seen it with crypto. We saw it at one point with software certainly in 2020 and 2021 when people would be launching paid newsletters writing about MongoDB and Elastic and all these companies. >> Fastly. >> Fastly. Exactly. Yeah.
18:54 Yeah. And so I think there's reason to be cautious there. I think for sure that you have these boring hardware companies for 10 years that are now all the rage among retail investors. I think that's certainly dangerous. The flip side is again this I think has a lot more legs.
19:11 It's got a lot more of an open-ended growth story. One thing I've learned is shorting an open-ended growth story can be deadly because there's no catalyst. There's no end to the story. You might get a data point that's helpful in your direction being negative, but it's really hard to put a nail in the coffin on an open-ended growth story.
19:29 And AI is a good open-ended growth story. I think it's a real one. You might have some hiccups, but in general adoption is moving up the S-curve. Capex is going up. I think that's the case for a number of years. And so I think that's what differentiates this. It makes me feel that this is not a 2022 environment.
19:46 It's not like June was the blowoff top and we're going to be, semiconductors are going to be bleeding for years. I think that this couple months is a little bit more of an information vacuum. People are latching on to every ARR leak of Anthropic or OpenAI. I think that'll get clearer when they become public.
20:05 I think it's good to clean out some of the leverage. I think it's healthy in general. I don't know that we're going to rip to all-time highs, but I certainly like short selling. I'm not piling into shorting AI here. >> We've made a few analogies back to 21. You just talked about 22.
20:21 We had this huge deleveraging event around situational awareness in 21. It was Archegos and it was a lot of the Cathie Wood ARK names. To compare those stocks and their businesses to what we're seeing from AI is completely disingenuous. So, let me just start by saying the fundamental case for these AI stocks compared to the speculative case that was being made for a lot of those names in 2021 is completely different.
20:50 With that being said, the market loves to extract as much pain as it possibly can. And that deleveraging event in 21 really did mark the top of the ARK names. The market continued to power higher until inflation concerns really started to take everything down besides energy in 2022 and we had that rising rate environment.
21:16 And so I just wonder how you think about the technical factors that we're talking about here and that as strong as the open-ended case is that we could still see a scenario where the market moves higher, a lot of these other themes like software that are continuing to have strong earnings. So many sectors are starting to participate in the bull market.
21:40 Is there just a little bit more pain to come for the tourist semiconductor investor before we start to see the market turn back to the fundamentals? >> I think the 2021 analogy is interesting. I might quibble with it in a few areas. I think you're right that it was blowoff top retail involvement and it took a while to get through that pain.
22:01 There's a lot of ironic similarities to just incredibly leveraged hedge funds driving up a few of these stocks. I think even in 2021 stocks like Tesla, software eventually kind of hit their peak in late 21. So they recovered — the quote-unquote real businesses did okay even if the SPACs and the GameStops and whatever may have cratered around Q1 of 21 as you mentioned.
22:27 So ultimately I do think it was inflation, higher rates that did a lot of the damage in 2022 for the real businesses and we were coming from I think arguably more indefensible multiples back then, 50 times revenue on software when clearly software had gone through a pandemic pull forward for cloud adoption.
22:47 AI today obviously there's some pull forward. I think there's more legs to it. I think the multiples are actually more reasonable. And so I think that there's interesting analogies. I don't think that this trade will break until you start to see adoption slowing. I think right now the reason the market gets nervous about ARR metrics or whatever is that's the best indicator we have is Anthropic, ARR, open source, whatever it may be.
23:11 But as long as in general the complex is seeing more adoption, which it is right now and it's explosive, I don't think you want to short them. And are they going to rip again? Unclear, but I think there's still a lot of open-ended growth in this story. And so that's where it's different, I think, than 21.
23:28 >> Now, it's another one of those sort of rhyming but not matching perfectly. Obviously, there is some debate about if and when the Fed might hike rates. We're probably going to get something in the next few months, whether it's the next meeting or not. The market is pricing that in with near certainty, but that by the end of 26, we're going to get something.
23:49 But it's not going to look like 22, right? We're not going to come off the zero bound up to the level of interest rates that we saw. That was obviously a huge hiking cycle, but we're seeing fiscal concerns come into play and arguably those real businesses that were hurt by rising interest rates are more rate sensitive now than they were then.
24:11 We're seeing a lot of this spending funded via the credit markets to the point that they've even commented that it's crowding out the Treasury market, the biggest market in the world. And so I just wonder how you feel about the rising long-end bond yields and that everybody seems really focused on whether the Fed is going to hike, but the market is pushing financial tightening much more so than monetary policy is.
24:38 And do you have concern about the ability for these companies to continue to raise if we have rates at this level at the long end of the curve? >> It's definitely a concern. Far greater minds than me have spent a lot of time trying to figure out which way the bond market's going to go.
24:55 So I won't forecast what's going to happen to rates. Certainly we've seen a number of these bond freakouts and rates go up and then they come back down. Who knows if this time's different. It's certainly possible that two months from now the 10 year is down 30 or 40 basis points for whatever reason and we have less capital concerns than we do right now.
25:14 But for sure, when you look at it now, it's a capital crunch in the sense that you're talking about hundreds of billions, pretty soon near trillions of what used to be buybacks now actually going away and also requiring debt issuance. My view is I understand it's a concern.
25:34 This is a more dangerous way to finance this buildout. I'm not that worried about it, frankly. Especially after the Anthropic IPO, these major — the labs and SpaceX and Google will have raised about $500 billion year to date. And the markets are fine. A lot more in debt as well, or maybe not more, but a lot of debt as well. And markets are broadly okay.
25:55 I think the cash — these companies are still growing top lines double digits, 20s for some. Frankly, I think there's probably more cost base they could take out if it ever got so bad. These are companies that employ tens of thousands of people. They did it in 2022. So I think that there's still enough room to fund the buildout.
26:13 I think capital markets are more volatile and subject to euphoria and fear. So you don't want to go through periods of the markets shutting down, but I think they've generally been open. The market's taken it in stride so far. And the other thing is the way that the Silicon Valley leaders think about this is they view this as existential.
26:31 This is essentially the greatest race of their career. This is the pinnacle. You see Sergey and Larry coming back to Google to work on this. You see Elon basically pivoting all of his attention to building data centers. This is the big one. And if they need to borrow at 150 bips higher than they thought they did, and it ends up in aggregate being whatever that math is, big dollars, 10, 20 billion dollars more a year, I don't think that's going to slow them down.
26:56 I think there's been maybe a disconnect between the financial analysis of this and the religious analysis that maybe comes from the West Coast. And I grew up on the West Coast, so I like to think I can do a little both. But I don't think they're slowing down. I think the credit markets might try to slow them down.
27:10 But unless we see yields really blowing out, I think there'll be enough money to fund it for a couple years. And if they're wrong and the ROI never materializes, we'll have a problem. But I think their bet is in a couple years it'll be obvious that this is a transformative technology and so a few extra bips of borrowing cost is not going to slow us down.
27:28 >> Now what do you think about — obviously we saw Google report its first negative free cash flow quarter. Is that coming for the rest of the hyperscalers and how do you think markets are going to think about the Mag 7 and the hyperscaler names that were the market leaders — they are definitively not the market leaders right now. Is that trade of just buy the big megacap tech companies, do you think that's over? Not to say that they won't keep up with the market but
28:02 that it's not going to be the easy trade to outperformance. >> I think so. I own some of them. The stocks, I think, are reasonably cheap. I think it's just a very different profile than it was over the number of years. Amazon, you can check me on this, but is probably up something like 10x since 2016 or 2017.
28:24 You made extreme returns in owning the most well-known companies in the world. Can you still make 15 to 20% a year or whatever it may be from here? Absolutely. But I do think you're right that the profile of return that people associate with these companies is probably behind us.
28:40 I think also the hyperscaler model which drove a lot of the performance of the Mag 7 — obviously not Apple and not Meta but the other three, a lot of it was cloud business. That market is dramatically changed from basically three vendors with thousands of customers to now there's probably nine or 10 scaled providers of compute.
29:01 You include Oracle and SpaceX and the labs who are procuring directly and the neoclouds and CoreWeave and Nebius at least. So you have a market that went from maybe three to 10. They still have scale advantages, but you have fewer customers. So I understand why those stocks are not trading at higher multiples given their growth.
29:21 I think there's a lot more uncertainty. There's more competition and you've got a little bit of the law of large numbers here just where you're talking about again GDP level market caps and profits and whatever. So to be able to get two, five, 10x returns on those I think that was a golden period over the last 5 to seven years but I think you're right that that's over but doesn't mean they can't still be good risk-rewards.
29:42 It's just a different profile. >> Now, I just pulled up Google's adjusted free cash flow estimates, and looking at annual numbers here, and it's like 27, so it's like 18.8 billion in free cash flow and then 28.5 positive free cash flow in 28. Does that presuppose like too much certainty on the return on this capex? Do you think that the path to return on the spend is that clear, in that immediate, or are these firms really betting on five, 10 years out that that growth that we said is really hard to predict? >> It is. I
30:23 think there's a lot of focus on what's the ROIC of this investment and there's obviously many opinions. The way I think about it is it's almost like buying an option. It is existential if you don't play this game and it wipes out your business so you have to play. So it's a defensive option number one. Number two, it's an offensive option because you see it in the cloud numbers.
30:46 There's other ways that you can generate new products. There's just sort of this open-ended upside that maybe if we become the leading provider of compute then whatever it may be there's open-ended upside. So it's really an optionality thing. And I think the way they've talked — I think Microsoft talked about this in the latest call, maybe it was one of the other hypers, but you can structure this where you're buying powered shells, you're not filling it with the chips yet. You're essentially
31:08 buying a couple years of optionality where if you need to, you can ramp up a lot of capacity really quickly. So that all makes sense to buy that option. Now, if the ROI on that option is low but positive, I think that's still a good investment for them. People think about this as a concrete —
31:28 you put $1,000 into the ground, what are you getting out of it? I think of it as you got to put $1,000 into the ground because if you don't, you might lose your whole business. You got to put it in because it might be this open-ended upside. And I don't know if it's going to pay you back 15. And frankly, I'm okay actually.
31:42 Shareholders might not like it, but it's rational to do it even at lower returns because the alternative is too risky. So I think that has a couple implications. One is we don't know what the ROIC is, but I think they're going to keep spending because of this need to accumulate options. Number two, it might not be great for shareholders because if the ROIC is only like four or 5%, that's not great.
32:02 It's kind of a waste of capital, but it's not incinerating capital. But I think the main takeaway is they're going to keep spending because the risks to stopping, I think, are much greater than the optionality you buy by at least continuing to stay in the game. So I think it's very hard to pin down the ROIC.
32:20 I think it's pretty good, which you can calculate in a number of different ways, either through numbers right now or just anecdotally about the profits that are flowing through the chain. But I think it's actually more driven by basically the need to maintain optionality than it is near-term ROI. >> Going back to what you said about the market is really rewarding short-term certainty right now, you could argue that the reason that the hyperscalers have not participated is the uncertainty around that ROIC. When I look at
32:49 estimates at least for Google's free cash flows, it says to me that the confidence people have in those free cash flows turning around in 2028 seems to me that they're willing to at least with a range of outcomes say, "Hey, we're going to start to get some return on this spend two years out from now.
33:10 " But what if 2027 comes around and we're not getting that visibility into 28 into this ROIC? And it's not that it's negative. It's not that it's going to be overwhelmingly positive. We just don't really know yet what the return is going to be. Do you think that that uncertainty is going to cause continued problems for the hyperscalers? >> Yeah, I think it's a good question.
33:32 I think there is a world where to the point earlier, this used to be a three horse business and now it's eight, nine or 10. The ROI, where the profits flow in this chain, is not clear. You have Anthropic and OpenAI building a lot of their own power and data centers, not enough but they're moving that direction. So I think that is why — I remember two years ago everyone was praying that AWS would grow like 18%.
33:58 And it's now growing 40s and stocks incrementally higher but not materially so given what we would have thought about a 40% growth two years ago. So I do think that there is a decent amount of queasiness on the ROI. I think there's also queasiness on the sustainability of growth because it's so concentrated in customers.
34:17 So you've seen it where Meta is kind of on the ropes with it. I think the market and I think a lot of people say the hyperscaler ROI is still positive and Andy Jassy laid out well and I think people are on board with that for now, but you're right that that can change quickly. I think sometimes we overestimate how sensitive these execs are to the stock price.
34:35 Again to the point of religious zeal, certainly they don't want their stock price going down, but it took a long time before in 2022 before Meta started making cuts and that sort of thing. So I think could they go down 10, 20% and everyone thinks it's a bear market for sure.
34:52 I don't know that that would cause them to flinch. And is the ROI so bad that the stock would go down more than that? I don't think so. I struggle to see the stock market forcing them to stop capex because in my view the ROIC is good enough that the stocks will never get so cheap that they change course.
35:09 But that is the question. >> I mean Zuck is a different animal when it comes to his ability to withstand the share price pain. But I do want to open up the can on Meta a little bit. There are so many bear arguments for Meta. There was at the point when they were saying like we're building out all of this compute capacity.
35:30 We're buying it up, but we're not going to sell it. It's for our own uses. And they were really being punished for that when they indicated that if they had overbought that they would sell it out. The market really rewarded them. And it has perplexed me and it might go to the short-termism that you — we're seeing in markets, but I'm like if Mark Zuckerberg can't figure out what to do with the compute that seems to me to be just like a bear case for AI technology in general.
35:58 What do you think about this bear argument for Meta that they haven't defined so much what all of this compute is going to be used for? >> I do think it would be a little shortsighted of them to rent out the excess compute. I think the reason they made this messaging is they want the market to give them rope and so they need to signal like look we understand your concerns and we will stop or release the capacity we have if need be.
36:24 Now if you do one of these three-month contracts you can earn a couple billion dollars in profits and that would be helpful to Meta but it's much more about the signaling to investors than people on a trillion dollar plus company getting excited about one year supply demand on a portion of your capacity.
36:40 So I don't think it's going to move the needle for the business. I think it's more almost telling investors, look, we get it and we will stop if need be. We will release our capacity if need be. I might disagree a little bit that Meta — they've had a lot of internal dynamics, hiring a new team. They were originally open source focused.
37:00 They've bounced around a lot. I think in general, there's always winners and losers within the foundational models. I think as long as the aggregate is making progress, that's more relevant to the broader AI trade. I think that's still the case. I think Meta has more of its issues.
37:17 I think Zuckerberg — there's a great quote in Sebastian Mallaby's book on Demis Hassabis about basically him asking Zuckerberg about AI as well as various other tech trends. And Zuckerberg was equally enthusiastic about AI and crypto and NFTs and all this stuff. And Zuckerberg far richer than I am, far more successful, but he has a tendency to go big on tech trends that it's not really obvious that he needs to do that for the core business.
37:43 I think as a shareholder, you have to believe in him for that. I think that's a more binary question as a shareholder. It's not one I can underwrite. And so but I also don't think that their troubles are necessarily reflective of broader AI troubles.
38:00 >> Well, to me, it's funny the people who are super bullish on AI, right? But then they're bearish on Meta because Meta doesn't have a plan. I'm like, well, there really isn't a plan at the corporate level for a lot of companies. And if you talk to the teams that are trying out AI, it's difficult. It's difficult to go to senior management and say, "Yeah, we want to bring on this technology that the bull case is we all get fired, right?" Like the actual adoption at the corporate level, how that's going to look is still
38:33 a huge concern. And so I just don't see how you could hold both views in your head at the same time that you're like, I'm super bullish on AI. It's going to change the way that we work. Compute is going to be the new oil, but at the same time, like Zuck is overbuilding or over — you know, like >> it's a good point.
38:52 I think that is Zuckerberg's view as well, which is basically, hey, we're believers in AI. We're building this compute. Don't worry about it because we can sublease it. We can do whatever we need to do. Again, to the point about acquiring optionality. It's not totally irrational for them to do.
39:08 I actually think it's actually quite rational. They're the biggest spender. They're relatively the biggest spender. They get the most questions. But I understand his concept of like this is existential to my business. It might be huge upside. It's not obvious yet. I need to keep playing the game and I need my investors to give me rope.
39:27 And so I need to tell those investors, look, I'm playing this game and I will back off if need to, but I'm not backing off yet. I think that's actually a rational strategy even though public markets will whip your stock around. >> Well, I want to shift a little bit from the analysis of the facts right now to what it means for you as an investor, as a portfolio manager.
39:48 So you have the ability to go long and short. What does the book look like right now? I know that you are very focused on the micro and the individual companies. We have to stay away from that a little bit, but in terms of the places where you're carrying a bit more exposure to these themes and the places where you're betting against them.
40:07 >> You want to be in the big themes. I think I have a little of a contrarian streak. There's a lot of headwinds about the politics around data centers. I think in general that is an overstated concern. I understand why there's concern. I understand why these people are upset about data centers, midterms are coming around.
40:23 I think there's a lot of stocks that are basically implying that data center starts are going to be challenged. Whether that's labor, whether that's some of these sites that are actually applying for power. I think that that is unlikely. I think that's probably short term.
40:37 I think that you look past the midterms, whether certain states are going to put moratoriums in place or not, I think they're going to get built. We've seen various technology moral panics over time and usually the market is strong enough. So I think that's one area where I'm looking at things that are pricing in a deceleration in data center starts.
40:55 I think there's actually some legs there. I'm always wary of extrapolation. I think anywhere that there's high near-term certainty I'm looking a little more contrarian and saying eventually you come on the back slope of those little mini cycles and extrapolations. So I think data center starts look good.
41:11 I think there's a lot of idiosyncratic growth stories that have been kind of left for dead. Things like e-commerce, some recent IPOs. Nobody's talking about those anymore. It's all — is software dead? Is it not? So I think that there are some independent stories, growth oriented, stand on their own legs that are pretty compelling because they used to be bid up to high multiples and I think they're not as attractive given there's other exciting areas.
41:34 >> Now, a bit about just your philosophy on the portfolio, right? So you said you want to have these themes. You're long short, right? Long short means a lot of different things. It can mean anything from like market neutral to there are people out there who are running essentially like 100% net long calling themselves long short and it's really more of an enhanced long portfolio.
41:56 What does it look like for you? What are your typical nets and how much can that swing around because I think it can help us understand how you feel about the environment. >> I refer to the fund as long biased which I count as having more than 50% net exposure, can obviously be higher than that as well. I think that there's so much money — one of the more common market debates these days is the influence of pod shops and momentum shops or whatever it is.
42:20 There's clearly a lot of money that is market neutral, that is relative valuation focused. I think that there's a lot of opportunity in not being constrained in that way. I like being long biased because markets go up over time. Capturing beta, you don't want to get paid for it, but you want to be able to capture it for your investors.
42:38 So I think that a long bias model with more duration that is not subject to having to be perfectly factor balanced is a pretty attractive model. I think it's probably always been the case, but when you have more and more money that's got to be quarter to quarter, that's got to be factor balanced or factor aware or has to delever at the wrong time, maybe it was July or whatever it may be.
43:01 I think that the kind of old school long-short model is probably in for a renaissance. I think there's reasons why it's tough to do that business. People don't want to pay for it. People want either a cheap long only or a high fee market neutral shop. And so I think that creates an opportunity where kind of the old school stock picking can actually outperform.
43:25 >> I alluded to it a little bit — you did very very well in 22. So many tech investors got so excited in 2021 and got carried out, rode their names down all the way. Some of them have very publicly come back from the ashes and legendary comebacks and hats off to them.
43:46 But many people didn't. For every Cliff Sawson who made it out and back, there was somebody else who got carried out probably times three. And so I'm just interested in like are you ever going fully net short and what does the environment have to look like for you to get that bearish on tech and what would you want to see from the current environment to really start to make you concerned? Because despite the uncertainty, you're still pretty bullish on this trend.
44:18 >> It's unlikely I would ever get net short. I think late 2021 was a once in 10 or 15 year environment where you had basically all the factors of incredible government support, significant retail involvement, clear cyclicality with COVID and pandemic beneficiaries in secular industries.
44:38 So you had a lot of reasons to be incrementally bearish. I think that multitude of factors rarely exists. I think markets move faster now. So it's very possible that we get one more regularly. Now I think the AI trade eventually has those ingredients. Huge change in underlying business activity to do this AI buildout. A lot of investor enthusiasm.
45:01 People are talking about curing cancer and money will be obsolete and whatever it may be and a lot of cyclicality in an investor base that's used to secular growth. So the ingredients are there. I think the things that are most important are number one basically the lab ARRs.
45:20 In other words, the adoption, the S-curve, which I don't think there's any crack in there yet. Number two is the capex. There's concerns it's not going to grow at the rates it has forever but if it levels off at pretty high rates a lot of these stocks will do well and right now it seems like growth is still strong.
45:39 Capital markets are still — they're tightening, but still provided hundreds of billions of dollars this year for infrastructure. So I'm not worried about the capital markets stopping it. And then the third thing that's relevant is the price of compute basically. And I think that will signal if we're in an overbuild, which would be problematic.
45:56 So I don't think we're there. I think the ARR is strong. I think the capex is strong. And the price of compute, the time it takes to build data centers is not suggesting we're at a glut yet, even though Elon might move that a lot. So those are the things I'm looking at. I don't see it yet, but I do think that if and when this cracks, it would be helpful to have a short selling skill because there's a lot of people in highly cyclical businesses trading peak on peak and even probably some stocks that I own. So there's going to be a
46:22 lot of pain, I think, if this ever rolls over. And I just don't think it's yet. >> The thing that has me most concerned is actually somewhat the AI tools that are available to a lot of investors and the echo chamber. And look, whether it's financial Twitter now X or it was the chat rooms back in the 90s where people were sharing stocks with each other, you now have like a true echo chamber where you're talking to an AI chatbot.
46:50 And there is so much happening at the technology level. And so people are doing a lot of fundamental work, a lot of work trying to understand the differences between these companies and where they sit in the supply chain. And it's really easy to have a strong almost feels like bulletproof fundamental thesis and you can take that and you can put it into an LLM and you can say sense check this and they can go, "Jeff, that is a really smart thesis," but sometimes the market doesn't really care about that. It feels good, the fundamental focus that people have
47:20 right now, but the market doesn't always care. >> I think you're right. The amount of AI generated content that I've seen, whether it's in sellside research or substacks or even just investor chatter is surprising to me because the whole job is you got to come up with your own thoughts.
47:36 And I think that it's very easy to get standard consensus thoughts, but I'm not really sure those are of value. And so I think AI is great for ramping up. It's great for facts and figures, but I agree with you that it can create an echo chamber. I think there's also a huge behavioral opportunity in fund management.
47:55 Just think about how addicted we all are to our phones. Think about how quickly these narratives spin up in politics or stocks or whatever it may be. Think about everyone's attention spans. So I do think that I don't think AI helps with any of that. I think AI probably enables a lot of shortcuts.
48:14 It enables a lot more confidence. I think that it's going to create a lot more behavioral inefficiency. I heard on another podcast that somebody was saying I think that if AI gets advanced, it's basically going to remove fear and greed from markets because the AIs don't suffer from fear and greed.
48:28 They can be perfectly rational. And I basically completely disagree. I think that you're going to get faster narrative cycles. I think you're going to get more crowded narrative cycles. I think that's in addition to the phones and terminally online nature of our society. So I think that there's an emerging behavioral inefficiency.
48:46 I think markets probably getting less efficient over time and I don't think AI really helps with that. Actually, I think it might for quantitative models, but I think it's much less likely to lead to inefficiency and more likely to lead to overconfidence. >> Yeah, you're saying to efficiency, you said not to lead to inefficiency.
49:01 More likely to lead to inefficiency than efficiency. >> Yeah, that — yes, thanks for catching that. >> I just wanted to clarify. But to your point about the speed — it's something that right, like the factor you talk about being factor aware and you're like there's an advantage for fund managers, you don't have to be overly factor aware but it does seem like it is a requirement to at least know what your factors are.
49:22 That doesn't mean — you can believe it's an advantage for you as a portfolio manager to knowingly take on more factor exposure than somebody whose job it is is to minimize that factor exposure as much as possible. But it doesn't mean you can be unaware of the factor exposures that you have.
49:40 And I think the speed at which things are being bucketed, factorized, and ceasing to be idiosyncratic company level alpha and becoming a bucket or a theme or a trend is happening extremely fast. How do you as somebody who likes to underwrite but do bottom up work sense check yourself and make sure that you're not letting the strong fundamentals that you're seeing cloud the fact that it's part of a theme that is extremely hot right now and that means a lot of the money that's coming in is not doing the work that you're
50:14 doing. >> I think you certainly have to be aware. I've spent a lot more time in the last year or two trying to be aware of factor exposures. It's not even something I was paying attention to probably four years ago. So I don't know if that's just my awareness or the markets are changing.
50:30 I think it's ultimately a time horizon question and I think it's an LP management question. There's certainly a standard in the business that you report monthly results and monthly results is completely insignificant in the scale of time.
50:45 Quarters are probably insignificant and so it's just a noise generator especially given how much factors move. If you were a software investor you had LPs saying what's wrong with your process, why are we losing so much money, and the reality was it might have been that AI funds were levering up and shorting software and driving negative momentum into your stock and it was a two-month thing and those stocks have round-tripped over four months. So I think you have to be factor aware. I
51:13 think you got to manage your LPs. I think you need to make sure that if you're making factor bets or you have sector bets or whatever that you're very well aware of that and that you're conscious about that. But I also think it creates a longer-term opportunity where you can say look clearly momentum is having whatever it is one of its worst five day stretches in 20 years.
51:30 Okay, let's take a look at that. One of the things I like to go to is extremes. When you have a washout in a sector, you have people saying something's uninvestable. You have, hey, this is the largest hedge fund shorting day that we've seen in seven years.
51:46 Okay, pay attention to that. Those are the types of signals that I think give a lot more opportunity. It seems like a new world. I'm sure there's always been parts of that. I certainly wasn't paying attention to it as much as I am now. So >> and so what did you do? Obviously, we all had to go through that, right? In July.
52:02 I doubt there's anybody listening to this podcast who didn't have at least one AI stock in their portfolio that was going down every day despite really no news about that particular company. You as a professional investor when that's happening to you, what are you doing to sense check your book? Because I don't think July is going to be the last time in this cycle that something like that is happening and we're all left wondering like why is this happening? >> Right? Yeah. There's kind
52:34 of a life cycle of at first you think, oh, this is great. My stock's down. I can buy more. And then there's a second where you start to get a little worried. And then there's a third where you say, okay, there's maybe bigger factors at play. And that's kind of when I think people started hearing about the margin calls and situational awareness and it made a little more sense.
52:48 An example there is a lot of their pain was concentrated in Asia, especially Korea. You saw some notable percentage of the Korean population getting margin calls. Who knows if that's rumor, but you saw a lot of margin calls, a lot of leverage in Korea, huge deleveraging. That's something that perks my ears up.
53:09 You have huge hedge fund deleveraging, huge retail deleveraging in a secular trend that I actually think is still going pretty well for the basically two large companies in that country. So that's one where once you have all the pieces in place and you can see what's going on, you say, "Look, I want to be a buyer when people are getting margin called and liquidated.
53:27 " So that's one where maybe I think the picture is improved. But otherwise, it can be hard to suss out whether it's fundamentals or technicals. So I think you have to really see a lot of signals like I think we saw in Korea to say, "Okay, this is maybe just even from a short-term trading perspective, a much more attractive proposition.
53:45 " >> And is that the type of thing where you are keeping some level of cash on the sidelines for these events where you're willing to flex up your leverage a little bit when you have that confidence as a portfolio manager? How do you like to maintain the ability to step in when you have that level of confidence? >> Yeah, I think I've talked about optionality before, but I think optionality is generally one of the most important things when managing a portfolio and that can be cash, that can be lower gross exposure, but being able
54:14 to move such that you're not the one deleveraging or maxed out when there's an opportunity. I think that's especially more so in these markets where it seems like there's again these three standard deviation events or once in a 20-year event seem to be happening every several months.
54:33 I like to run with a little bit more flexibility. I like to be able to step in once there's a little more pain. I'd say in general, my time horizon is a little longer. I'm trying not to trade too much. I'm trying to wait until something hits me in the face. I think we had a little bit of that in July with as I mentioned, but it's hard to say.
54:48 And I think that there's a graveyard of folks that try to trade short-term moves. So that's why I'm really trying to see something that's three standard deviations, four standard deviations really stands out to me. And then you can move and keeping that optionality open — this job, we all spend all of our days looking at models and companies and reading and all this stuff.
55:10 But you make your money on one or two good ideas a year, a couple big ideas every couple years. And so making sure you have the optionality for that, I think is more paramount than making sure you've got great stocks maxed out at all times. >> Yeah. And look, obviously the vol has come down quite a bit, but I just pulled up like EWY and socks vol.
55:30 They're realizing 73 vol in EWY on the one month look back and socks is 55, and they peaked out at like 185. It's like how much leverage do you need for sectors that are realizing that level of volatility? >> Right. Right. Yeah. I'd say in general I've learned that I comparatively I run with very low leverage.
55:56 I didn't realize some of these folks are running with as much. I agree with you because I think leverage is a great way to reduce optionality. I think it's interesting the realized — usually one of the most basic indicators you can see for how markets are feeling is just the VIX and you can tell how much people are panicking and paying for protection.
56:15 This market has been different because there's such dispersion within sectors and even within stocks that the markets are very tranquil. The headline indices are very tranquil and you're actually seeing these incredible vol spikes in certain sectors or to your point EWY or socks, which it's like a duck kicking below the surface but if you go below the surface this year's had a lot of volatility and a lot of rotations and a lot of opportunity to be nimble at extremes.
56:43 My guess is frankly it seems like one of those trends that's going to increase. It seems like there's probably only shorter time horizons, more uncertainty, more money flowing into less fundamental strategies. So my guess is that that will increase and the behavioral advantages will increase. >> Well, look, I want to talk about what's coming up next and close out.
57:05 We've got pretty much with the exception of Nvidia at the time that we record this, they haven't reported. I'm not going to ask you for any predictions on Nvidia earnings or anything like that, but we're kind of through the meat of earnings season by and large. We have a little bit of Fed uncertainty, geopolitical uncertainty that people are concerned about, obviously what's happening in the bond market, but what are the factors that you're going to be paying most attention to as we close out 2026? >> Yeah, I think we'll get the
57:31 Anthropic S-1 probably, it sounds like next week or the week after. I think that's going to be — I don't know that we'll get a ton of new information that we don't really already know. We might get a little more precision, but my guess is the numbers are broadly leaked, but I think that those companies coming public will be hugely helpful in that we just have metrics and we're not guessing and trying to hear third-hand.
57:53 But I think the most important things are, especially for the AI trade, it's ARR of the labs and open source, some form of adoption of ARR. Number two, it's capex, which seems pretty locked to be strong for 27 and people start looking at 28. But I think if anything, people are feeling better about the hyperscalers than they were in the first half.
58:12 So that seems to be strong. And then number three is probably the forward prices of compute, which very tight right now, probably tight for a while. I think if you're going to see an overbuild in data centers, you'll see it there. Those three I think are broadly strong right now.
58:28 What will the stocks do? It would not surprise me if we chop around for a while. I think you had a lot of retail enthusiasm, a lot of investor enthusiasm in June. Usually, you don't see those things rip back to new highs. I think that we're waiting for the next breakthrough or acceleration.
58:43 It wouldn't surprise me if we get one. I don't know what it is, but there's recursive self-improvement is kind of floating out there. So we could have some technology breakthroughs that get people excited again, but it would not surprise me if we chop for a bit. But I think having the labs public will help with some of this narrative volatility. So I'm excited for that.
58:58 >> One follow-up question to that before I let you go. Do you think that there will be any sort of like capital vacuum that comes from that? We saw it with space, right? Again, not quite the same fundamental business quality that we're talking about in many of these AI themes that people are investing in where there's already like strong earnings, but you had all these also-rans in space really have the wind taken out of their sails.
59:22 They had been doing incredibly well and the second people could buy SpaceX, they sold a lot of those other space names to go put it into SpaceX. We don't have a pure play lab, right? Like Anthropic. Obviously, if you're buying Google, you're getting the whole Google business. If you're buying these semiconductor companies, yes, it's directly tied to the AI theme, but you don't have really AI model exposure available on the public markets in a pure play.
59:50 Do you have any concerns about certain pockets of the AI trade having a SpaceX effect? >> I think it's possible. I don't have a ton of concern. I think that equity issuance is a huge signal for euphoria and blowoff tops. I think it is different when the equity issuance is into more real businesses.
1:00:11 Now, I would certainly quibble with the SpaceX valuation, but in general, these are not dollars going into — I'm trying to think of some of the highlights from SPACs or whatever it is. So I think that equity issuance into realistic businesses at realistic prices is less capital sucking than we need to put $10 billion into Nikola or whatever we were doing back in 2021.
1:00:33 So I think there's some worry. The space stocks, they doubled and then got cut in half. So they're kind of back to where they were. Could the AI stocks run up into the Anthropic IPO and then come back down for sure, but I think that the euphoria and the optimism and the positioning has kind of cleaned out in the semi trade.
1:00:50 I don't know that means it's going to rip, but I think that the market's ready to digest Anthropic, and I think it'll want lab exposure. Frankly, I think it's going to trade at a crazy price. I'm not saying I'll buy it, but I think that there's going to be a lot of enthusiasm for it.
1:01:06 >> All right. Well, Jeff, we will leave it right there. Where can people find you out on the internet these days? >> I'm on Twitter, Jeff Keller1. And then always on email, JeffKeller.co. I didn't pay for the com, so it's just the .co. >> Nobody's paying for the com anymore. All right, Jeff. Thank you so much.
1:01:24 We'll do it again soon. >> Great. Thanks, Max. Appreciate it.