Ep12. Tech Stocks vs the Rest; AI VC Bubble, Calcium CT Scan | BG2 with Bill Gurley & Brad Gerstner
57:29 Share

People talk about the military industrial complex. We may have created a healthcare industrial complex that really can't stop maximizing profitability and different ways of making money and not focus necessarily on the lowest cost, best, most preventative process. For sure. That would be my guess as to what's going on. Yeah.

Hey, Bill. Hey, Brad. How you doing, man? Enjoying the summer?

I, you know, it's, it's a good start to summer. I was down in, uh, I was down in DC. I picked Lincoln up from an internship in DC, I guess, just over a week ago. Now, you know, it was actually the night of the debate. We hosted a poker game on Capitol Hill and we had members in the game from both sides of the aisle. And we're kind of watching the debate in the background. And it was pretty wild because, you know, we got through as we're watching the debate

you know, nobody was really surprised at what was happening. You know, the biggest surprise of the night is nobody was surprised. They're like, oh yeah, like this is, this is, you know, the condition of the current president. And yet, you know, so I, I, we ended the night and ended the debate. And I thought like, there was not a lot to see here, um, because we've been talking about this for months. Um, and then of course we had the furor of the last couple of weeks, um,

You know, so that was good. And, you know, it seems to me the last few weeks have been like all politics all the time, every every every pod, every show. You know, and I looked at the calendar. We're a few days away from the Republican convention. It starts next Monday in Milwaukee, which is really the kickoff of, you know, the the final push in the campaign. The Democratic campaign I looked is five weeks from now in Chicago.

So it's interesting. Sometimes they're closer. And this time they're about five weeks apart. And then I counted up. We have 90 trading days until the election. So 90 market days until the election. And so I thought, you know, maybe we kick off here talking about how funds are positioning, particularly technology funds in the public market are positioning ahead of the elections and

Do you believe people are weighing the elections as a critical factor in what they want to be doing with their funds? For sure. For sure. But we'll get into that a little bit because you pushed me a little bit on how much markets are moved by active versus passive dollars, etc. And then, of course, you and I have been talking about it. A lot of our friends are talking about it. I think the biggest debate of the summer in AI continues to be the thing you've been talking about, I think, for quite some time on this pod.

which is, you know, is the AI hype ahead of the meat? You know, is there beef on the burger? And you've been making the case that, you know, the spending, I think, and some of the commentary is out in front of the AI benefits. And we've seen and we'll dig into, I think, a lot of articles out this week on this.

And then finally, maybe, you know, it was about a year ago that I was talking about a topic near and dear to me, this calcium CT scan.

and why it should really be standard of care for people over 35. I sent you a video, you know, on some of the politics of why calcium CT scan, partly because of stent and partly because of other things, didn't become the standard of care. And I thought, you know, maybe just as a little bit of a summer wrap, we'd revisit a couple of topics on that that have come up. Okay. Sounds great. That's awesome. Do you want to lead us in?

Yeah, I mean, tell us where you think the markets are right now. I agree with your point that politics have dominated the conversation so much, it's almost like people aren't even aware of whether the Nasdaq's up or down in a given day or a given week. It's not really on people's mind.

Tell us what you think is going on. And then I do want to dig deeper. There's a lot of data out that suggests consumer discretionary spending is dropping pretty precipitously. And I'd like to get your feedback on that. Yeah.

Yeah, well, let's get into it. I mean, it is wild, right? Like AI stocks continue to perform really well. Let's think Mag7. NASDAQ's now up 20% on the year. We have Mike Wilson, you know, of Morgan Stanley fame, who called the 22, you know, crash, come out and say recently he thinks we're going to have a 10% pullback ahead of the election. Right.

But yet the people who did the sell in May and go away, that was a pretty bad idea because May and June and early July have continued to move pretty meaningfully higher. But, you know, if you ask me to prognosticate on kind of what I think is happening, you know, listen, the market's pricing in now 60 percent chance of a Trump victory.

And with that, they're pricing in the permanence of corporate tax cuts that we've talked about before represent 20% to 30% of the earnings growth since 2017. So this is a really big deal. As a reminder to everybody, the tax cuts that were passed in 2016 that went into effect in 2017, those terminate by design next year unless they're renewed.

So they terminate at the end of next year. And so that would be a significant headwind to markets. And so I think as, you know, as Trump is kind of surged in the polls, the permanence of those tax cuts have also come, you know, into focus. And then I think the other thing is the market's now pricing in a 73 percent chance of a rate cut before the election.

So we have two Fed meetings before the election. We have one on July 31st and we have one on September 18th. And the Fed's been dropping some breadcrumbs, right? Referencing things like the slowing in the discretionary spending bill, referencing the fact that we've seen a tick up in the unemployment rate.

referencing the fact that core PC continues to roll over. It's given itself, I think, a window here to begin to cut rates. And remember, the market's most interested in that second derivative. Are we going up or are we going down? And I think the market's pretty well convinced that rates are going down. And so,

The economy is slowing, but it seems like a bit of a soft landing. And so that's the setup. And that's why I think we've seen such elevated performance out of big cap tech. But we have Q2 earnings starting in two weeks. And I think there's going to be a really important earnings season. Is it going to show acceleration? The reason stocks have been going up is because earnings have been beating.

You know, will it again occur, you know, in this quarter? I think there's a chance that this may be one of the trickiest quarters where the expectations have now creeped far enough ahead that it's going to be a little bit difficult for expectations to keep up with them. But I pulled a little bit of data, you know, courtesy of our friends at Morgan Stanley, Ashton Curtis. You know, he sends out this weekly email. You know, he had some data in there last week that I thought was interesting.

So he said, May and June have been two of the top three largest months of tech selling since 2010. Right. So I feel like there's a view in the world that, you know, everybody's crazy about tech and everybody's crowding into tech.

But here's some data that supports the opposite view. So long short equity funds have been adding to their tech shorts. Tech exposure has fallen to the bottom decile over the last year due to outsized selling. Semiconductors and semi stocks have driven most of the net selling during these months.

And overall, hedge funds are more underweight growth than at any point Morgan Stanley has seen in the last 10 years. What do you think is causing that?

Well, I you know, we've had this huge run up unabated over the course of the last six quarters. And I think that people are just saying, listen, we've got uncertain elections coming up. We've got an uncertain earnings season. The margin of safety is compressed. So whatever number of units of risk you want to have on, you're going to take a few of those units of risk off.

Right. If you've been, you know, like we're having a good year this year. There's no need to be heroes over the course of that money. Go, Brad. It's got to go somewhere. Well, you know, it's just gone into dry powder. You have more cash on the sidelines. You have more ability to buy on any of these pullbacks.

And so, you know, I think you see that through some of this active selling. But Bill, you asked me a really important question and I sent you some data on it. You know, you said how much signal is there in the fact that you and, you know, KOTU and Tiger and other people, Capital Group may be buying or selling. You said what percentage of that action today is active versus passive?

Do you see that chart I sent you? Got it right here. And was that a surprise to you?

Let me hear your analysis on top of that. Well, I mean, you know, so basically what it shows is that about 11 to 12 percent of the dollars in large cap tech, think MAG7, are quote unquote active dollars. So think long only in hedge funds. Yeah. So if you would have looked at this in 2010, that would have been above 20 percent.

So clearly today, fewer dollars that are being traded in the market, one out of ten dollars is what you would call a fundamental manager. Right. Somebody who's sitting there running the numbers. And this is obviously higher for the mag seven because they're so big. And so if you're buying an index, you're buying more of them than you would have.

on a percentage basis. I think that's probably true. And so I guess the point would be, or the pushback that you gave me is, yeah, these are interesting data points out of Morgan Stanley, but it may not tell the whole story because if fund flows continue to come in, right? If people generally are saying we're going to,

we feel good about the markets. We want to continue to invest in our 401ks. We want to continue all of these dollars passively flow into these ETFs. And so fund flows have continued to be strong as markets have gone up. So there was this tweet that you sent me, which is kind of the tale of two cities, right? It's what you referenced earlier. If you look at consumer stocks,

from McDonald's to Nike to Lululemon, SoFi, Walgreens, Papa John's. Just in the last week, they all hit 52-week lows. Yeah. Right? And then there's a group of stocks, which is casual dining stocks, which are, you know... Really that mid-market, that mid-market, the PF chain types. Exactly. Yeah. Exactly. And the Papa John's of the world that are down, you know, 30% to 40% on the year. And remember...

A lot of what happened here, Bill, is this is the pull forward that happened in 2021 because of COVID. We had huge build outs in these businesses. Earnings went from flat to really accelerating in these businesses. And everybody started valuing them at kind of peak multiples like they were growth stocks.

And then what we have now is, oh, got a little dog action over there. What we have now is them really just normalizing, I think, back to trend. So, you know, I think we have a good chart here.

courtesy of Goldman Sachs, which shows the earnings growth over the course of the last 12 months. So if you look at it for Microsoft, NVIDIA, Amazon, Google Meta, earnings are up 38% over the last 12 months. And the stocks, not surprisingly, are up a lot.

If you look at the S&P 500 as a whole, earnings are roughly flat during that period of time. And if you take out those tech stocks, the other 495 earnings are actually down 5%. So that's the reason...

There's this massive dispersion in the public markets. And the reason shorts have worked so well this year is you could be short hundreds of these S&P stocks that are actually down on the year and long just the mag seven, and you'd be having a terrific year. Is there, in your mind, is there not an implication of this that is a question about the broader economy? I also saw a...

Someone had tweeted out some credit card data. And one of the things that was down year over year was travel, a sector you know quite well.

So is part of what's happening with these casual dining and these consumer stocks a lack of spending? Is that something you're worried about? For sure. I mean, listen, everybody – if you look at folks other than wealthy folks in this country, they burned through their stimulus checks.

Right. Their excess savings are gone. Credit card borrowing is hitting, you know, every single month a new high. And so I think you're seeing the results. People are just having to make tradeoffs. And, you know, travel happens to be one of the last things people want to trade off. Right. They'll forego the new refrigerator. They'll forego eating out every night before they'll forego their summer vacation. But, yes, that's even being touched.

So this is exactly what the Fed wanted to orchestrate, Bill, right? It wanted unemployment to go up. It's getting unemployment to go up. It's getting consumer spending to slow down. It's why they're in a position to cut rates. And the gift that the markets had really is that at the same time, like we would otherwise be talking about a recession, right?

But we had this surge in infrastructure spending around AI, right, gave life to all of these technology companies. And one of the things that, you know, we're going to put this chart in here. One of the things that you asked me is, you know, is the big PE, you know, names, are they overvalued? And so I asked my team to go through an exercise, look at the peak multiple performance.

that they were trading at in 21, Q4 21. Look at the trough multiple they were trading at in 22. And then where are we today?

Right. So, you know, take Apple, for example, at its peak in 21, it was trading at 28 times at its trough last in 22. It was traded 22 times today. We're at 32 times. So we're above peak for Apple. 32 times for P forward P. Yeah. Okay. Consensus forward P in the case of Meta. Yeah. Right. Through them all. Just because some people will be listening. Where are they? Yeah. In the case of Meta.

right? In 21, you peaked at 22 times. We troughed, remember, at 90 bucks a share in 22 at 12 times. And now we're back at 24 times. Okay. In the case of Google, peaked at 24 times, troughed at 17 last year, now back at about 23 times. In the case of Amazon, we peaked at 64 times.

Right. In 22, it got as low as 46 times. Now we're at 33 times. So Amazon is actually as a multiple of earnings trading below where it was during those periods. Why? Because they've been cutting costs, et cetera, and driving earnings as growth has slowed. In the case of Microsoft, peaked at 37, troughed at 23. We're back at 35.

So you get the point here. And in case I skipped NVIDIA, 66 times was the peak. The trough was probably the start of this year, actually, at 20 times. And now we're on consensus number back closer to 40 times. And so if you look at these, I think you can make an argument on both sides, Bill. We certainly are nowhere near the trough.

But these don't this doesn't feel to me like the stuff that bubbles are made of. Right. I can make an argument that you could these things could all be down 10 percent in the next three months and none of them would be screaming buys. Right. Right. None of them are in the territory where I expect that they would be down 30 or 40 or 50 percent.

You know, they're not trading at hundreds of times sales like a lot of software companies, you know, were at the end of 21. And so I think that, you know, as we sit here, I think it's a tricky moment.

There may be an earnings error pocket in this quarter. I think there's less margin of safety in this quarter than prior quarters. We got the uncertainty of the election. So, you know, if I were playing from home, I'd have a little less on, take a few units of risk off the table. But ultimately, I can paint a pretty bullish scenario for you, right? How do you frame the fact that, you know,

It's the same thing we've been talking about for several quarters now, where you've got these seven stocks that are performing, right? They're performing from an earnings growth standpoint, from a cash flow, free cash flow standpoint. Yet, what we just talked about, the entire rest of the market's not performing.

So what does that mean? Do you have to own them because they're the only thing that's working? Or is it scarier to, is there anything about the fact that the rest of the entire rest of the S&P isn't working that should cause concern?

Yes, I think it should cause concern other than the fact that the Fed is in a very enviable position from a monetary policy perspective to cut rates and add some juice to the economy when it needs to. And I think that's what's going to happen, which is part of the bullish scenario, right? The market is saying there's a 73% chance the Fed's going to cut rates. And the reason for that is

is unemployment's ticking up and you have 490 stocks whose earnings are not growing. And that is a recipe for more economic distress if we don't get in front of it. - How do you, if you're looking amongst these seven, are you willing, I'd be curious if you're willing to make a comment like,

Which one? And maybe with the backdrop that AI might be driving some of these, which ones feel safer to you here? Yeah, it's a good question. Well, first to say we own them all. OK. And, you know, the second thing I would say is altimeters peak exposure is.

to these names was about 70% over the course of the last 12 months. So we had 70% of our dollars at one point in time in these names. Remember, at one point, I had 30% of my dollars in meta alone off of the bottom. So we had a lot in these names.

We've taken a lot of, you know, almost half of that off the table. Now, we haven't redeployed it further out on the risk curve because for all the reasons you and I talked about, I'm nervous about things further out on the risk curve. But that's just, you know, comes to us probably dry powder. What I would say here is, you know, again, if I turned the tables on you, Bill, and said you and I have been doing this for a really long time –

And do you think more money has been lost trying to time these things or made by timing these things? And my instincts here are you can't just like totally be out of this market. If you have, you've missed a tremendous move over the course of the last six quarters.

And the bullish case I was going to make to you is I think, you know, if the market, if the betting market is right, there's a 60, 70 percent chance of a Trump victory. Set the politics of that aside. But that means you're likely to get a permanent tax cut at the same time that you're getting a reduction in rates. Right now, imagine if you get a resolution to the conflict in the Ukraine.

Now imagine if three or four quarters from now, some of these AI benefits start catching up with the software companies and their investments, right? That is a highly constructive scenario that you would want to be invested against. And so, you know, like, I don't think you have to be all in all the time. But I think that's what this market is beginning to sniff out. And that's why, you know, after a blockbuster year for technology last year, you're having another good year this year.

My quick reaction would be that Meta is somewhat unique in the group in that if you

If there were an air pocket or something that came out of the LLM AGI world, they're not really. I mean, they're playing with the open source models, but they're not really dependent on it. And their core AI use case isn't even LLM based. It's driving their ad matching and their content engine. So it seems like they get all the good without any of the risk. Yeah.

And anyway, for what it's worth. Listen, I think that there's a maybe good time to shift to this conversation about AI costs being way out in front of AI revenues and whether or not that constitutes a bubble. But certainly meta is, you know, however you want to define it. It's one of the most profitable AI companies in the world, as is ByteDance, TikTok.

But, you know, you've been saying for some time. And who knows what I don't have a strong point of view on which candidate is more likely to follow through on something on TikTok. But that would obviously be a massive boon for for Facebook and Meta if it happened.

No doubt. No doubt. That's, you know, again, if we look at the calendar, that's something else that, you know, the clock's ticking on. I mean, shortly after the election or I think shortly before the election, the ban is meant to go into effect. Now, I think it is in the courts. So maybe there's a stay of execution there. But there are a lot of things that are potentially going to drop here and that this election is going to have a lot of consequence on.

Coming back to this topic of AI, you know, it's interesting to me.

You and I have been talking on this pod. I remember when we went over the $2 trillion build out for Nvidia over the course of the next four or five years. And you asked the question some six months ago, what the hell do you have to have in terms of revenue on the other side of this in order to justify these investments? And over the last few weeks, a chorus of high profile firms have seemed to come out and say versions of the same thing, Bill.

Maybe we just start there and you break it down a little bit. We go back and forth. But there was, you know, maybe the first piece that came out, David Kahn over at Sequoia. You know, you've got Goldman Sachs with a piece, Modest Proposal had a piece. McKenzie did some work that was all quite skeptical. But David...

David Kahn had this piece called, you know, AI $600 billion question. And in it, he said, the AI bubble is reaching a tipping point. Navigating what comes next will be essential.

Right. Effectively arguing that there's a big gap between revenue expectations implied by the build out and the actual revenue growth that he sees in what he calls the AI ecosystem. Yeah, that sounds like a pretty stern warning to venture firms and entrepreneurs out there. Yeah. And yeah. And look, look, the main thing, the main point I was making was that the more you promise

the more you set up the risk that someone's going to start taking the other side of the argument. And so I'm not surprised that this happened, mainly because there have been people willing to say the most outlandish thing or the most optimistic thing, or this is going to go on forever, this is going to scale forever, it's going to solve every problem. No one's going to have to work again. We're going to have UBI for everyone. These are pretty massive claims. And so

And I think the other big thing to keep in mind that I don't think we've seen really before in a tech ecosystem or a tech cycle is this one is very CapEx forward.

And that's what's driving NVIDIA. That's why NVIDIA is the biggest winner in this whole thing. But this Mag 7 is spending on a CapEx level at an accelerated rate to anything they've done before. And so it all puts it out there. What David did was simply try to aggregate the spend that was being put in place and then...

and think about that as supply and then say, what will it take for there to be demand on the other side to suck all this up? And it's a public link. We'll put it in here so people can go read it. And I recommend people look through all this stuff because it's all super interesting. But he's having trouble adding up the other parts of the other side. And I think one reasonable question to ask is why would someone at Sequoia –

go out on a limb with a pessimistic view. We all know that the competition to get into, whether it's early stage, mid stage or late stage,

deals is somewhat dependent on your ability to be seen as an optimist and to be seen as someone that has a positive point of view. And so why? Why would someone take that point of view? And I reached out to him because I was just curious about it. And his view, which actually is somewhat sensible to me, is if you let the markets run crazy

you end up with a higher risk of a reset. And I will also tell you, having lived through what happened with SoftBank and the last cycle,

Having crazy money in the market is not necessarily consistent with creating positive return for your venture capital portfolio. It can start to trend the other way if you end up with hyper competition, if you end up where every single player has $400 million, which I think is true in the market.

in the co-pilot AI coding space. And so I think, I think his intuition is let's keep things reasonable and let's keep the card on the tracks and let's not go out over our skis. I mixed a bunch of metaphors there. Okay. So, so, so let's just break it down a

a little bit because I think it's always difficult to know out of, uh, out of these arguments, you know, Sequoia is one of the biggest investors in AI right now. They've probably done more rounds than just about any firm. And, and, and so here's a firm that's doing a ton at what some might argue are pretty high valuations. And in fact, go on Twitter and find a lot of these, uh, critiques. Um, and at the same time they're saying, but you know, don't go out there and do too many deals. Um, yeah,

You know, it seems to me that, you know, as I look there, of course, it's what Satya said at East Meets West a couple weeks ago, Bill. Investment always comes before the return.

And it's almost never perfectly matched in terms of timing, right? It's like, it's impossible to think that it would be perfectly matched in terms of timing. I would argue here, like we don't really see big rounds out of, you know, the soft banks and tigers of the world. The people we see big rounds out of are, you know, the Microsofts and NVIDIAs and Amazons of the world. And, you know, Meta's investing a lot of dollars, right?

And so, yes, I do see that there's no doubt that AI investments in the venture landscape are about the only thing getting done. They're super hot. The valuations are really high. Again, I think we've been somewhat circumspect on AI.

a lot of those rounds simply because we think they're great companies, just hard to earn a return from where a lot of these rounds are getting done. But, you know, my sense is that this is more a critique on, you know, a timing mismatch than it is a critique on whether or not we're going to get there. Like I didn't read anything in that article or the Goldman Sachs article that caused me to lose hope whether or not AI is going to be, have a major impact.

I think that would come, let's, let's transition to the Goldman. So, so in addition to Sequoia, who you might ask, why would they have an incentive to take a skeptical point of view? Goldman put out a rather, I mean, it's a readable link. It's like 30 pages. I recommend anyone read it. They can get ahold of it. I don't know what the current state of affairs are on, on research dissemination and we'll find out if we can post it. But, um,

They started with two people. They had a professor from MIT, and he took, I think, a pretty strong point of view. He said many people in the industry seem to believe in some sort of scaling law that doubling the amount of data and compute capacity will double the capability of the models. But I would challenge this view in several ways. And he goes on to suggest he doesn't believe in

in, in kind of either linear scalability or exponential scalability. And, you know, that gets back to a conversation we've had in the past. Like we'll, when, when, when Sam says four, five, six, seven, and says, Oh, they're all going to be as better than the one before. Is that real or not?

And I think it's kind of a big outstanding question that's out there today. And then their head of research, who used to be a semi-analyst, he came out and took a point of view, I think very similar to the Sequoia point of view, that the build-out of over a trillion dollars will need to have massive investments.

you know, return coming off of it in order for this to have worked out. And it gets back to this point that we're putting all the CapEx up front. And I pushed back on you a little bit in that whether it was the mobile wave or the internet wave, like it didn't require,

this much betting on the cone. You started to get some results right away that, that, and, and, and, and people were able to leverage technology. I mean, the internet had been built out over 30 years prior to when it actually kind of flourished. And so there wasn't this, like every player didn't have to put up so much money. Um,

So let me just say a couple quick comments. First, we got an MIT professor saying, I don't know if I believe in the scaling laws. I think you have all the best folks in the industry today from OpenAI to Anthropic to Elon, et cetera, who have all said they were surprised to the upside of

by larger models leading to really breakthrough outcomes. And in order to believe that professor, you have to believe that Elon doesn't know what he's doing. He just bought a, is investing in 100,000 H100 cluster, just raised a ton of money to do it. He clearly believes that there are upsides to scaling.

Open AI, you know, Microsoft clearly believes there are upsides to scaling. Google believes there are upsides. So the folks on the field, not the professors critiquing from the towers, are spending their money because they see the benefits. So, you know, I guess that critique doesn't hold a lot of water there.

you know, for me, I think that, of course, they all also acknowledge, like Jan Lacunas argued, that this may not, in fact, be the path to AGI. And we don't know how many more, you know, rounds of scaling are going to continue to lead to enough beneficial improvement to justify the incremental spend. But most people, I think, on the field today would argue that we're not there yet.

And then the second thing is, it's not true to say that prior cycles didn't have big investment. I mean, by our analysis, for example, just Amazon spent over $100 billion on AWS and

invested for over eight years before they saw profitability in that business. We know that Reality Labs was willing to make a $20 billion investment per annum for a decade in order to bring us virtual reality. We think they brought that down and pivoted more toward AI. But the fact of the matter is that we have companies that are

printing cash, multi-trillion dollar businesses. Think of it like a nation state bill. If you were a nation and you had surplus like these companies have surplus, what would you do? You would invest in your national strategic advantage. And these companies are investing in the single greatest vector of potential advantage in the future. And the fact that they all agree on that, that

you know, suggests to me, you know, coming at it from diverse points of view, that there's probably some logic to the stuff that they're investing in. And then finally, I would just say, I do see benefits. Granted, I agree that there is hype that is running ahead of the reality. But to the tune of, you know, $3 billion, OpenAI has people reaching in their pockets and paying real money, including myself for their service.

And I do that rationally because I use it every day. It's a great service. Now, I'm sure people will come and compete and those prices may get driven down, etc. But I think we're still at the early phases. And finally, and I'll shut up, you and I both came on this pod and talked about FSD-12.

I mean, we're seeing breakthroughs in self-driving that you and I did not think three years ago were possible because of model breakthroughs that have occurred. Now, granted, those may not be LLMs, but it is still the same AI infrastructure that we're talking about and these chips that we're talking about that are unlocking those advantages. So,

I just, you know, the Goldman piece again, didn't, didn't hold, except from saying, don't get too over your skis. Well, great. I agree. Don't get too over your skis. But outside of that, I don't, I'm not really sure the argument that they're making. Well, I would, I would, um,

I would definitely double click on one point you just made, which is I do think that traditional AI models that live behind products like FSD12 are very, very different than LLMs from my point of view. And those are the same things driving Meta's internal engines. And if you have a problem,

that is set up to be solved by that type of solution, I think there's unlimited potential.

And so I buy into that. I think the real question about scalability is specifically tied to LLMs and not pointed at AI. In addition, there's so much interesting content came out. McKinsey also put out a piece that was basically a bunch of survey work from organizations that have been

playing around with AI. And a couple of things I think really popped. One, people still have trouble, right? They have trouble with hallucinations. They have trouble with errors. They are making progress. They're making progress in the areas we would expect customer service, number one,

sales enablement a little bit number two and and it goes back to a point i i just feel very strongly about which is llms are great at text they're horrible at numbers and um even kevin scott on a podcast he did ironically with sequoia that's over this past week um he he made that point and so um

I think there are a bunch of interesting questions. So let me lay out a few of them. One, could the CapEx spin be ahead of itself, which is what Sequoia mentioned and what the second piece of the Goldman piece is. Is LLM scaling possible?

linear, or will it diminish? And in addition to the MIT professor, I mean, there are a number of people that have been around these problems for a long time who believe that the LLM models will have diminishing return on spend. And clearly, Dario doesn't think that. He had a podcast out this week on good company. I would encourage people. Sam doesn't believe that. And Kevin Scott, obviously,

Clearly, on this podcast he did with Sequoia, he's the CTO at Microsoft, shares the enthusiasm that Sam and Dario do that this is just going to go on forever. Like, we're going to put more compute into more training and more data, and the LLMs are going to get better and better and better. And I think that's an unproven point.

And you sound like you're in their camp. I'm probably more skeptical. I think that the way that LLMs work is actually set up for diminishing marginal returns. But we will see.

Hopefully we'll see very soon. Um, but, but we haven't, one of the reasons we haven't seen is because things keep getting pushed out. The amazing voice demo that we saw on the last open AI demo day, um, they had talked about releasing the voice product in two weeks and it's been months, months. It's not out and people aren't even talking about when it's coming up. Um,

And so anyway, things do. I think that's a big question. Another big question that's hinted at in the McKinsey piece and a little bit in the Goldman piece is how much of this is going to be cost reduction versus revenue creation? And if it's only cost reduction, is that enough?

And even in the cost reduction case, I think some of the initial claims, the Klarna case, modest proposal called a prockerful, when you talk to people about how much benefit you get in coding, it was originally like 40%, now 30% and 20%. People have come off.

of their original claims. So there's no question it's adding value, but I think, and then, you know, where are the revenue creation pieces? We do have some companies in our portfolio where they're charging more for the AI piece. That's revenue creation. I think the more of that stuff we see, the better. So-

No, listen, I would ask you, it seems to me the greatest danger, Bill, right? Like if Meta invests a little bit too far ahead of where the benefits are, right, the stock will be down a bit.

And if Microsoft does, the stock will be down a bit. But there's nothing lethal to these businesses, right? They can move the training clusters over to inference. A lot of things they can do if some of these things change. It seems to me very different for a venture-backed business, right? And we see a lot of these venture-backed businesses, right, that raised billions of dollars out of the gate. They exist at the beneficence of the capital markets, right?

And people have to be convinced that they're going to have revenues that give them payback. And we saw another announcement this week that Adept had sold or been parted out to Amazon. We've seen this with a couple other businesses that are effectively, it seems to me, like take-unders. Did you have any perspective on that? And it seems to me if there's real risk in the ecosystem, the risk is to all of these venture-backed companies

Because they're the ones who have to become self-sustaining. The guys in the public markets, these are trillion-dollar businesses throwing off tens of billions of free cash flow a year. If they waste a year's worth of spend, that's not a question of existence the way it is for the venture-backed businesses. You're right. Look, the Adept thing is – which was –

by the press is similar to the inflection outcome. There's no way that's a win for venture capitalists. Like I'm sure venture capitalists try and spin it. And we heard some spin around inflection, but you know, it,

It's just not possible. I mean, we need to be realistic. Like if you're a venture capitalist and every one of your deals turned into one of these weird human acquihires with a licensed deal back, you're not going to generate a return for your portfolio. Like it's best case. This company had raised $400,000.

and $15 million, or how much are they going to pay for the license deal? And then you leave this thing kind of pseudo alive, like, and then the people that are running it, are you going to demand a dividend on that amount? Like, how do you get, like, there's no way this is, uh, optimal or even, even really something you should celebrate. And, um,

It is driven, I suspect, by the limitations on acquisition by the big players. But it's also just an alternative to a low-priced sale. It's not a win.

And and I wonder, you know, if part of it is the founders raising their hand and saying and if you read the press release they put out, it kind of says this, like they're kind of afraid to be on the field, you know, to your point. Right. And and when you put this much money in these companies and you raise. I've said this a lot and I'm sure I borrowed it from someone much smarter than me, you know,

Valuations represent discounted future expectations. They're not an award for what you've done in the past. And so if you raise at a billion, if you raise it 2 billion, if you raise it 14 billion, the shit you have to accomplish...

to live up to that expectation you just signed off on is really high. And so you have to find revenue. You have to find performance. And it may or may not be there. And so I agree. I mean, I think you phrased it the right way. This is where the risk lies. And the playing field is difficult to understand. I mean, if you listen through all this content that came out even in the past two weeks,

The, you know, Kevin Scott made a comment very similar to one Sam made a while ago, which is don't try and add value just a little bit or don't even even said in this thing, don't go do a small focused model on, you know, let's say a particular vertical, because we're going to keep investing in the in the foundational model. And it's going to it's going to be great at everything. It's going to be great at the big stuff, but it's going to be great at that focus stuff and we'll run over you.

And so where are you supposed to go if you can't focus and you can't add value? You know, Sam said people are using only X percent of OpenAI. They need to use 90 percent. But what is it you're doing if you're using 90 percent of the other model? I don't know. I don't know where you're headed as a startup. So it's a it's an interesting playing field.

Well, I think that I think it's well said. And perhaps that if we had to tease out a red thread that I think is actionable from all of these pieces, the Goldman piece, Sequoia, McKinsey, et cetera, you know, it is just an acknowledgement that things have run up a lot, that billions are getting invested in venture, that we're back to high valuations, that expectations are going up, that public multiples are going up.

And we're getting closer to a reckoning, and it could fall either way. But the distribution of probabilities make this a more challenging time for those venture companies that are lucky enough to raise at these high-priced rounds. But let me just shout out one that you and I met with, Bill, which is Arvin from Glean.

You know, here's an example of a company that I've talked to 10 of their customers. They absolutely love the product. They're getting, you know, real benefits. They would not want to give up the product. You know, it's enterprise search meets AI. And I do think that there's a lot of real work getting done under the covers. And I see with them...

like I see with open AI with, you know, with some enterprises and with lots of consumers, that feeling of magic that it really changed. You know, our friend Rich Barton said it's the one piece of software he couldn't live without, right? Like that's, you don't hear that out of CEOs very often. And so, you know, I think kudos to the guys who are grinding and creating that. But I do think that there are a bunch of folks who,

I see a bunch of grifting in the AI landscape. It's where you can raise money. They're raising money at high valuations. I do think that the mortality rate in AI is going to be very high.

And the winners are going to be incredibly large. And if this is all as big as we think it is, this is going to play out over decades. This isn't going to play out over the course of the next 10 days. And, you know, again, I'll just repeat the warning Satya gave East meets West. You know, he's like, I'm a total believer, but I also am preparing the company that we may have to have a big drawdown between here and there. Um,

You know, if these things aren't perfectly sequenced. And I think the whole ecosystem needs to be prepared for that. Yeah, fair enough.

You know, I got a note the other day that really impacted me. And, you know, I've been talking a lot about these calcium CT scans, you know, and just in the spirit of mixing it up a bit, you know, Bill, you heard me talking about them a year ago. You got a calcium CT scan. So many folks I know have gone out and gotten one done. And, you know, I've been talking a lot about these calcium CT scans.

And, but this was a woman who said, you know, she got it done. She didn't think women were really at risk, but I, you know, she had heard me mention it enough times. She thought she'd get it done. She got a score in the 900s. It really scared her. She found a lot of plaque and blockage. And so I did a little bit of digging and here was the thing that a stat that blew me away. Five times as many women die of a heart attack each year in this country as die of breast cancer.

Five times as many women die of a heart attack in this country versus breast cancer. And while the mammogram is appropriately and widely accepted as standard of care, right? Very, very few women do a calcium CT scan, which I might argue, or I've heard argued is the mammogram for the heart, right? David Marin, Dr. David Marin, the head of preventative cardiology at Stanford says it's the mammogram for the heart.

As a reminder to folks, it's 150 bucks. You can Google...

You know, calcium CT scan, you can get it done at any number of clinics in whatever towns you live in. I've had hundreds of people over the course of last year, text, email, and they just found things. And, you know, they found either blockage or plaque or other things that they didn't know were going on with their heart. But even the folks who came back with a zero score say, I'm so happy I did it.

because I'm at ease. Like I know what the situation is. And we've all been trained to track this cholesterol, which is not predictive of heart health, right? People put you straight on medication off of a cholesterol score. Fast, fast. And by the way, and I had one of these mini pods I did with Marin. He said that half the people, right, who have good cholesterol have a bunch of plaque in their arteries.

And plenty of people who are in the red zone of cholesterol have no plaque in their arteries. So wouldn't you just rather get to the source and say, take a calcium CT scan, just say, do I have plaque or do I not have plaque in my arteries? But Bill, the thing I want to talk about, and I'll post this, there's a documentary on YouTube I would encourage people to watch called The Widowmaker.

And this refers to the artery, which most often gets blocked and causes a lot of death. But this documentary painfully outlines the politics behind the calcium CT scan and why the stent lobby and other forces, including the insurance lobby,

appeared to have prevented the calcium CT scan from becoming the standard of care. And a really interesting part of that documentary bill was a battle that took place in the Texas legislature.

You have the head of several medical institutions that kind of went to war with the state of Texas because the state of Texas passed legislation out of both houses that would have required the insurance companies to cover the $150 for these calcium CT scans. So this is just, it's another interesting place if you have 45 minutes and you want to see how government can get in the way of technology.

and how a bunch of vested interests try to keep this from happening. I would encourage you to watch it. We'll post the link. What's your thesis as to why the establishment is non-supportive of a low-cost

preventative scan? Well, what I've been told by heads of cardiology at Harvard and Columbia and Stanford and all the folks I've talked to, right, I ask them all a very simple question.

What do you do for yourself, your friends, and your family? Just tell me your care, standard of care for yourself, your friends, and your family. And 100% of them follow the same protocol, which is a calcium CT scan no later than 40. And if they have any signs of plaque, they go on a statin to reduce their LDLs, their bad cholesterol, as low as possible.

And so I actually don't think the establishment cardiologists are against it. I think that there are some folks connected with various lobbyists, the stent lobby. Now, there was a trial around stents, Bill, a few years back called the Courage trial, which basically disproved the

that stents are actually that valuable. What they found was that a calcium CT scan and medical intervention was at least as valuable as a stent. So for example, I think today, most people will not even have stents put in, even if they find high rates of plaque, they'll just do it through statins, et cetera. But it's interesting. I've asked them why they wanna spend as much time with me as they have. And they said, because you can get the word out.

Right. Millions of people are losing their lives and it can all be prevented by taking a calcium CT scan. But this thing, this thing's cheap and fast. So I don't think you really answered the question. Like, what's the why would people be against it? Like, what's the point of being against it? Listen, I think the economic point for being against it is insurance companies don't want to pay for it.

The folks who are in the business of repairing the heart after the fact, like those in the stent business, right? It's good for their business that you aren't preventing it medicinally. And then finally, I think, Bill, it's just never underestimate the inertia of the way things are.

You know, my mom, 88 years old, she goes to a doctor and I said, Mom, why are you taking this medication? She said, I don't know. It's just what the doctor tells me to take. Yeah, I hear you. I don't I don't buy into the insurance one just because I don't actually think insurance companies are interested in cutting costs. They make a percentage of what is spent. And the more the merrier, the more that's in there, the merrier. Yeah.

I do think that the medical industry, I listened to this podcast once, I need to find it, but where a gentleman argued that over 90% of NIH grants go towards things that are potentially monetizable. So either a product or a drug.

and that the NIH is less interested in funding things that might be just preventative behavioral things like what you eat or your diet or things that don't lead to monetization. And we may have created, people talk about the military industrial complex, we may have created a healthcare industrial complex that really can't stop

maximizing profitability and different ways of making money

and not focus necessarily on the lowest cost, best, most preventative process. For sure. That would be my guess as to what's going on. There are a bunch of states that seem to require referral. A lot of people don't even have relationships with a doctor. This is particularly acute in poorer communities. Yeah.

And so, you know, if you want to get a referral, you can simply Google referral calcium CT scan. There are tons of referral services on the internet. It costs you 20 bucks. You get a referral. But I would encourage you most...

imaging centers, you know, if you follow me on Twitter at AltCap, you'll see just search calcium CT and you'll see people from almost every city in the country who've posted where you can get them in their city, the cheapest price you can get them in those cities. There's a mobile unit up in Seattle that does it for 55 bucks. Walk in, no referral needed.

The bottom line is, you know, I have no dog in this hunt except for the fact that I lost my father way too early to heart disease in a situation that was 100% preventable. So I wanted to have this done for myself, for my siblings, et cetera. I had a low score when I got the calcium CT scan done. I went on a statin. It

reduce my LDLs, and I'm in control. And I have the information and I can watch this longitudinally. And I was shocked to learn some of these facts behind the scenes as to why it wasn't being treated as a standard of care. And so, you know, if we can move the ball down the pitch just a little bit by spreading the word out there, it is simply one of the easiest things you can do. And I think you should really think of it as the mammogram for the heart.

And as seriously as we all take things like colonoscopies, prostate exams, mammograms, et cetera, this is the number one killer in the world by a long shot. So the fact that we're not doing the preventative care on it makes no sense. Good seeing you, man. Yeah, great seeing you. And thanks for doing that. Hopefully you're safe from the line. Let's get her done. I'll talk to you soon. Take care. Bye-bye.

As a reminder to everybody, just our opinions, not investment advice.