Bill Gurley, Brad Gerstner, & Jason Calacanis on the state of tech markets, hosted by David Weisburd | E1875

28:29They're late stage investors. And, you know, they have a self-interest in keeping these companies private for a long time because like private equity, they can continue to put more and more capital in. It's like a private club for public market companies that only 50 people can compete for. It's a much better market for you to make money in than the actual public market. I don't think it's better for the founders. I don't think it's better for the companies, but you know, like club, right? Well, I mean, the names that, that, that, you know, you know, well, the crossover investors, the tigers, the go-to's, you know, the green oaks, the altimeters of the world.

29:04But then I think there's you know all of these companies stripe stripe stripe matches that matches that stripe bite dance spacex you know go through the list all of these companies could be public and i think unless you have a really good reason for the company you know to stay private i think the public markets as amazon as google as salesforce and so many others have proven this idea that you can't innovate in the public markets is total nonsense this idea you know it's a cheap source of capital. It keeps the business efficient and honest. And so my sense is that you're going to see the markets really open up in 24 and 25.

29:43We have a backlog because we've had almost no IPOs over the course of the last two years. And so it's just a matter of price. IPO markets are always wide open, Jason. It's just a matter of price. One interesting thing that facilitates what brad was talking about is we we've become quite comfortable with secondary and and in fact when we were going through zurp and there were people with tons of money in their pocket and they wanted to cram it into these companies they would encourage secondary as a way for them to get ownership because the founders and the early investors didn't want more dilution so they would literally you know knock on doors and say let us help you do this and if you're a rather large angel or early A or B investor or a big founder, it's hard for you to get liquidity through that type of mechanism, but it relieves a ton of pressure on the management team for the employees to be able to get liquidity.

30:45And so you find yourself in a forever cycle where these companies might stay private for a very, very long time. And I think we will see a few examples, And I don't know exactly which company, but where that proves to have been a mistake, like for the reasons Brad talked about. And I, my favorite, my favorite metaphor here is to, is to imagine a, uh, football player that goes from high school to college to pro. And, um, you know, could you imagine, you know, on draft day or a week before draft day, the number one quarterback candidate says, you know, I think I'm going to opt out because. You know, if I play on Sunday, they're going to, they're going to be way more scrutiny.

31:26They're going to look, they're going to study all my stats. I don't think I want to play at that level or I don't want to play in that environment. And everyone would laugh at someone that did that because the problem is once you've started handing out options to your employees, once you've taken investors onto your cap chart, you're in a game where your job is to constantly increase price per share in one day to provide liquidity to those people. And so there's no way to opt out of that game. You're in the game. And the better you do, the harder it gets. And so it just keeps going up. but that's the obligation you have.

32:05Hey, Bill, did you guys wait too long to take Uber public? Would you have liked to see an Uber go public before, you know, growth started slowing and, you know, at a valuation less than whatever it was, 50 billion bucks? Possibly. And so, you know, we had some pretty big mistakes in leasing and stuff that might have been exposed and dealt with sooner. You know, as Jason has said on this and other podcasts, If you bring up Uber, I have to take a second to just really, really thank Dara for all he's done. Because the first job was to clean things up and put out the fires, which he totally crushed.

32:45And then the second job was to increase the value of the company, which is he's starting to deliver on that. And yeah, it makes me very happy. So there's no doubt, Bill, in your mind, we're on the topic of Uber, and this sort of speaks to conviction. When you see a billion rides going on and the company's losing a billion or two dollars, you never thought like this could never not be profitable because you knew customers would pay a little bit more money. But the press was kind of fixated on this. This can never be profitable. This can never be profitable. This business is a terrible business.

33:18But the same people who are saying that were using the product every day. And when the prices went up and the venture discount and the competition with Lyft started to recede, people had no problem paying an extra$5 a ride,$3 a ride. My belief, it was all a Zerp thing. Exactly. And you had massive amounts of capital. You had a clear winner in a network effect business, but the benefits of that winning were delayed because you had capital being thrown at the second place player in a way that we've never seen before in the history of venture. But we may be witnessing now in the LLM market. As Bill has described it, which I love, You know, capital was used as a weapon of economic destruction because the cost of capital was free.

34:11And the healthiest and best thing that's happening to the venture market is the fact that we have interest rates back at three and a half percent. Because you can't have number three, number four, number five, number six, raise capital to compete diseconomically against the market leader. And that allows the backers and the founders and the market leaders to emerge with profitability profiles much, much faster. Yeah, it was interesting. What a wild experience to see that amount of capital used that way. First time in history. And people looked at Amazon as the playbook for that bill, I think. They said, hey, Amazon was never profitable.

34:53They broke even, broke even. but they built this huge, you know, revenue base. The TAM was huge that, you know, look at all the prime subscribers, look at all the adjacent businesses like AWS, they were able to build an advertising business. So there is something to this. I don't want to say lose money at an alarming rate, but a break even and, and build the business for growth. And how do you balance that? Yeah. I mean, well, well, yeah. If you know, if you know, you have a network effect, like how far do you want to push it and how aggressive do you want to be? And Bezos, you go back to that moment in time between 01 and 04, there were multiple times where the press was convinced that Amazon was going bankrupt.

35:34And he did have people chasing him, especially in verticals, and he put them all under. Part of it was what we started the podcast talking about is just the cycle, and he had to adjust his game to the cycle. So in, in 70, 97, 98, 99, he had unlimited capital being thrown at him. And when that changed, um, he had, he did, he did cuts. Like he had to do cuts. He had to get to break even. Um, and eventually he did. Um, yeah, so, but look, he's one of the, there, there are, if you look at, at Bezos, Benioff and, uh, let's say Reed Hastings, they all push the edge of this like like trading off profitability for growth um and i think they all did in a way they were rewarded for but it it makes for a a a uh somewhat frightening ride for the people on the yacht it's like going up and down the pch at like 70 miles an hour you're just right along the edge you're like shouldn't there be a guardrail here or something no we don't need guardrails.

36:38We'll just, we'll build those later. If your landing page looks terrible, customers are just going to leave. They're going to bounce. It's 2024. There are no excuses for having an ugly website. So stop settling for okay or good enough and have an excellent, a beautiful and extraordinary website using Squarespace. It's out of the box. Beautiful website designs that will engage your audience, allow you to sell anything, whether it's content or an actual physical product. And you know all these amazing Squarespace features. You've heard me talk about it before, but they have the most amazing, gorgeous templates to get you off to the races instantly.

37:12And those templates, they're all designed and optimized for mobile. This is one of the things that makes me crazy. People will design a website on their desktop, but the consumers, 60, 70, 80, 90 % of them are experiencing it on their phones. You want to start mobile first. And these templates at Squarespace, they are gorgeous on mobile. Of course, they're gorgeous on desktop as well. And they have an amazing drag and drop web designer. And you get all kinds of analytics now. You don't have to get third-party tools to do analytics. Nope. Squarespace baked that all into their core product. You're going to get marketing analysis, sales data, and more directly inside of Squarespace, where you can create an online store, or start a blog, or create a subscription business for members-only content, and so much more.

37:50You do this all simultaneously on the same platform. It's the simplest, most effective, and best-looking way to start a business online. So here's your call to action. Check out squarespace.com slash twist for a free trial. And when you're ready to launch, go to squarespace.com slash twist for 10 % off your first purchase of a website or a domain. So moving on, given the reset in the market, what do we see in 2024 looking forward? Maybe Brad, what do you look at at the market today? And where will it be in 2024? Well, maybe just a level set. But Nick, do you have that chart of, so this is the state of Series B and Series C for software deals.

38:30So we just looked at PitchBook, all the software deals that were done. And as you see, we're down 90 % in 2023 in terms of B and C software deals from the peak in 21, in many ways back to the 2017 trend, which makes a lot of sense to me, given that we're back to 2017 level of interest rates, etc. etc. But you know, what this means is that we had in 2021, this explosion of early stage seed and series A software deals that were done. And the reality is they're not finding a home. And so now maybe I click in a little bit to altimeter. And I think there were a lot of private venture deals. And I think there are a lot of people who looked at that period of time.

39:13And frankly, prices had not adjust back to Bill's point. The market was delayed in the venture industry into getting its head around the fact that the world had totally reoriented around a different axis with respect to interest rates. And then, you know, if I just look at the list for altimeter, here's the deals. We did nine deals in Q3 and Q4 last year. We did four series A, two series B, and three pre IPO deals, okay, late stage deals. And I think that is reflective of the opening up the, you know, prices coming back down to planet Earth, the prices that we invested in are back to 2013 2014 level pricing, again, set aside the AI enclave that the bill's talking about.

40:00And so that's what I see happening in the world around us. Those were competitive deals with the Sequoias and Andreessen's and Red Points and all the other funds that folks know about. So I think the market is getting healthy because pricing is resetting. And Jason can comment on what's going on in the earlier stage market, but unless firms like ours are willing to do those Series B and Series C deals, then you've got a real problem of the pipeline of seed in Series A that can't find a home. And Brad, you look like a genius today, but at the time, you must have been getting a lot of pressure from LPs, a lot of people internally.

40:38How did you resist that pressure? And what was your operating principles on doing so few deals? Really, we didn't get a lot of pressure from our LPs. LPs were pretty pissed off at the cadence of activity in 2021. Venture firms were not getting the message. There were a lot of venture firms that were accelerator to the floor through Q2 of 22. Because remember, venture firms are not in the public market business. So it's easy just to say, oh, that's a public market problem. That's not our problem. The problem here, though, was that the melt up the balloon was created by a macro event, namely Zerp, and the deflation was created by Zerp.

41:20So I think because we have a foot in both markets, we were quicker to the set of concerns that caused us to hit pause. And, you know, I would also say that, you know, founders don't stop doing interesting things. In fact, I think the most interesting time to back founders are during the tough periods. And that's why you've seen these nine deals we've done over the course of the last couple quarters. it's you know these are some of the best series b and series a deals that we've done in a long time in terms of the quality of the team the tam the revenue profile of the businesses the product market fit the efficiency they all they all took the religion and got fit in fact every inbound that we get that it starts with we've you know we've read your letter time to get fit we're efficient you know like nobody's coming in you know burning insane amounts of money anymore and And I think it's that health restored to the market that's causing venture firms like Altimeter to be pretty foot forward and in market as we head into 2024.

42:24If you want something to put a lot of optimism around, I do believe it'd be an incredible time to go start a company if that's something you've been thinking about doing. It's just a much more rational environment in the whole ZERP environment. The number of competitors you would have would be dramatically lower. Your ability to hire people is obviously a ton easier than it was before. If you're having to want real estate, it's near free. We'll pay you to come to the office just to make it not a ghost town. And so, you know, it's an incredible time to start a company and benchmarking others are wide open for series A deals.

43:02So, you know, reach out to whoever you need to reach out to. But like, it's a fabulous time to start something. And I think history has shown that these windows of pessimism are a really good time to get going. And so I would highly encourage anyone who's kind of been debating starting something that this is actually a fabulous environment. Yeah. Just on the early stage to build on that, the biggest issue for the past 10 years that I hear from a founder is not raising money. Raising money was easy. Go to an accelerator. They're trying to fill seats at accelerators. You get your first 100k or maybe get 50k from your friends and family 100k from your accelerator raise a 500k to two million dollar seed round pass the hat Party round.

43:49It was all so easy raising money. But then once they had the money, they couldn't find a CTO. They couldn't find a VP of engineering. They couldn't find a sales executive to be their first sales executive. And when they did find those people, they said, well, I've got an Uber and Airbnb and a Google offer, and I'm going to try and get a Microsoft offer. And, you know, then they'd come back and say, well, I need 300, 400 ,000 in total comp. Now, this is a startup that's raised a million. So they're going to give 400 ,000 out of the gate. They can't match the Facebook offer. And now Facebook, Amazon, Uber, everybody's on a hiring freeze.

44:24And they're still doing, I believe, the quiet gentleman's riff where, you know, maybe the bottom 10 % of people are cut in a performance review on a rolling basis across the entire organization. What this means is now we see applications for Foundry University or pre-accelerator of two or three former Uber, Airbnb execs who are working on a side hustle and they're not incorporated yet. So the talent availability is incredible. It is incredible. and when you don't have comparatively yeah comparatively yeah when you don't have those it's also become international so there's a there's two very subtle things i want to point out that founders at the earliest stage i'm talking year zero founders where we invest there's two things they're doing that nobody's paying attention to one is they are so good at running remote teams that they are able to manage a developer in argentina india ukraine or in silicon valley or you know working at lake tahoe but demanding a google level salary but once you learn how to manage those people in a slack room on zoom using huddles whatever it is um you you don't need a silicon valley person anymore you don't even need an american and now they're telling me oh well i'm getting people in portugal for 60 000 a year as a developer i'm getting people in manila for my you know ea operations role sdr roles for 3k a month 2k a month and they know how to manage that so their burn rates are coming down i'll see people who have 12 employees and they're spending 50 000 a month total that's their spend not their burn bill yeah and i'm like how's that possible that's 4k a month per person like we have no office we got 100 000 in free credits from azure or 250 000 from azure and 100 000 from google and 50 000 from whatever amazon yeah and we're hiring people internationally uh and then the second piece is ai and the idea that you would hire somebody to write copy to make a logo to be a designer with co-pilots writing content faster with sdrs researching things everybody's getting 20 faster 30 faster at their job every quarter so now you see three four five person teams hitting a quarter a million in revenue, 10 person teams hitting a million in revenue.

46:44And this is what gives me massive hope, the efficiency that founders have, and they expect this round of funding to be their last. They when they raise that 500k or$1.5 million seed round, they're telling me, we're going to get to break even on this. We don't want to be in the series they race, we don't want to be dependent on venture capital, we just want to have seed investors get our 500k checks, and preserve the cap table. So I think this is going to be a revolution in funding. And this vintage, I think, will be the best vintage since the Uber vintage, the Airbnb vintage. Jason, does your selection criteria change in a post-AI, post-ZERP market versus the pre-ZERP, pre-AI market?

47:26Are you fundamentally looking for different teams, or is it just different valuations? It is a great question. I think when you're a venture capitalist, listen, I've gotten to sit at a poker table with Bill Gurley for over 20 years, and as a journalist, ask them questions and ask other folks, Ruloff, Michael Moritz. You know, I've literally had a group of unofficial mentors who I've been able to ask questions to. And that's helped me refine process. Because if you think about what a venture capitalist success is based upon, and I've really studied it, it's based on their brand. Bill Gurley's got a great brand.

47:56Brad Gershaw's got a great brand. He just said before, when people contact him, they first say, oh, we're fit. So what Brad did by writing that fitness letter was he put out a flag and said, hey, if you are a fit company, come talk to us. And so he's now he's got a deal flow that is different and differentiated from other venture firms. It's the fit people come to him, right? Or Bill Gurley has the people with network effects because oh, he did web van. Oh, he did Uber. Oh, he did, you know, this other network. Oh, he did eBay. They want people who understand networks. So your deal flow, and then your decision making, and then your ability to double down on an investment, I believe those are the three criteria for success in venture.

48:36Everybody's got different theories on this. And I'm an LP and 20 funds that aren't mine. And I look at those things, deal flow, decision making, doubling down. On a decision making basis, I was having this talk with Ruloff two weeks ago, we had lunch and he said, as the number, as your deal flow goes up, you're, you can become more discerning, right? You can, you can expect more. So now when people come to me with a deal, that's 30 times revenue, and there's no technical person, and it's a solo founder. And you know, they, the customer churn is too high. Well, we have other competing deals that have three founders, and two of them are technical.

49:11And they have a lighthouse customer, and they're asking for 12 times revenue. So you can actually skim the cream as it were. And I think when I started my career, it was just investing in my network. But then over time, I sure deal flow goes up, we have 20 ,000 applications for funding a year now. And we invest in 100 new companies per year. So 50 bps, you know, one in 200. And so I've really thought about that question a lot. And basically, more deal flow equals a higher benchmark, so to speak, for investing. What a David, one of the things that Jason reference, and I'd be curious Bill's point on this, because I don't think we've ever seen this in our 20 some years, Bill here, that But if you look at the expected, the consensus headcount growth in engineering in 2024 for Meta, for Google, for Uber, etc., it's flat to negative for these companies.

50:10And this is coming on the heels of a year where you already saw negative headcount growth for Google, for Meta, etc., as they tighten their belts. And this is because of their teams getting to 20 to 30 % more productive out of the gates with Copilot and their view that that will continue for years. So this point to Jason, if the finite resource in Silicon Valley is engineering and technical talent, if that is the most important resource, that's the resource that you couldn't find and you couldn't afford during ZERP. It's not just that Zerp's ending. That would have in and of itself created a much better environment to Bill's point to start a company.

50:54It's that Zerp is ending at precisely the same time that we have this inflection in AI-driven co-pilots, which is causing the biggest companies in Silicon Valley that sucked up all of the talent, that set the price level for all the talent. They're no longer hiring those folks. And so you have two major factors contributing to this abundance in terms of technical talent at this moment in Silicon Valley that I think is unique, certainly compared to past cycles. yeah i agree with that and i think that's one of the reasons one of the many reasons why it'd be a great time to start a company going back to the people availability i do i do push back a little bit on this notion that oh you'll always be able to run a company with only four people what i've seen in the past is we evolve with our tools there was a great a great story about bjorn borg trying to make a comeback and he insisted on using the wooden racket even though everyone had moved to the graphite racket and he got his ass kicked and like once once we get the benefit of the graphite racket like once everyone has it then you're just back to where you were right and so then you're then you're competing on all the same terms so i i don't think there's i don't think there's an infinite lift in productivity because it eventually gets whittled away by competition and and if you don't use it you die so that that's true the thing that i'm curious uh bill your take on um if it is as things got cheaper to start companies the outcomes became much more dramatic in venture so cloud computing was i think maybe the biggest one where you know and we work um you know getting rid of that quarter million dollar check that you needed to give to a landlord and getting rid of the AWS getting rid of the quarter million or half million dollar check you had to give to, you know, a co-location facility and having to hire a chief security officer and a bunch of now with AI, it does feel to me like, you know, three people are going to be able to get companies to 500k a million in revenue.

53:03You didn't see anything like that in the first part of your career, the road to a million in revenue. What did that take back in the day in terms of investment to get that first million in revenue. Yeah, I think you're absolutely right, Jason. I think there's a flip side to that though, that is, if you want to build a company that's going to do a billion in revenue one day, you're eventually going to have to hire people. You're going to have to spend money. So maybe there's more opportunity for angel-backed companies to never do venture rounds and sell at$20 to$50 million. And that could be a huge win for those founders and their early investors.

53:47And all that gets into people understanding whether the opportunity they have in front of them is something that, how much potential is really there in that opportunity and how much value can be created and how much runway do you have, run room, how big can it be? and then adjusting your capital strategy to that opportunity. I don't think, you know, the other thing I would add to prove that point is if you happen to be on something that is a really big opportunity, there are going to be two or three other startups. And if you're the one sitting there saying, I'm only going to hire five people and someone else steps up against it, Bezos style, then you'll get run over.

54:29Yeah. I mean, there were competitors to Uber before Uber, right? We had Sidecar, I think, came out before it and Taxi Magic. There were a couple of other folks that you had actually met with them. Yeah. Yeah. And they didn't have the ambition of our guy. Absolutely. So you got to get all that figured out. So let's talk a little bit about AI investments and AI in 2023. In 2023, there was a total of$87. billion deployed by VCs into AI. A lot of that went to a small handful of companies. You had OpenAI that raised$13 billion. Anthropic raised$4 billion from Amazon and apparently is at a run rate of$850 million, as reported by the information.

55:17InflectionAI raised$1.3 billion. Just those three deals made up over 20 % of all VC AI funding. So the question is, will 2024, will we see the same amalgamation of capital among the incumbent startups? Or are we going to see more of a distributed funding of new AI startups? Hey, David, I want to clarify one thing because I think this happens quite a bit. I want you to go back and find that information article. And I think what you'll see is they claim that they would hit$850 million in 2024. So their actual run rate is$100 million. I end up here, yeah. And this is a classic little trick that people play.

55:56Do you mind if I record that? No, no, I think it's actually good. No, no, I think it's a good learning moment here. I feel like getting in there because people hear these numbers and they start saying they're real, even though that was just a forecast that they had put out. No, I think it's the nuances is super important. And I've had founders take last weeks. Let's say they had a record week. They take that week and they times it by 52. and they're like you know like but but that's because like some sale came in of a quarter million dollars and they're like oh a quarter million dollars times 52 is this amount so that's our run rate look this stuff's unbelievable like like the you you can't sit here and say that uh that it's all a hoax that it's all a scam that it's all like like there there is real revenue coming in at very high volumes i mean you can see it in the nvidia sales um and and so it's It's happening and it's interesting and people care.

56:52And that's why this is an exception and it's a mania and it's still acting like Zerg. It could be that gap revenue is one of the only regulatory things that are positive in today's market. Yeah, I mean, and there's some interesting elements of AI in that. And that because these big companies have, at least today, I'll say gotten away with giving away credits for equity, these companies now have credits. Most all of them want to be the revenue of records. So you don't pay them separately from the hosting service. You pay them with the hosting embedded. And this creates massive opportunity for gamifying revenue.

57:40because you're reselling hosting services. So if you lower the price enough, you're definitely going to create revenue. And so it's a world that could get sloppy because of everything that's happening and all the components. And when I say that, it's really important to say, I know there's real stuff happening too. So it's a manic world like it was in 01 for everything else. And so you're going to see a lot of really good positive stuff and you're going to see a lot of crazy manic stuff simultaneous. And that's really the lesson, you know, that I would apply from that 99 to 2001 period. It's very difficult as an analyst, and we're all analysts.

58:24We study the world around us. We study the world of startups. We try to make sense. We try to predict the future. We try to make, you know, an asymmetric bet in our favor. At this moment, you have to hold two simultaneous truths, right? One, this thing, in 99, it was the internet. We knew it was going to change everything in our world forever. And I think now we would probably say, there's probably consensus on this panel, that AI may very well be bigger than the internet itself, bigger than mobile, bigger than cloud computing. We'll be investing against it for the next several decades of our investment career.

59:00But the simultaneous truth you have to hold is that there's a lot of stuff going on that is radically mispriced, because there's no ability to forecast durability of revenue, whether LLMs are going to be important, getting smaller, getting bigger, etc. And the prices that are being paid are applying a discount to them that almost assumes certainty, like a certainty of the forward view. And so if you believe that AI is as big as I believe, and I think everybody else here believes, it's going to play out over decades, right, not play out over days or weeks or months. And so you have two camps that have assembled, you have the dogmatic camp that thinks it's all a fad, everything's radically overvalued, you shouldn't do anything.

59:45And you have a dogmatic camp that says, you know, AI everything. And the truth is, you've got to be in the game, you've got to be studying, you've got to be understanding what's working, what's not working, what are the end use cases that are driving productivity for enterprises? How is this going to change search forever? You know, what does this mean for Google, all of those questions. And then as Warren Buffett, you know, has famously said, the hardest thing is to do all that work and then do nothing. To wait, to let the next card be turned, to understand a little bit more, right, whether or not you truly have asymmetric risk reward.

1:00:21And I think that's kind of this moment we're in. Every venture firm in Silicon Valley in 1997, 98, 99 was scrambling to get a search logo. go altavista excite infoseek uh you know uh go go through the list how did that work out lycos etc they all thought they were smart and guess what they got most of the bet right they knew the internet was going to be huge they knew search was going to be the gatekeeper to the internet it was going to be the tax collector and that would absorb a lot of the the profits right but you would have been better off not placing any bets and investing in the ipo of Google in 2004, and you had captured 98 % of all the profits ever generated by internet search.

1:01:02And so, you know, being too early is tantamount to being wrong sometimes. And I think this is one of those moments, whereas a firm, we're just trying to study work hard, but be cautious. I think what you just described with these language models, and it was very interesting when these language models got funded, they skipped the normal startup path, they just went right to late stage funding, which, you know, I think is a function of the deep mind people were so rich, these AI folks were already making three, four, five,$10 million a year with packages and places. So the money instantly came in in nine figures for these language models.

1:01:38I think the language models could be like us excite in that, you know, there it may wind out there's just one of them that just is superior to all the ones and is most updated and who knows who's that's going going to be or it might just be that there's 20 of them like publishing platforms like WordPress and other publishing platforms on the web. And the whole angel community didn't get to participate in language models. And then what I see with startups is they're taking the same prompts, the same code, and they spit it out to five language models. And whoever and they're looking at it, and they're telling me like, yeah, chat GPT four is a little bit better, but Claude is good.

1:02:15This is good. So I think they're going to quickly get commoditized. And I don't I think they're going to be indistinguishable for 90 of jobs and if they're all indistinguishable for 90 of jobs it's a commodity business it might be like storage it might be like compute does anybody care where they're storing their images or videos you know what do you think bill and the and the open source models are a particularly disruptive play here um launched by meta and then mistraw has really good numbers out of europe and and you know to the extent that amazon and and others can host those without fees like it would be very very um i i personally based on everything i've studied in conversations with with our friend sonny um i do think that the the variables that have been used to drive the non-linear growth both in the parameter count and in the um and in the uh the window where they where they study relevance um i think those are probably topped out and they're trying to differentiate based on merging multiple models but the user experience on that is janky and slower and so i i do think there's a chance that we're up against a on the llms up against a bit of a of a um ceiling and we'll see though and and and i think that's why the startups that you're seeing are finding that there's not much difference because they're all kind of heading towards the same ceiling?

1:03:45Well, I will say coming out of last summer, we already started to see a cooling in terms of valuations and up rounds for these AI startups. I think that a lot of firms are coming to that same conclusion. I don't think it's quite as euphoric as it was in Q1 and Q2 after ChatGPT came out and there was this mad scramble. I think everybody's trying to discern what is real. And remember, David, the valuations and the companies that you pointed out, almost, I think, in every one of those cases, it was set by a non-financial investor. If Amazon or Google or Microsoft or one of these companies is marking or leading one of these investment rounds, you should significantly discount the importance of that valuation.

1:04:30because while there may be some financial investors who jump on that bandwagon, they're not setting, you know, that price in the market. And so, you know, I think that we've seen a relative cooling. I think LLMs has been where what consumed a lot of the action. We're starting to see applications emerge. I'm personally very excited, as Jason and Bill know, about the impact this is going to have on Google and what's coming next for the consumer. And so whether it's the really interesting work that Humane's doing, humane.ai or whether it's the work that's going on at Apple. I think Apple's going to have a large language model and a full new Siri come, you know, Q2, Q3 this year.

1:05:11You know, Llama 3 is coming in Q2 of this year. Zuck's going to be in that game. Microsoft wants to be back in the consumer game. What's the best case, Brad, for meta being a leader in AI? Bill just sort of mentioned, hey, they did this disruptive thing. They're behind. They know they're behind. So they did the open source model, they may have even leaked some of it. What's the best case for meta? Is it that their products are just that much better and their feeds are more addicting? Or is there a new product that comes out of meta where they just heads on compete with chat GPT with an app with a co pilot?

1:05:47What's the best case scenario for meta in AI? Yeah, so I would, I would direct you in three, three directions. Number one is reducing friction from existing products to make them way more pleasurable to the to the end user, you know, thrill customers. So, you know, if you talk to people today, the number one source of apparel leads for anybody in the apparel business is Instagram. But it's still a lot of friction, I'll hover over something, and then I have to go to a sub site, I have to fill out all the information, etc. Instead, there's just going to be an agent that says, Hey, do you want to buy this?

1:06:19I'll take care of it for you. Right? So it just tells us Brad's wearing this shirt right now, right? Click here to buy it. Shopping agents are coming in 2024. that I think they'll radically unlock a whole new level of commerce and activity that occurs on Instagram. Number two, WhatsApp. Remember, the interaction model for so much of this is chat. We see startups in South America that have 20 million MAUs that are bots built on top of WhatsApp as a platform. So expect in 2024, you're going to have vertical agents and horizontal agents that are built on top of WhatsApp. This is a platform that has two and a half billion And DAUs, right, in India and South America and other parts of the world, this will be the platform of choice for launching your agent into the world.

1:07:06And then, of course, you saw, Mark, last year experiment with things like meta AI and characters, you know, expect a lot more on that. I would say this company, from the time, you know, the fall of 22, the time to get fit letter, it's extraordinary to watch the turnaround. mark singularly has doubled down on ai the whole company is focused on ai um and i think there's going to be a tremendous amount of innovation and i would say one of the if you said what's a surprise prediction over the course of the next two years um remember everybody got out of the hardware game right yeah apple won the game siri was your agent i think this opens the door for everybody to get back in the hardware game ray-ban glasses i think when meta reports in q uh in january their q4 returns meta meta glasses i have everybody in my office is wearing them they've been a blockbuster hit this season killer use case is listening to music but you're going to charm you're going to see a lot more innovation coming whether you have devices that are in your pocket like humane uh that are going to connect you because it allows you to bypass the operating system.

1:08:17It allows you to bypass iOS. And Apple knows this, which is why they're coming with their heater in Q3, Q4 this year. And you're going to have, I think they're going to scrap Siri. I think they're going to go to a fully new LLM enabled AI that's going to drive your search experience. They're sitting in pull position. They ought to be able to win. But I think you're going to see Microsoft and everybody else back in the game. Gurley, what do you think in terms of those Magnificent Seven type companies? Google with their blue links is the franchise at risk with just getting an answer. Meta with the glasses.

1:08:50Apple may be throwing a Hail Mary and Siri 2.0 actually working. What do you think of those major players? As Brad hinted at, I do think voice recognition is a key part of all this. A lot of the major disruptions out there that have happened over the past 30 years relate to the UI that the customer is using. when we went from mainframes to PCs, to browsers, to mobile phones, like they all changed the UI. And it's hard for me to imagine AI being great without voice recognition. Like the efficiency of talking to something is so much better than having to type, you know? And, and we also got excited about, uh, um, like chatting as a potential UI and there were actually startups created like four or five years ago and it never happened i just don't know that people want to hype i think if you could get the talking right so i do i think kind of combining a great llm with great voice recognition is a is a really important milestone such a great insight because if you think we've we've been trying to have this voice interface for 20 years right bill dragon dictate you probably looked at some of these companies back in the day and i my perception was the payoff from speaking was very low.

1:10:13Siri, lower the music. Siri, pause the music. It's not enough of a payoff for me not to just click the volume up and down button. But if you say, get me a reservation at one of the five best sushi restaurants within 20 blocks of me while I'm in Manhattan and I want it in the next hour, and it actually does that, that's a major lift. That's a major payoff. So I'm willing to do that. And I found myself when I was skiing this time, I answered and reply to Siri can do this where you can like reply or you know, and I find myself doing that. And then on the iPhone 15, I don't know if you gentlemen have it.

1:10:53There's a thing called the action button. Do you know about this? The action button, the action button is just a button that you can program to go directly to an app. I set my action button when I press it, it does a shortcut. And my shortcut is the chat GPT-4 interface. Voice interface. Voice interface. So I'm driving in my car. I got my Tesla on self-driving. I got this and I just press the button. And then I'm having a conversation with chat GPT-4 about history. You know, and I was with my daughter and we were having a conversation. She wanted to know the history of a certain war and a certain people.

1:11:28And it was unbelievable. It was like having a tutor in the car with me. And so I agree with you. I think this could be the big breakout. By the way, one cool kind of tactical minor example of this is the Roku remote. It sounds like you had a great experience using chat. Not now. Not now. Oh, God. Now people know what you're really connected to, Jacob. Both on the mobile app and on that small remote that Roku has, there's a microphone button. And you can say, fast forward 10 minutes or turn the volume up or turn the captions on or turn the captions off. And it's awesome. Well, again, the payoff to put captions on or off is like five or six steps, right?

1:12:12And so when the payoff is greater and AI kind of makes the payoff greater, then it's worth doing it, right? The juice, the squeeze is worth the juice. The juice is worth the squeeze. There's a second part to get to the reality you talked about. So I think this first hurdle is can you get voice recognition great and combine it with an LLM? The second part is equally difficult, which is can you build an infrastructure for transactional, you know, things in a digital world that you described, Jason? And I think it's harder than people realize. I think Google struggled over the years with can they do deals with, you know, the aggregators in a way that's mutually beneficial.

1:12:52And for the most part, there's just been a ton of friction in trying to make that happen. And if you don't do that, what's an example of that? Like Expedia or something? Like trying to work with Expedia or booking, you know, dot com, because there's an ad relationship and Google always wanted there to be a jump ball competition between the players. But the user that that example you just shared, you don't want it to come back and say, do you want me to get a quote from booking and a quote from Expedia and a quote from like you don't want that. overhead just give me my reservation yeah yeah so how do you make that happen because if you can't work out a deal with an aggregator that's mutually beneficial that's hidden under the covers now you have to go out to each individual small business to make this all work and that's nearly impossible also yeah that's a multi-decade process well this this this one though i mean i you know you got me in a place of passion bill this this one is the one that i think is going to get cracked.

1:13:50This isn't Jason, a 10x moment. This is 100x moment. This is when we talk about having a personal assistant in each of our pockets, personal assistants don't just tell you, you know, some information about World War Two, they actually do that makes your life easier, better, they book the restaurant, they book the hotel, they book the airline ticket. I've talked to lots of friends at Meta, at Google, at Apple, etc. They would all echo Bill's point, It's harder to do than you think. Keparthi, in fact, went to open AI, I think in 2018, tried to do this, worked on a project called World of Bits and said it was much harder than you think, but he thinks now is the time.

1:14:28The timeline they give me is that, you know, you have the rebirth of these models on devices like Apple this summer, but it doesn't get you to action. The action bot is sometime in the 18 to 24 month time horizon. I will say that Apple and Google have a huge advantage in this regard. Because if you think about the app ecosystem, they have payments on both sides, they already have a relationship in the app store with booking.com with Expedia, etc. And so they're able to connect that plumbing up in a way that may be harder for meta to do maybe harder for open AI to do. But that's that's the work that's going on now.

1:15:03Sounds to me like an argument, Brad, that the incumbents, OpenTable, Yelp, you know, united has a great app now or emirates when you and i go to the middle east we take emirates typically they're all starting to have good apps it might just be you to just ask siri to open open table and that open table gets you the reservation yeah i i think for certain my my i've thought about this i think for something that's a very important transaction in your life like you know real estate or something you might have a a separate ai relationship with with someone for the more tactical stuff, you know, I don't think that's going to work.

1:15:43I think it violates the premise of what you said. And the other thing I would highlight that I think makes this sense. The premise being simplicity. Yeah, yeah, yeah, yeah. You just want it to happen. One click and done. The other problem I think all of the incumbents have with this second part of facilitating this transaction, they've all had monopoly-like power and they've all tended to treat partners in a win-lose way. And I would put that on nearly every one of them, Google, Facebook, Amazon. They don't play nicely. Apple. They don't. They don't. They haven't built win-win relationships with partners.

1:16:25So that's an opportunity for somebody? No, they just, they always, well, Google in particular was born, were putting these people up against each other and actually led to revenue growth and all that. But yeah, they've had so much market power that they've created cultures where the attitude is, we win, you lose. Squeeze the partner to death. And Facebook has done that. Name the companies that have been built on the backs of these companies. Facebook, I remember Mark Pink is telling me that his personal relationship with Zuckerberg would allow Zynga to flourish in the Facebook ecosystem. And that's where all their apps would work.

1:17:04And then he woke up one morning. and zuckerberg's slit his throat so like so i just think it like whoever figures it out i think they're going to have to have a bit of a mindset shift because if they could work out a win-win deal with one of the leading partners it would allow them to knock off some of these verticals seems like a good possibility there if they just let each if they just let everybody do it for free do it for free and just make the 20 bucks a month on the pro version that that seems to be the best can do it for free this is a huge moment of potential massive disruption to these monopoly models you know i've said google is the greatest monopoly in the history of capitalism they're in the best position but they also face the biggest innovators dilemma it's almost impossible to me to see how they go from the age of 10 blue links to the age of answers and actions and maintain the same monopoly profits they'll be in the pack of competitors but to maintain the the monopoly profits you have to assume that everybody else in the game possible you know it's impossible on a per user on a per visit basis is impossible it truly appears to be impossible i would say that maybe another solution is that the the the leading llm lets the consumer choose which of these services companies they want to work with and then they have open you know connections to each of them and then it uses what you have selected.

1:18:30That's another way. That would be wonderful. I mean, just trying to get another browser to work on your iPhone as the default is just so much lift and so hard. Like when you ask Siri for music, it just automatically goes to Apple Music. I want to use Spotify. It doesn't make it easy. They kind of put these hurdles up and Siri doesn't go deep into Spotify. Well, this gets into my point. Yeah, and then there's other, there's some of these potential players have vertical services that might compete with it so anyway that's it's it there's a lot to have that has to happen to get it right it's super fun to watch so but when you have a fiduciary working in the background for your b on your behalf right it's not going to be an auction that my my meta will simply say hey brad do you uh you know do you prefer booking or expedient i'll say you know booking um or expedient they'll ask they'll say do you prefer uber or lyft and i'll say uber they'll say, do you have an Uber number?

1:19:24And I'll say, yeah, I'll text it to you in a minute. I'll text the Uber number to him of my frequent flyer or my loyalty on Uber. It's going to be this ambient thing that's really seamless that occurs in the background. They asked me once, just like my personal assistants. And once it's done longitudinally, it gets better and better and better every single day. It's assuming the leading front end LLM doesn't say, well, I'll only do that deal if uber or expedia whoever bends over and gives me crazy economics the reason i don't think that happens is because this is no longer a you know google you know determining the terms we're going to have seven or eight people and by the way for meta who gets zero money in the travel vertical why wouldn't they do it for free why wouldn't chat gpt do it for free they disrupt the hell out of their competitors they deliver a more thrilling consumer experience your margin Customers.

1:20:19It's my opportunity. Right? I mean, this is. I hope that's what happens. I really do. I've just, I've lived through the past 15 years of these companies not respecting the vertical marketplaces. So, it'll be interesting to see. So, I'm going to force a decision on you guys. You've been talking about a lot of the Magnificent Seven, Alphabet, Amazon, Apple, Meta, Microsoft. We didn't talk as much about NVIDIA, but obviously NVIDIA is a very significant player. And, of course, Tesla. If you had to choose one prediction for 2024 in terms of the best performing on a relative basis and a worst performing, what would your choice be?

1:20:56We'll start with Brad. Well, first, I would say that I think 2024 and 2025 will again be two more years that tech oriented companies that are represented by the Magnificent Seven will outperform non-tech. So if you look at the performance of the NASDAQ versus the SPY or the S &PX technology, I think you're going to continue to see meaningful outperformance because of all the benefits of AI that we've just talked about. You've also heard us say that all seven of the incumbents generally benefit from what's going on here. All their cloud businesses get bigger. All this AI stack has to get built.

1:21:40But if you put a gun to my head today, I would say the non consensus pick I would have on the long side would be Nvidia. You know, it's trading low 20s PE multiple. I think the demand is going to continue to outstrip their supply for a long time. I think the runway is still very long and very wide. And there's a wall of worry about Nvidia. A lot of people like Chema think that they pulled forward all the demand for the next two or three years and, you know, that this is like dark fiber in 2001. And so that wall of worry causes the multiple to have compressed pretty dramatically over the course of last year.

1:22:15I think they will continue to surprise to the upside. And then on the downside, again, I think I would pick Google. I think Google will still be positively returning in 2024. But I just think it's very challenging for their multiple to expand in a world where 10 blue links are clearly dying. I don't know how long that timeline will be. It's not going away overnight. Um, they've got an incredible team over there, but I think it's very challenging. The only way I think I'm wrong on Google is if Ruth and, and, and, and team get really aggressive on the cost side, they were probably the least aggressive in terms of getting fit in 2023.

1:22:53And I think there's massive opportunities for them, uh, in the future to, to tighten their belt. Yeah. I, I, I, you know, I'm going to choose a biased answer on the long side and, and, and go with Uber. I think that the new CFO brings a mindset on cost management that the company hasn't had. And we saw the powerful stock performance that Meta was able to deliver when that kind of attitude was kind of brought into the company. And I'm hopeful that that can happen here. And it's coincidental or simultaneous with um the core business just having you know a lot of wind at its back um with with with lifts inability to kind of fund losses forever off the table um it's just a really powerful dynamic for the company um and then on i i would probably echo the google comment on the on the other side when when the world is worried about disruptives to your business it's just really hard to have any multiple expansion whatsoever and so you you find stocks that look cheap on a pe basis but they just can't get anywhere and jason you get the final word well i'm yeah i'm obviously long uber uh as well and and i do think yeah they're more financial management is a great thing they could probably do what they're doing with half the number of employees um and i'm not advocating for half people to lose their job but the point is what we saw at Twitter, X, and what Elon did there, it is probably the case that every tech company could cut half their staff and still perform at a high level, and they might even function better.

1:24:40And we saw that when Google cut and Facebook cut 10, 20 ,000 people each. So I'm super long Uber, but when we look at the Magnificent Seven, I think Apple has some fundamental problems in that. And I like anecdotes about consumer purchasing behavior, because I do think that they can be super insightful, especially in a company that's launching a lot of products and has a lot of buzzwords, etc. I skipped for the first time, iPhone 14, I skipped a generation of iPhone. Now I was the person who would send somebody to line up and pick up a phone for me. And I would always buy the interim. So between, you know, iPhone 5 and 6 you know, they would do the s or something and I would buy that too So I would literally just trade in my iPhone every year for the newest one because it really had it felt like such a profound change now I skipped I went from 13 to 15 and my reason for skipping was Not financial.

1:25:34It was just I didn't want to go through the process of unboxing it There wasn't literally there wasn't enough Juice from that squeeze of unboxing and setting up the phone that one hour for me to buy it and upgrade it And then I talked to my family this Christmas. And some family members are skipping three or four generations are going from 10 to 15, 11 to 15. And you're starting to see that in the revenue numbers. And I don't think the glasses, the goggles are going to make up for it. I don't think they have enough new products. And I think it's like the Steve Jobs product roadmap has been exhausted.

1:26:07And I don't know that goggles gets them there or services gets them there. So I think Apple's got a lot of headwinds. Google, I'm actually bullish on. I think that the franchise of search is going to take a long time to deprecate and I think people are going to do more searches and they're going to do more queries because it's going to be so fun to do queries. So I think that the lift from doing more queries will will out trip the disruption and what they have in YouTube as a base for a language model and what they have in your Gmail and what they have from the browser and what they have from Android.

1:26:39Remember, they have like 5, 6, 7 billion user franchises. Chrome, Android, YouTube, Search, obviously, Docs. What am I missing in there? Billion franchise. So I feel like they're going to get a lot out of that data. So I don't think they're out of the game. All right. 138.83. You can write that down, Dave. We'll watch it. By the way, one area where Apple has momentum, I think, is Apple Pay. Oh, yes. Really on fire. On fire. So good. Yeah. Both online and on. You know what? It's interesting. I was skiing and because you can't get service people in ski towns now, right? It's just like a major problem.

1:27:23Any tourist town, because of Airbnb, et cetera, people can't live there. People can't drive to Lake Tahoe to work. What they did is they've replaced like with a toast kind of system, whatever the ability to order from your seat at the lodges and Because of Apple pay, you know the idea of typing in your credit card is just Why would I do that if my watch does Apple pay or my phone does it instantly and I'm starting to see Apple pay come up for Like clothes purchases like I didn't think it was gonna get there so quick It's not just like getting yogurt, you know frozen yogurt. It's it's happening everywhere So yeah, I agree with you.

1:28:00That's a really interesting business. That's your negative pick for the year, right? I think Apple is a negative. And I think I'm not long necessarily Google. I'm trying to figure out who I'm long in that group. What are you talking about? You chose Apple as your short and you chose Google as your long. I heard it. No, no. I'm just that I'm not as negative on Google. We were restricted to the Magnificent. I'm not as negative on Google. I think I'm trying to think. Oh, come on. I think if we're betting on stock prices, you probably would go with NVIDIA because of that reason. So I would have to pick my long being NVIDIA and Apple being my short.

1:28:32So he's already changed for me, Bill. No, no. I'm just saying I think you guys are the wall of worry of Google. And I think it's a little overblown. Okay. I think they have opportunities. Anyway, this has been an amazing episode. David, great job moderating reading the news. Bill Gurley, awesome. And Brad Gerstner, amazing. We'll see you all next time on this week's service. Bye-bye.