AI Spend To Kingdon (Capital) Come with Parshv Shah
47:21 Share

When will M&A activity pick up? Will this year mark the return of IPOs? Listen to Strategic Alternatives, a podcast from RBC Capital Markets to get insights on these questions and more. Explore the trends in market forces impacting deal flow and find out how companies' investors are shifting their strategies to drive growth and unlock value. Listen and subscribe to Strategic Alternatives today, available wherever you get your podcasts.

To be invested with Betterment is to believe in better. Whether it's saving for today or building wealth for tomorrow, they help people and small businesses put their money to work. They automate to make savings simpler. They optimize to make investing smarter. They build innovative technology backed by financial experts for anyone who's ever said, I think I can do better. So be invested in yourself, be invested in your business, be invested in better with Betterment. Get started at Betterment.com.

Welcome to the Risk Reversal Podcast. I'm Dan Nathan, joined by Parsh Shah, Portfolio Manager at Kingdon Capital Management. Parsh, welcome to the pod. Thank you for having me. All right. This is a good opportunity for us to kind of level set where...

investors kind of perceive this generative AI trade. I'm going to call it trade in fast money terms here a little bit because it's one of those things that over the last two years in my career, going back nearly 30 years, I've seen very few sort of shifts that have kind of just taken the markets by storm, if you will. And so I'd love to go back a little bit. When you started investing in the markets, how long have you been at Kingdon? Kingdon's a firm and I just got to tell our listeners if they're not

familiar with it. It's over 40 years old. Mark Kingdon, the founder, he's a legendary investor. He was a legendary investor by the time I got in the business, um, in the late nineties. So tell me a little bit about your seat at Kingdon capital. Uh, absolutely. Thanks again, Dan, for having me. Uh, I've been investing for over 19 years and I've been with Kingdon capital for over 12. Uh,

I joined as a portfolio manager covering primarily software and services and expanded to broader tech and TMT overall. So my seat encompasses everything that we're going to talk about today from generative AI to software and security to

consumer AI, as Dan had mentioned. So quite a broad landscape. Yeah, and it's interesting when you think about, we've spent so much time on this kind of ecosystems that relates to the hardware and then the use cases, whether it be on the public cloud, the hyperscalers and the like, and who are the customers and how are they using this technology and the like.

but it really does feel like it's about ready to broaden out a little bit from, you know, to software and services. So we'll hit all that as it relates to security too. There was just on the street this week, there was a couple upgrades of some security stocks, but I think one was Bank of America. They did a double upgrade.

upgrade on Cloudflare, which I always love the double upgrade. But I think some of these folks are like, wait, I got to play a little catch up here, right? I got to think about what the balance of 2025 looks like into 2026. And it might not be NVIDIA. It might not be the hyperscalers. So

excited to talk um a little bit um about that all right you and i met in october we were down in dallas our mutual friend imran khan was holding the conference with a k um down there it was really a great group of speakers evan spiegel linda yaccarino and there was a whole bunch of others it was like really interesting i mean i actually wasn't dominated by a there was a great panel i think it was somebody from gcp and there was somebody from openai and that was pretty interesting so

When you think about where we were in October and there was a level of enthusiasm that still existed, now we flash forward. At one point this month, we had the NASDAQ down 13%. We had the S&P down 30%. It looks like that we've taken a little bit of a stall. And I'm just curious from a market sentiment standpoint, what do you think has happened over the last six months or so?

No, great question. Quite a bit has happened. Obviously, we had the elections come and go, needless to say. But when you move beyond that, I think a major debate entered the landscape of AI investing, and that was with what we learned about DeepSeek during January. I think over the last couple of years, the assumption was that

The more capital and the more capital intensive you are, the better the models that you will be able to build and thus your moat will be strengthened. And so that question got thrown for a debate because when DeepSeat came out, all of us were left to ask the question, well,

Can you build more efficient models? Do you need to be this capital intensive? And can this broaden out whereby now the hyperscalers will have significantly more competition, perhaps Nvidia's revenue growth won't be as strong as we previously anticipated.

And so that's a major debate that entered the investing landscape in January and still continues, quite frankly. Yeah, so that date was about two months to the date when we're recording this right now. It was January 27th. It was a Monday. The data started coming out over the weekend. I was getting blown up by text by a lot of folks who maybe are not like tech experts

specific in their investing, but this was something that I think the markets were not set up for in a way because there had been so much enthusiasm about a handful of names. When you think about going back to the launch of ChatGPT back in late 2022, there's been like 10 names in the public markets that have really seen massive appreciation since then. But some of the folks

that initially thought, let's shoot first, ask questions later. NVIDIA closed down 17% that Monday. A lot of folks have come around saying, well, actually DeepSeek and just the ability to kind of train that model and the efficiency in which they did is a positive thing for NVIDIA. Where were you initially and where are you now two months later? Well, initially, I did not view it as a positive because the narrative started to shift and the narrative became

As I mentioned earlier, the more and more CapEx dollars that are spent, the more GPUs you buy, the better it is for NVIDIA. And then all of a sudden, you seem to have a thesis drift saying, well, actually, the less that you're spending, the more efficient you are, that'll lead to higher generative AI usage and thus still great for NVIDIA. Perhaps that's true, but the market has to go through a digestion period, which I still think we're going through right now, to go from

massive capital intensity to less capital intensity. In terms of, is this ultimately a positive?

Well, there's a lot of nuance in that, but I think the one thing that I kind of keep thinking about and pointing to is that last week at GTC, Jensen Wang made it clear that relative to last year, the demand for inference and advanced reasoning is 100 times. And so that's a pretty remarkable comment. And it's not just a comment, but it's being backed up by the spend. So while on the one hand, we had a debate shift,

on the other you are seeing this shift towards inference time reasoning and as a result you are going to see more compute enter the system but again we have to digest some of these new data points that we've seen over the last few months and that's jensen being jensen this is gtc it's their massive user conference right and it feels like he spoke for like three days straight and you know part of it is is like yeah they have this headwind of like perception at least from these you know kind of um

you know, more efficient models. And it wasn't just deep seek, right? It was Alibaba came out with one, Baidu came out with one and financial this week just announced one. And I think this headline is kind of interesting. This is from Joe Tsai. He's warning of an AI data center bubble. Now, Joe Tsai is the chairman of Alibaba. I think we,

both agree that he's a pretty brilliant guy. Absolutely. He and Jack Ma, what they did with Alibaba and Ant and the like here. But it seems like there's something brewing here. It's like a little sovereign AI in a way. And we also have to go back a couple months ago where President Xi got all those tech champions together and it looked like they were making a united front. This is years after he squashed all of their technology companies. And now you see what's going on with DeepSeek and these other models. And so

I think about it as a way where we're putting through all these export bans of our best technology, right? These guys are trying to say, we don't need your best technology. We're actually innovating on our own side with different sort of chips, which I think is kind of interesting. How do you walk away from that sort of debate? Because this is something that is going to go on for a bit. It's going to go on. Absolutely. And

Yeah, we actually saw Joe speak at Imran's conference, and it was a great conversation that Imran had with him on stage.

And one of the things is that Alibaba is clearly not backing away from AI. And in fact, if anything, their own capital intensity is going to accelerate over the next three years. But it's not to the tune of hundreds of billions the way that Microsoft and Google- Yeah, but Dan, they're catering to the China market. Our hyperscalers are catering to the rest of the world. So it's a much bigger TAM and a much bigger pie that the companies here are going after. So yes, fair.

It's not the same magnitude, but it is still multiples higher that Alibaba is going to be spending. Yeah, no doubt. And so I guess the point of Josiah in this – he was speaking at HSBC conference in Hong Kong the other day. So he's talking about a bubble. He's talking about what the U.S. are doing. That Stargate was –

It was Masa-san from SoftBank and it was Oracle. Right. And it seems like there's been no shortage of announcements like that. But the one thing I think is really interesting is that we are seeing a slowdown in the rate of increases. Right. So on a period over period basis, we're getting down to kind of low single digits as far as like Microsoft and Google and the spend there.

Does that sound right? Year over year, it's still a lot. But we're starting to see, and especially after Q4 earnings where we saw slight revenue growth decel, and we saw the numbers which are staying put. Amazon had a huge upgrade to their cap back. So I'm curious how you're thinking about what Joe Tsai is saying, what these hyperscalers here in the US are saying, what Jensen Wang is saying. Do you think we're going to be at overcapacity sometimes? Yeah, that's a great question. It's something we talk about.

internally all the time. And the answer is yes to all of them. What I mean by that is, is there an oversupply risk? Absolutely. I mean, you think about all of the spending plans that have been announced this year, to your point, Dan, well over $300 billion just by the big hyperscalers this year. Is that all going to be put to productive use? We still don't know. And many of them have signaled that we will continue to see increases after this year.

This reminds a lot of seasoned investors of what happened at the turn of the century with the dot-com bubble. I think that said, where I'm a little more balanced is that if we take a step back and think about why are these dollars being spent? Well, the promise of AI is substantial. I mean, and I think

Have we realized that promise yet? No. I mean, a lot of what we've been doing is kind of chat GPT and chat-like type applications. They're kind of consumer-facing at the moment. Very consumer-facing, absolutely. And some companies like the Metas of the world have talked about how it's helped to accelerate their own internal revenue growth. But promises that have been out there in terms of healthcare applications. This morning, Charles River, one of my colleagues mentioned to me,

talked about incorporating AI into their new lupus drug. You have security use cases, which we can talk about some more that are just barely at the start that have, you know, we're at the top of the first inning. So the point that I'm making here is that the benefits of AI have not accrued to the companies or to society

significant dollars have been spent and we do have a risk of oversupply, no doubt about it. But at the moment, we are still in the mode of, we're discovering as a society and as an investment community, what are the benefits that are going to accrue above the AI infrastructure layer? Yeah, the Meta is a great example because it's one of the few tech companies that they're able to monetize their own investments in it right now. And you could say, well, Microsoft, they launched this co-pilot using open AI technology. We haven't heard a whole heck of a lot of

about their customers using that technology yet. I think it's pretty universal that a lot of folks feel like, you know, health care and, you know, drug research and the like, or whether it's like, you know, intelligent assistance for doctors and stuff like that's going to be really game changing. But again, it's probably

a couple years away. I want to go back to that Jensen 100X sort of like need for compute. You know, it seems to me that when it's useful, when we hear these conversations about Microsoft doing this deal with a nuclear company to kind of put, you know, a reactor back on, you know,

you know, on the grid and stuff like that. It seems like that's a great headline there. But then the flip side of that, when you hear Jensen talking about 100x compute, where's all that energy going to come from? And I'm just curious, do you guys spend any time thinking about that? Because that seems like one of the things that could be totally a fly in the ointment of the sentiment towards this trade. You know, Satya Nadella, he was on Bill Gurley and Brad Gerstner's podcast a few months ago. And do you remember they asked him specifically if he's still chip constrained? He said, no, we are power constrained. And I just feel like

That could be something that takes a little air out of the excitement in this trade. Well, power was a big theme last week at GTC, and it took more airtime probably than the GPUs themselves. So absolutely, power is a major constraint. And I think one of the promises going back to NVIDIA and now this upgrading product cycle with Blackwell

is that it is significantly more power efficient. It is a transition fully to liquid cooling, which is more space and power efficient than air cooling, and then air cooled data centers, excuse me. Also, NVIDIA talked about it with their optical cables moving from more of a copper-based optical network to photonics and what's called CPO, co-package optics. The point is that that is a key focus of the Blackwell product cycle and cycles moving forward. So,

Satya Nadella and the team at Nvidia and others talking about power efficiency as a major theme, something that we've been looking at as well. And quite frankly, future iterations do need to be much more efficient. Otherwise, this won't be affordable. Yeah.

What do you make of – and there was a great article in the Information the other day talking about kind of covering GTC and talking about incremental performance from Hopper and the like. But they actually were quoting a few folks saying, listen, I haven't even gotten my Blackwell chips yet, and they're already talking about Rubin. You know what I mean? And so what does that mean to you? Because –

In some ways, a lot of folks are going to kind of think about, OK, this upgrade cycle, you know what I mean? And then I have to do it again in a year. It just seems like that seems like there's going to be a glut. There's going to be bottlenecks. I mean, yeah, I'll be completely honest. I'm worried about that as well. And one of the things I get concerned about with that is when I look at the hyperscalers, what I mean by the hyperscalers, meaning the meta Google Amazons of the world,

is that they spent the last few years actually elongating their depreciation cycles and the useful life of their servers from call it four years to six years.

Now, all of a sudden, these chips that they're investing tens of billions of dollars will be made redundant over a year or two. So the depreciation is going to have to be sped up after being elongated. Well, that is a concern. That is a potential. And Amazon actually did a slight adjustment on their last earnings call where they slightly, very slightly, I think it was by half a year or a year, decreased the useful life.

That's not a cash flow impact, but it is an earnings impact. And so, you know, there is some adjustment that has to be had on that front. And in addition, does it create potential for a glut? It absolutely does. Uh, but the checks that we do and, and, and, you know, you look at the orders, um,

They don't suggest that if anything, you still have a line outside the door for Blackwell. So I think that is a key concern. It's almost being in an annual product cadence, kind of like Apple with the iPhone. Yet this is an enterprise product, not a consumer product. So it's an interesting dynamic. Yeah, but Apple, Tim Cook doesn't come out at WWDC and just say, here's iOS 19 and this is going to be great for this generation.

new piece of hardware that we're going to introduce in September and say, by the way, iOS 20 is going to be off the charts and it's going to have a new form factor as it relates to the iPhone. It seems like Jensen, in a lot of ways, when you've seen your stock go like $300 billion to $3 billion,

and a half trillion dollars in market cap you got to keep a narrative going you know what i mean now he's clearly like an absolutely amazing like you know visionary ceo that we haven't seen the likes of maybe since like steve jobs in many ways but um i just wonder like when i think about satya at microsoft sundar at google um

jazzy at Amazon, like these guys, you know, they're asking for forgiveness, not permission, but this is a really new sort of situation for these companies, right? Because they were just messing around with like ad models or e-commerce model, you know what I mean? Or, you know, enterprise spending on seats and all this sort of stuff. Now this is a totally different kind of ball of wax in a way. And I'm just curious when you think about Google and you think about Amazon into their prints, um,

Last month, they were trading at all-time highs. And the stocks actually got slammed. I mean, literally, they were down, let's say, including Microsoft, I think between 4% and 8%, all three of them. And then they kept on going. Now, you can tell me one way or another, well, did the market start selling off in mid-February because we lost confidence in these companies' ability to grow and the like here, and valuations were high?

And competition was getting more severe. And this is weeks after deep seek, right? So there was a whole list of reasons why these companies should continue to maybe like, I don't know, take a break, take a little bit of a breather and then the stock should have followed. So I'm curious how you think of the 20% drawdown in all three of those and throw meta in there too. Like how, is that just taking a lot of the euphoria out of the, you know, out of this trade in general? And is that a healthy thing in Europe?

Yeah. I mean, like you said, the stocks, these MAC 7 names in particular, went into earnings season at all-time highs. And each of them had a slightly unique situation, but there were some common themes. So if you look at especially Google, Amazon, and Microsoft, all of them in their cloud business saw a slight, Google a little bit more, of a deceleration in growth rates while they announced on their calls that they'd be accelerating, or at least

seeing bigger absolute dollars of CapEx in the market anticipated. So a lot of investors are sitting at their desks asking themselves, wait a second, here are these names at all-time highs. I already own probably a max position. And they're decelerating their growth and increasing their CapEx. So I have to take my free cash flow estimates down

It wasn't a recipe for the stocks to continue to work. On top of that, you had a broader market backdrop with tariffs and some geopolitical dynamics, which are obviously still present. So you take all of that together and it created a pretty...

messy setup, I guess you could call it, going into those prints. Isn't it crazy though? We're almost in April. We're going to be getting to Q1 earnings really quickly. Yeah, do it all over again. One of the things that I thought interesting going back in early January, all the Mag7 basically reported, I think that last week in January, early February, and they had all kind of given some soft guidance on CapEx. It kind of snuck out in the week or two weeks before that. And so I wonder if they're going to do that again because the idea of kind of

Let's just say, I don't expect an increase to 2025, you know what I mean, CapEx. If anything, and I know like a lot of,

people who are much smarter than me about this stuff. They're like, there's no way they're going to pull back on those CapEx numbers. But to your point that you just made about if there's a greater decel and say in revenue growth, then they're going to have to, to some degree. Yeah. I mean, no, it's a great point. The one company I think to watch really closely on that front is Microsoft because Microsoft will be on its fiscal third quarter

earnings call. And typically, they don't update for the out year CapEx dollars on that Q3 call, but they have provided, Amy Hood has the CFO, broad strokes on what the next fiscal year could look like in terms of revenue and operating income growth. Could she provide an update on the CapEx? I'm not banking on it. Most, if not nearly all investors are not banking on it, but it certainly is possible because if you take that in conjunction with

channel checks have been coming out to suggest that they've been canceling some long-term data center builds and the evolution of their open AI relationship. It's certainly possible. I'll certainly be watching closely for that. As for the others, it seems unlikely, but you never know. But Microsoft will be the one to watch.

In today's hyper-fast markets, it's never been more important to consider every option to raise capital, drive growth, and create value. Stay one step ahead with Strategic Alternatives, a podcast from RBC Capital Markets. This season, RBC's experts examine how corporates and investors are adapting their strategies, reassessing their portfolios, and reallocating capital to navigate uncertainty and volatility in the current macro environment. Tune in to Strategic Alternatives, the RBC Capital Markets podcast today.

Imagining a better future is the first step. Investing that future with Betterment Advisor Solutions is the next. Whether you're launching your own practice, looking to streamline client onboarding, or

or just searching for efficient ways to scale your firm, Betterment is here to help. They automate to make tax optimization simpler. They provide support to make administrative tasks easier. At Betterment, they're building innovative technology all for anyone who's ever said, I think I can do better. So grow your RIA your way with Betterment Advisor Solutions. Learn more at betterment.com slash advisors. Investing involves risk, performance not guaranteed.

Let's talk about Alphabet Google really quickly here, because this is one where I don't know why there was that enthusiasm heading into the print. It was trading at all-time highs. But the stock is down. And its lows last week or so was down about 24% from those recent highs. It's rallied a little bit here. But this is one going back to, again, early 23 with the Bard launch. It just seemed like a train wreck, one after another after another. But when you think about Meta, which you just mentioned, they're

ability to pivot a lot of that spend from the metaverse and kind of reinvest it or repurpose it, right? And they're getting those benefits within their own ad delivery and the processes within what they do. And then the valuation was something that people felt pretty good about. You could make a lot of the same arguments about Google valuation easily, the platforms in which they have to put this Gemini technology across, whether it's Gmail, whether it's YouTube, I mean, the list goes on and on. Why do you

Why do you think Google is in such a penalty box on a relative basis over the last couple of years? Yeah, I think that ... No, it's a great point. Valuation, to your point, Dan, is extremely undemanding. It's probably ... Last I saw, maybe a 20% plus discount to the market.

The issue is the classic innovator's dilemma question that scares a lot of big investors, which is essentially that, is my cash cow, and it's not just a cash cow, it's a growing cash cow, one of the best business models ever, search, going to be disrupted and massively so. And going back to the conference where we met, really a student investor asked a panelist from Google, how do you

How do you basically grow through a period of what could be one of the biggest innovators' dilemmas we've ever seen? And it's one of those unprovable dynamics in terms of Google will get through this and they'll be okay. It's difficult to prove, and so investors just are not yet giving them the credit they'll be able to do it. Now, I think it's important to remember that in 2024, Google search revenue accelerated in the second year of this massive AI wave.

Last quarter, they grew over 12.5% on a nearly $200 billion business. So what gives? Because I know you and I have talked about this. I've talked about this with a lot of others. Everyone says, well, my search activity is so much lower. I'm not searching anywhere near as much as I used to. I'm using Perplexit. I'm using ChatGPT. I'm using Grok. I'm using Claude and others. And I don't know if the search TAM is expanding and search usage is just increasing that much.

But Google came out with a data point that was a little under the radar a few weeks ago that they've hit 5 trillion searches annualized. And doing some back of the envelope math and connecting the dots, that could be up 20% year on year. And so over the last two years, commercial searches clearly have not slowed. Search activity continues to be up and hasn't seen signs of waning, it appears.

So we'll see. But at the moment, it's just in a bit of a pickle because, again, investors don't want to be caught in an innovator's dilemma, whereas Meta doesn't have that issue right now. Yeah. I mean, and the other point is like search is only useful if you're clicking on

the links and you're getting the ad impressions and like, you know, and some folks in like, you know, or can get spooked in a like recessionary environment that you're going to see a pullback, right. In advertising spend, you know, so that's one of the things that I think some of us forget how cyclical advertising and is in general. And then you think about how messy those search results are on Google. You get the overlays, you get the,

the sponsored links, you get the blue links, that sort of thing. And if the overlay is you don't trust them the way you would chat GPT or cloud or perplexity, which is obviously an aggregator of a lot of those models, then you're going to go to one of these others. I mean, my behavior has changed. The perplexity user interface looks like Google did. You know what I mean? It's really simple. But the answers in which you get are just...

I think they're just far better than anything you do on Google. We talked about the Mag7. I call it the Faithful Eight. Once Broadcom kind of entered that trillion-dollar market cap, now it's since come back again. What did you make over the last couple of years as a portfolio manager of this concentration among these, let's call them these top 10 stocks that were

half the weight of the NDX and NASDAQ 100. They were like 33% or so, the S&P 500. Is this something that you've gotten used to over the course of your career? Or is it just that the numbers are so big? We had eight stocks with trillion dollar market caps, some with two, some with three, that sort of thing. How did you guys shake out on the concentration? Well, I mean, I think the, you know, you hear about market concentration and you, you know, you talked about, you

the history of Kingdom Capital and Mark, our founder, often talks about previous episodes where you saw this concentration, whether it be the dot-com bubble or even back in the 60s with the nifty 50. And you think, oh man, that's troubling. That's bad. I think

There are some differences in nuance, though, again, because you have a situation where these companies are just compounding and have been, at least especially over the last couple of years, at such high rates. These, let's call it MAG-7, Faithful 8, whatever, however you want to look at it, Faithful 8, have been growing at a faster earnings clip than the rest of the 492, let alone the small mid-caps in the Russell 2000.

And so as fund managers, we're always looking for the new story, the kind of diamond in the rough, something interesting that's not these top seven or eight companies. But at the end of the day, you have to perform. And if they have presented opportunities, granted recently it's been a bit of a shakeout, and so the concentration is there, but

Each company is different. They have massive TAMs, and some are better positioned than others. And I don't think it necessarily spells a major trouble just because we have this concentration issue, but it is worth watching, and we are. Yeah, I guess the point I would make is that, okay, so they led over the last couple of years a raging bull market. Let's call them 10 names. They were a big part of the earnings up

We're earnings revisions for the S&P 500 that continue. I mean, a lot of folks thought, you know, coming into 23, we're going to have a recession. We had an earnings recession in 2022, the recession possibly going to 2024. We've never had one. So if we were to see, let's say these companies where we've already seen revenue growth rates kind of tweaked down by the street. But if we see pressure to earnings, you know, and by the way, the weak dollar should help that, right? Like, you know, kind of buoy that a little bit.

That's where the concentration becomes more interesting to me of what the potential downside is if we do kind of hit growth kind of issues one way or another. Let's talk about custom silicon. This is something you and I have talked about. When Broadcom introduced that guidance in, I think it was

mid-December. Yeah, the stock went up in two days 40%, okay? And it got over that trillion dollar mark. We'd just seen Marvell do a very similar sort of thing on a much smaller scale as far as market cap terms, but it had this huge move. And the narrative was that all those major customers of NVIDIA, once they've gotten through training the models as they go more into this inference phase, they want more custom chips, right? To do these processes. That was the excitement about that. Hocktan at Broadcom, he gave this crazy

crazy TAM, right, for custom silicon going out to the end of 2027. You could have driven a truck through it. I think it was like 30 billion to 90 billion or something like that. Yeah, 60 to 90. All right, 60 to 90. Okay. So midpoint obviously being 75. And then the question is how much were they going to capture of that? It wasn't just them. So like,

But Broadcom has kind of filled in that entire gap. It went to basically 250 from 180, and here we are right now at 189. And you would have thought that even the earnings that they were able to put up in the guidance just a few weeks ago, that would have been enough to kind of move the stock a little bit higher. It's gone nowhere. Well, I think with Broadcom specifically, it is – to your point that you just made, I mean –

These are trillion-ish market cap companies. Hard to decouple, even with great results, if the rest of the group is trading weak, isn't trading well. I think that's one thing. And I think the other is that in December, the...

the promise that was bought now has to meet reality. And Broadcom has a great track record of doing so, but now the proof has to be in the pudding. And that's what's really on the come here. And so you had the euphoria, excuse me, the enthusiasm, the excitement. Google is one of their big custom silicon customers. They will be spending with them. They are ahead of the game on that with TPU. And now we just have to see that come to light. And

To the extent that these names are somewhat, at least on the semi side, somewhat tethered and anchored to NVIDIA, well, it trades at a three or four point premium to NVIDIA. So again, hard for it to run away when NVIDIA has been kind of stuck here. All right, let's go down the ecosystem a little bit. We spent plenty of time talking about hyperscalers. Let's go to the...

servers, and then let's go to the components other than NVIDIA GPUs a little bit. Because what we've seen, first of the hyperscalers, we saw a revenue decel. We just covered all that, right? And then if you think about what we've heard from the servers, if you just think about Dell and Supermicro, let's spend a little time on there. Maybe we don't have to waste a lot of time with Supermicro. There's a lot of hair on that story one way or another. We talk about this stuff on Fast Money all the time.

You know, there's like a gazillion other stocks out there. You don't have to get involved in the ones that have investigations going on and the like here. So let's talk a little bit about Dell because it feels like it's kind of a one-horse race in AI servers with them. Well, you mean a one-horse race? From a competition standpoint. Gotcha. Well, in AI servers, I mean, they do have Supermicro as a major competitor there. And we've seen announcements from HP, Lenovo, and the like.

AI servers is interesting. Last year at this time,

People could not get enough of Dell in their portfolio. You probably recall, I think it was the last year in early March, the stock gapped of 35% when they put up a big AI server order number. And I just talked about euphoria and excitement and then the aftermath. The aftermath has been a lot more challenging. And it's not because they haven't had good AI server backlog. In fact, they won a $5 billion AI server deal with XAI.

But the challenge has been the margins. And Dell's gross margins have come down. This year, if they achieve their guidance, will have been about 250 basis points, I think, over the last couple of years. And they've been emphasizing... To be fair, Dell has been very clear that AI servers come with lower margins.

But the gross profit dollars will be higher. But the market simply, I don't know if it's just they haven't cared or some of the other segments which you can talk about have been a bit challenged such as the PC market. And as a result, that stock has been in a range for quite some time despite what I would argue is executing really well on AI servers. And just one final point on this is that while the gross margins have been challenged,

they've met or exceeded their operating margin targets. And in fact, their OPEX was down last year. They talked about AI actually contributing to that. So you look at it and I'm like, yeah, they're slightly beating earnings estimates. They're executing well in AI servers, but there's a lot of fear about, is AI

our AI server is going to be commoditized. Yeah, for them, fiscal 2024, they had 24 and a quarter percent gross margins, expected to be just a little bit under 22%. So the operating margin, the way you just kind of broke it out, that's kind of the more important, I guess, metric right here. But the flip side of that is,

you know, super micro, I think that they had 11% gross margins on like two quarters ago when they announced. So like when I say it was a one horse race, I guess I meant from a profitability standpoint and not 22% gross margins are particularly, you know, phenomenal. But when you think about hardware, you know, ultimately this all gets commoditized and, you know,

I don't know. I find this like the least interesting part about it. If I go back 25 years and I think of the server market in general, you know what I mean? And then I look at the underappreciation. Other than that period you're talking about, March of last year, and you saw that big run-up, but then the expectations got so high in the report of the next quarter, the stock went down 27%, and then didn't see an uptick for like three months. Well, and one of the – a few things with Dell I should highlight is that

They have that much higher margin profile than a super micro because Dell does have some of the bells and whistles, if you will, that corporations like. They have the storage. They have services. And so they have an ability to layer those in and attach those to AI servers. And they've talked about a two times attach rate for their storage and their services. But

this has been largely a cloud market and the cloud buyers are saying, "Hey Dell, thanks for the AI server, but we can handle the rest." And so I think Dell has the challenge of convincing the market by convincing its customers that they have a big attached play at hand here. They might when the AI market brought in some bigger enterprises that are non-tech oriented, but so far that hasn't played out I think as quickly as investors would like. So tough end market for sure. AI servers are

growing top line very, very quickly, but difficult to extract margins, which has been the big challenge. Right. So let's talk about storage and memory. Yeah. I'm looking at Western Digital. It's basically trading very near 52-week lows. Micron is trading very near 52-week lows. Right. And I put that together with Dell, which is also...

not far off. It just seems like as anything away from Nvidia hyperscalers, you know, no one wants any piece of it. And then again, we just talked about that all of the hyperscalers were down 27% from their recent all-time highs. So it just seems like to me that the fever has broken in this trade to some degree. Well, no, absolutely. It's a great point. Micron just reported last week and initially the stock went up, you know, they beat earnings. They

had very bullish commentary in the press release and going into their earnings call, mostly it was very good. There were some good vibes going around because what they highlighted was their high bandwidth memory, which is tailor-made for AI GPUs, was beating and exceeding expectations. But then as folks looked more under the hood, what they saw is that gross margins are, again, down quarter on quarter. Now, Micron had highlighted that throughout the quarter at conferences,

But nevertheless, it left people asking the question, "Well, if I'm seeing all this strength, why?" And AI is providing so much goodness, if you will. Why are your gross margins going down quarter on quarter? They had some issues with the NAN business that they have that's subscale for Micron and has been probably negative margins at the moment due to oversupply. But if we just take a step back and think, "Well,

Again, if AI is supposed to be providing all of this value add further down the chain, why am I having margin issues at Dell? Why am I having them at Micron? Why am I having them possibly at some other optical component companies and the like?

And so I think that's been the challenge and one of the reasons why Micron kind of quickly faded last week, and I think it was down 7%, 8% on earnings. - You kind of alluded this before, where do these use cases, where do they emerge, right? And cybersecurity is one of them. And we talked about the,

the double upgrade in Cloudflare. There was also one in CrowdStrike. Everything with the crowd was doing pretty well this morning. This is on Tuesday morning, Tuesday afternoon as we're recording this. Let's talk a little bit about what you think this means for cybersecurity and is this an area that you're really focused on for the balance of the year? - So I think with cyber, I mean,

The companies, obviously a lot of companies with a lot of different flavors here, but I think there's a few things that are very positive for the cyber companies if I look at it from a top-down perspective. So generative AI is expanding the attack surface. What I mean by that is sophisticated hackers have...

have a bigger toolbox to work with, essentially, with generative AI. As a result, you need greater protections. And not only that, but relative to, say, traditional software or SaaS seat-based software, you have the potential to expand your market in cyber. And the reason is that agents...

these AI agents are on the come and they too will need protection. So not only us as humans will need to be protected inside the workplace and outside as well, but also you have an increased proliferation of agents and you have companies like Okta,

uh, one of the, one of the leaders in the identity security market as well as cyber are, you know, much more vocal about this. So I think that's really interesting. It's early days to be very, very clear, but you have TAM expansion again with the increase of taxer attack surface and the proliferation of these agents. Yeah. So talk to us a little bit about agents in general. This is something that, um, you know, Mark Benioff at Salesforce has been, uh, pushing really hard. Um, they've, uh,

know released a suite of products it doesn't seem like it's getting he's talking about it um you know it's seen this massive uptake off a very low base but then when you start talking the financial contribution it's it's not particularly you know that great and you're thinking about a company that's growing probably mid to high single digit revenue right now so it's not a hyper growth company is this something you look at their ability to kind of grow this market and integrating across a lot of different industries right because they touch a lot of them is this the potential you

in the not so distant future to be a real growth engine for some of these SaaS companies like Salesforce? - Potentially. I'd emphasize that word potentially because you take a Salesforce that is a, call it 35, 40 plus billion dollar revenue company currently. And so I think Mark Benioff's enthusiasm combined with the guidance

perhaps a little bit more muted, you know, is something that people have been asking about, like, you know, how to marry the two. And you'd say, well, look, like this is a massive scaled company. It has four established clouds already. And then you have this agent force, great product, you know, feedback in terms of the usage and customer feedback has been positive, but you're starting to your point, Dan, on a very low base. So to move the needle,

in any given year, particularly this year, is really difficult. If this doubles and it contributes a few hundred million dollars to the overall revenue, that's barely a drop in the bucket. And that might be an aggressive assumption. So agents are evolving. This is going to be a market that

is going to play out over many years. But they're smart to stake out a position here, but it's still just not going to move the needle in the near term. Do you spend a lot of time thinking that's a good opportunity looking a year or two out for them? But investors have looked at a company like Adobe,

I think that all this innovation is coming at their core suite of products. And the stock is down considerably from the highs in early last year. It's down nearly 40% or so. It's barely seen an uptick. So I'm just curious, how do you think of some of the ones that have major headwinds?

Yeah, Adobe's been a really interesting one and it just is not getting the benefit of the doubt. And I know it's frustrated them, you know, unsurprisingly. And this is a company that you think about over the last decade plus, they were a leader in going from license-based traditional software model to SaaS. And they essentially executed beautifully through that period and had, you know, pseudo monopoly in their creative business. And so...

As growth has slowed coming out of the COVID pandemic, and we're seeing closer to quality, low double-digit growth, investors are saying, wait a second, I have what was a monopoly, maybe not so much anymore. And I have all these generative AI tools. I remember last year when OpenAI Sora came out, the Adobe stock got really hit hard. And it's been just kind of like we were talking about with Google earlier. It's just been difficult for them to sort of...

unproved this narrative that's out there. And, you know, I think it was during the holiday season, Coca-Cola put together this really interesting ad, mainly this ad campaign, mainly using Gen AI. And it also caused a bit of a freak out moment in terms of, oh, wait, are Adobe's tools going to be as prevalent, you know, in these advertising and marketing campaigns going forward? They're prepared. They're not going to, you know, go away. But by any means, I mean, it's a phenomenal company.

But again, we have a lot of questions and again, a big reason why the stock didn't hit so hard. Yeah. Before we get out of here, let's talk about some of these consumer internet names that you focus and how you think that generative AI either serves as a headwind like we just described with like an Adobe or you see it as something that could be really additive. And maybe we could throw an Airbnb, an Uber, Instacart, like some of those sorts of names because

Again, I mean, they heavily rely on certain components of this sort of technology, but maybe in some instances, you know, they could be, you know, kind of headwinds, especially when you start thinking about agents and how they might work, you know, for folks, you know, who are kind of using these services and products in general.

No, it's a great point. So when OpenAI came out with, I think it was January of this year, a released Operator. Operator is essentially an agent that can go on your browser and essentially order. I think the vision is ultimately be able to order food for you or buy plane tickets for you or what have you. And so it was a bit of a, I don't want to say a freak out, but definitely a oh my God type of a moment in that are they going to basically render Instacart or Booking.com and companies like this

essentially useless or at least circumvent them. But then these are their partners with the OpenAI operator launch. Instacart is one of them. And that's an example, Cart is, of a company that is

still fairly manual intensive. The shopper goes around picking up the items that you order. And while the shopper dynamic is not going away, one of the things is that they can be so much more efficient if they have an update on what inventory does my grocery store have? Are the avocados that I asked to pick up, are they available or what have you? And so the potential for

to improve the unit economics for a company like an Instacart, for example, which is a leader in grocery delivery, you know, is immense. And, you know, this is a company that, you

unit economics make all the difference in the world, delivering items from point A to point B is a fairly labor-intensive task. Any efficiency that can be gained is tremendous. They, too, have an advertising business that provides some pure profitability. There's generative AI potential in terms of improving conversion for a company like them, such as what, say, a Meta has done already and has shown that can be achieved. I think there's a lot that can be achieved that we barely scratched the surface on some of the consumer internet use cases

But again, I think I want to watch really closely what an OpenAI operator and other products like this coming out will do. The OTAs, like a booking, have been called into question on this very dynamic as well because Google search is such a big part of their lead generation.

But again, they have the relationship with the airlines. They have the relationship with the hotels. And so that's going to be difficult to overtake. And I wouldn't count them out. All right, man. We could have done this all day long. At least I could have done it all day long. I'm sure you're getting a little tired. No, thanks so much. Really appreciate it. So that's Par Shah. He is a portfolio manager at Kingdon Capital. Really appreciate you being here. Hopefully after Q1 earnings, Q2 guidance will come back. That would be great. All right. Thanks. Really enjoyed it. Thanks. Thanks.