Ep10. Pre-IPO Market, AI Hype Cycle, Is Software Dead? | BG2 with Bill Gurley & Brad Gerstner

00:00:00.000 | there is absolutely no doubt in my mind that excess capital

00:00:04.640 | distorts company behavior, okay?

00:00:08.180 | And in particularly these early phases,

00:00:11.140 | it's very difficult, it's not impossible,

00:00:13.560 | but it's very difficult to stay fit and efficient

00:00:17.480 | when you have a buffet of all options

00:00:20.740 | sitting in front of you and you can fund all of them.

00:00:23.240 | Scarcity breeds necessity, scarcity breeds innovation.

00:00:27.740 | So I think if you're on the board of a company

00:00:29.400 | or a founder of a company or CEO of a company,

00:00:31.620 | you have to think long and hard about the negative cultural

00:00:35.420 | and negative fundamental effects to your business

00:00:38.540 | of taking too much capital.

00:00:39.920 | (upbeat music)

00:00:42.500 | - Hey man, good to see you, happy Saturday pod.

00:00:55.060 | - Good to see you, sir.

00:00:56.440 | - What's been on your mind?

00:00:57.440 | - Well, look, I mean, so much has happened

00:00:59.400 | and I think we took a few weeks off.

00:01:01.160 | So like, it almost feels like the world

00:01:05.280 | has changed dramatically from when we last talked.

00:01:08.580 | So there's a bunch of different things on my mind.

00:01:10.400 | I've been thinking a lot about the,

00:01:12.560 | I'll call it the pre-IPO market.

00:01:15.280 | I don't know why I don't wanna call it late stage

00:01:18.440 | 'cause I think some of the money has moved down

00:01:20.040 | to companies that are even pre-revenue,

00:01:21.840 | but the pre-IPO market I think is changing dramatically.

00:01:26.840 | I think there's always this interesting question

00:01:29.320 | of where are we in terms of reality and perception

00:01:34.320 | on whatever the newest trend is.

00:01:40.000 | And we've seen that with other things

00:01:41.760 | like crypto and whatnot.

00:01:42.960 | And I think it's interesting to think about that

00:01:45.280 | relative to AI.

00:01:46.360 | And then lastly, one topic I really wanna get

00:01:49.080 | your opinion on, after some of the earnings releases

00:01:54.040 | since we last talked in the software category

00:01:56.400 | particularly, people have thrown out this question,

00:02:00.560 | like is software as we know it dead,

00:02:03.400 | which is obviously overly provocative.

00:02:06.760 | But there are a lot of questions around that

00:02:09.520 | and a few people weighed in and I'd love,

00:02:11.840 | so anyway, I'd love to talk about all three

00:02:13.440 | of those things if you're up for it.

00:02:15.480 | Well, no doubt about it.

00:02:16.520 | And it sounds like we have our agenda,

00:02:18.400 | but before we jump in, I have to say,

00:02:21.200 | I was driving in this morning and I don't know,

00:02:24.800 | I'm just reflected in a reflective place.

00:02:27.000 | I mean, my oldest son, Lincoln turned 16 this week

00:02:30.600 | and my mother turned 88 and it's just a moment,

00:02:33.400 | just a moment.

00:02:34.240 | My mom sent me this note.

00:02:36.100 | Well, she actually sent it to all the kids

00:02:38.800 | and I just, I have to read a bit of it

00:02:41.520 | because it goes to how I'm thinking about things

00:02:44.000 | and she's 88, right?

00:02:45.520 | So she starts by saying, I was born on this day in 1936,

00:02:50.080 | right between the Great Depression and World War II.

00:02:53.760 | Times were tough, we didn't have much,

00:02:55.800 | but we shared what little we had.

00:02:58.120 | Does that sound bad?

00:02:59.060 | Nope, these have been the best 88 years possible

00:03:02.360 | to be alive.

00:03:03.720 | Sure, I've seen lots of bad stuff, assassinations,

00:03:07.440 | bad politicians and wars, lots of wars,

00:03:11.040 | but the good outweighs the bad by a long shot.

00:03:13.840 | Just think about what I've seen.

00:03:16.000 | The first TVs, the first commercial flights,

00:03:19.060 | the first computers, the first mobile phones and Google

00:03:23.320 | and now even chat GPT.

00:03:25.260 | The world's gotten better in almost every way.

00:03:28.320 | We live longer and healthier.

00:03:30.460 | We solve problems we could never have dreamed of

00:03:33.340 | and we're all more connected.

00:03:35.040 | I wish people weren't so negative.

00:03:37.620 | Let me tell you the struggles of no food or housing

00:03:40.480 | in the Great Depression or being shipped off to war at 18,

00:03:44.400 | knowing you wouldn't come home, that is hardship.

00:03:47.680 | But the truth is I understand people feeling overwhelmed

00:03:51.640 | by all the hustle and bustle today.

00:03:54.000 | I too have a love-hate relationship with technology myself.

00:03:58.480 | While I know it's made the world better,

00:03:59.960 | I didn't realize how lucky I was to grow up

00:04:02.200 | in small town America at a quieter time.

00:04:05.280 | It was a simple life, long walks and card games,

00:04:09.400 | fewer friends, but you shared everything.

00:04:12.460 | So enjoy your technology,

00:04:13.960 | but never let it rob you of the personal touch.

00:04:16.360 | It's a tool to use.

00:04:18.280 | You should run it, don't let it run you.

00:04:21.480 | But never turn against progress

00:04:23.240 | because without it, we go backwards.

00:04:25.080 | And personally, I can't wait to see what comes next.

00:04:27.880 | That from like an 88-year-old woman.

00:04:30.200 | I mean, I was so blown away by just that perspective,

00:04:36.240 | her optimism and just like the framing of her life.

00:04:41.400 | It really was like all the innovation we've seen,

00:04:45.480 | like it compressed into such a short period of time.

00:04:49.040 | And so I know today we're gonna talk a bunch

00:04:52.280 | about where markets are going and is software dead

00:04:54.600 | and where are we in the age of AI?

00:04:56.520 | But it just caused me to think we live

00:04:58.680 | at such a unique moment in human history.

00:05:01.960 | I mean, I think you would agree,

00:05:04.160 | just innovation of all sorts,

00:05:06.960 | it's raging at a pace that you and I have never seen.

00:05:11.800 | And it's not an accident.

00:05:13.680 | Like there's been all this attack on capitalism

00:05:16.240 | and free enterprise over the last few years,

00:05:19.560 | but it is in fact that system,

00:05:23.120 | that and a direct result of that system

00:05:25.840 | that inspires and incentivizes people to dream

00:05:29.560 | and to build this stuff.

00:05:30.560 | And I was looking at the Starship launch this week,

00:05:34.480 | which is just this insane feat of human engineering.

00:05:37.320 | I mean, in four launches, right?

00:05:40.000 | We now had a soft landing in the ocean.

00:05:42.720 | This is a ship that nobody thought was ever achievable.

00:05:46.480 | It's gonna make us a multi-planetary species.

00:05:49.440 | And I saw the looks of exhilaration

00:05:52.480 | on the faces of those young people in the control center.

00:05:57.080 | And I just told my kids when we were watching that,

00:05:59.400 | I just said, find something in life

00:06:01.640 | that makes you feel that way, right?

00:06:04.200 | And I think that may be Elon's, frankly,

00:06:07.120 | greatest legacy is just the motivation,

00:06:12.360 | the inspiration he's giving to all the generations

00:06:15.520 | that are coming up, right?

00:06:17.280 | To dream big, to think big, to build stuff that matters.

00:06:21.360 | I think that was only half of your mother's message,

00:06:23.680 | by the way.

00:06:24.520 | Yeah, I mean, like, you know,

00:06:27.480 | I think it was the part to me that really inspires me

00:06:32.280 | and motivates me.

00:06:33.560 | And, you know, I'll get off the soapbox,

00:06:37.080 | but, you know, as we start diving in here,

00:06:39.960 | the one thing I'm just thinking a lot about

00:06:42.160 | is we as a country need to make sure we don't screw this up.

00:06:45.760 | It's the system that's creating all that prosperity.

00:06:48.920 | It's the system that's creating, you know,

00:06:52.480 | the advances in biology, the advances in space,

00:06:55.240 | the advances in AI.

00:06:57.080 | And certainly they'll have challenges,

00:06:58.920 | but it's an extraordinary time to be out here

00:07:01.920 | in Silicon Valley doing what we do, so.

00:07:03.960 | Well, and I, you know, adding on top of that,

00:07:06.880 | something we talked about in one of our first episodes

00:07:09.320 | was that Reagan speech that he gave.

00:07:11.280 | And one of the reasons why the U.S. has been so successful

00:07:15.960 | is open skilled immigration

00:07:18.840 | and getting the best people in the world

00:07:21.120 | to come practice their craft here.

00:07:23.960 | And why we have limited that and not advanced it

00:07:28.480 | is shocking to me.

00:07:30.760 | And there will be ramifications

00:07:32.600 | 'cause the technology allows those people

00:07:35.600 | to stay in other places if they so choose.

00:07:38.160 | So you gotta make it easy for 'em.

00:07:40.360 | Yeah, no doubt about it here, here.

00:07:42.000 | Why don't we jump into, I guess, the first topic,

00:07:44.640 | you know, about this new reality in the late stage market.

00:07:47.480 | Why don't you take us through

00:07:49.320 | what's piqued your interest there?

00:07:51.080 | So let me walk you through this.

00:07:52.600 | And you and I have talked about it in the past,

00:07:54.280 | but I think I've come around to a new perspective.

00:07:58.560 | So I've lived through two different major cycles,

00:08:02.720 | venture cycles, and, you know,

00:08:05.760 | I've watched there be high periods of liquidity

00:08:08.680 | where a ton of returns are made in '99,

00:08:11.960 | kind of 2021, even kind of '07, '08.

00:08:16.280 | And then frequently on the downside, like '01,

00:08:19.320 | you just see a washout, right?

00:08:21.200 | And I've seen people completely cleared out.

00:08:24.280 | And I've often thought about there are changes

00:08:28.560 | to the venture industry that are systematic,

00:08:30.600 | like increasing competition has been completely linear

00:08:33.720 | and systematic since I joined,

00:08:35.960 | but other things are cyclical and they come and go.

00:08:38.840 | And I think I had always thought

00:08:41.320 | that the presence of large amounts of money,

00:08:44.240 | presumably easy to get in the private,

00:08:47.680 | in the late, well, we say late stage,

00:08:49.480 | but it's come so early,

00:08:50.440 | I don't think that's the right word,

00:08:51.800 | but the large, let's just say large round private market.

00:08:56.800 | I've come to believe that it may be a systematic trend

00:09:04.040 | and not a cyclical one.

00:09:05.680 | And there was a mini correction.

00:09:08.560 | I think you have to call it a mini correction

00:09:10.560 | in venture in 2022, 2023.

00:09:14.760 | A lot of companies did layoffs.

00:09:16.280 | You had right sizing.

00:09:17.320 | You had a lot of talk about free cash flow and profitability,

00:09:20.880 | but then the AI wave came

00:09:23.120 | and then the AI wave got so big, right?

00:09:27.200 | What's AI as a percentage of venture capital right now?

00:09:30.520 | 50% at least.

00:09:32.440 | Yes.

00:09:33.280 | And so, and that market is behaving almost like it was

00:09:37.560 | prior to this mini correction.

00:09:40.000 | And so I look around and at some of these data points.

00:09:45.000 | So there was an FT article since you and I talked last

00:09:49.880 | that someone aggregated the cumulative losses

00:09:53.480 | in the food delivery business at $20 billion.

00:09:56.920 | And that's just not your grandfather's

00:10:00.760 | venture capital industry.

00:10:02.440 | That's something new.

00:10:04.840 | There are four companies in the coding co-pilot space

00:10:09.840 | that are not named Microsoft

00:10:12.240 | that have raised over $200 million each.

00:10:15.200 | And we're just, these companies are all of a year

00:10:18.080 | and a half old, right?

00:10:19.720 | Right.

00:10:20.560 | And so it's just a different world.

00:10:22.720 | I look at someone posted just the investments Sequoia

00:10:27.160 | has made in Elon's companies.

00:10:29.720 | And they were rounds that were in the, you know,

00:10:32.640 | 500, $600 million range.

00:10:35.680 | And there were three or four of them.

00:10:37.400 | And once again, just not the historic venture capital model.

00:10:42.400 | Lots of money, lots of, you know, you're in this world.

00:10:47.640 | So don't take this the wrong way,

00:10:50.440 | but like these people are getting two and 20

00:10:53.920 | to write a check for three or $400 million

00:10:56.680 | and not take a board seat.

00:10:58.280 | And for the listeners that may not know

00:11:01.480 | that two represents an annual management fee,

00:11:04.840 | not a one-time management fee.

00:11:07.440 | And it's taken generally over seven, you know,

00:11:10.600 | seven to eight years.

00:11:11.480 | So that could be 10% of that money.

00:11:14.320 | So they're getting-

00:11:15.160 | Although I would point out, Bill,

00:11:16.320 | I think on some of those very large

00:11:18.920 | multi-billion dollar rounds

00:11:20.360 | where the investment size is multi-billions,

00:11:22.560 | I think there are creative fee structures.

00:11:24.800 | I think what's more standard is like

00:11:28.080 | zero and 10 or one and 10.

00:11:29.920 | I've heard those things are changing.

00:11:33.120 | Yeah, yeah, fair enough.

00:11:33.960 | I think that's a fair point.

00:11:35.880 | But I guess my point is,

00:11:38.800 | I think this is more permanent.

00:11:40.640 | I think it would take a massive shake out,

00:11:44.760 | like a gargantuan '01 style

00:11:47.880 | or more to change this at this point.

00:11:50.880 | Well, I mean, listen, I totally agree with you.

00:11:53.200 | In fact, you know, you and I have talked about this

00:11:55.440 | for well over a decade.

00:11:57.280 | You know, in fact, part of the reason I started Altimeter

00:11:59.840 | and part of our thesis in 2005

00:12:01.680 | is we thought companies would stay private longer,

00:12:03.840 | more of the value capture would occur in the private markets

00:12:07.160 | because they would scale faster.

00:12:09.200 | Now, at the time, we thought it was because

00:12:10.680 | they would also be more capital efficient, right?

00:12:13.360 | Think Google raising less than $40 million pre-IPO.

00:12:16.960 | But, you know, I think a lot of this has to do

00:12:20.360 | with the function of changing market structure,

00:12:22.840 | both technology, the way it scales,

00:12:24.960 | regulation, making it less desirable to go public,

00:12:28.200 | and frankly, the development, right?

00:12:30.480 | The market development and responding to that

00:12:32.560 | is just a much deeper and much more liquid pool of capital

00:12:36.000 | for companies that choose to stay private.

00:12:38.080 | So, you know, I pull up a couple of charts here,

00:12:41.960 | but let's ground it in some facts.

00:12:44.160 | Over the last 10 years, there's no doubt

00:12:46.360 | there's been more multistage funds,

00:12:48.560 | larger fund sizes, right?

00:12:50.880 | And this accelerates the trend

00:12:52.320 | because you no longer have to tap the public markets.

00:12:55.320 | So if you look at this chart,

00:12:57.320 | it just shows you the share of private capital

00:12:59.520 | that was raised by the large funds,

00:13:01.880 | and that's just becoming a much, much bigger part

00:13:04.760 | of the total market.

00:13:06.960 | And what I think is particularly interesting about that,

00:13:10.600 | it said through Q2, 2024,

00:13:13.760 | 521 private market funds have raised a total of 295 billion

00:13:18.680 | across these asset classes,

00:13:20.600 | but fund counts fell by 45%

00:13:23.800 | during that same period of time, right?

00:13:26.280 | So yes, this is the case that there is a structural

00:13:29.640 | and dynamic change in these markets.

00:13:32.360 | You've got these multistage funds.

00:13:34.120 | You've got funds like Altimeter

00:13:36.000 | that's been doing this for a long time,

00:13:37.680 | where we do early all the way through public markets.

00:13:41.880 | And that's the amount of capital that's been raised.

00:13:44.280 | And then, Bill, if you look at the amount of capital

00:13:46.600 | that's been deployed and where that's coming from,

00:13:50.240 | you can pull up this chart from Crunchbase,

00:13:52.040 | and it also shows you, right,

00:13:54.040 | that you just have a lot of these late-stage dollars.

00:13:56.680 | Now, what this chart doesn't show you

00:13:58.480 | is the valuation of the rounds when they were getting done.

00:14:02.640 | So I think you're exactly right there.

00:14:04.720 | And then, of course,

00:14:05.680 | this isn't just VC and growth funds anymore.

00:14:08.680 | We have world-class sovereigns

00:14:10.840 | that have moved into this domain.

00:14:13.080 | You can see here this new fund out of Abu Dhabi, MGX,

00:14:17.520 | which is run by, frankly, extraordinary investors,

00:14:21.160 | long-term time horizons

00:14:22.840 | aligned with national sovereign interests.

00:14:25.520 | And it's not just them, right?

00:14:26.960 | The Saudis, the Kuwaitis.

00:14:29.480 | And then don't forget,

00:14:30.560 | we've also seen the private wealth,

00:14:33.800 | the high-net-worth platforms move into this space.

00:14:36.480 | So Goldman Sachs, JP Morgan, Morgan Stanley,

00:14:39.800 | those platforms are now aggregating

00:14:42.120 | and put big dollars into this space.

00:14:44.880 | So the most fascinating part of this

00:14:48.680 | is that we do now, in my estimation,

00:14:51.240 | I think this has been the case for quite a long time,

00:14:54.000 | we have the permanent emergence

00:14:56.160 | of what I've been calling the quasi-public market, right?

00:14:59.400 | And the reason I call it the quasi-public market

00:15:01.680 | is because VC has a certain connotation to it, right?

00:15:06.200 | You're taking that first, second, third round of risk

00:15:09.160 | into a company.

00:15:10.000 | You often own somewhere between 7% and 20% of a company.

00:15:14.400 | You take a board seat, you actively engage

00:15:16.600 | and help build that company's success.

00:15:18.600 | That's very different than investing in a company

00:15:21.880 | at 10, 20, $30 billion that frankly

00:15:25.440 | is in a very different place in its life cycle oftentimes.

00:15:30.200 | So, you know, I don't know.

00:15:31.600 | I look at that and if I ask the question,

00:15:33.840 | is that good or bad?

00:15:35.880 | I don't know.

00:15:36.720 | More liquidity allows guys like Elon's to better experiment,

00:15:40.600 | to take bigger swings of the bat.

00:15:43.840 | Maybe it compresses the margins

00:15:46.280 | for those of us in the investing business,

00:15:48.400 | but I think it leads ultimately to a lot more innovation.

00:15:51.960 | Although I'm sensitive to the point you put out there.

00:15:54.960 | You know, it was just two or three years ago

00:15:57.160 | where we were talking about, right?

00:15:58.840 | The soft bank effect, weapons of economic destruction.

00:16:03.560 | You know, we certainly led to excess competition

00:16:06.320 | in the ride-sharing space.

00:16:08.760 | And so what it did is it slowed down

00:16:10.840 | and prevented the natural order of things.

00:16:12.960 | The 80/20, the winner take most

00:16:15.320 | from developing in a profitable way.

00:16:17.200 | And once that capital dried up a bit, Bill,

00:16:20.960 | obviously you saw the profitability of Uber skyrocket

00:16:24.320 | and the natural kind of market structure set in.

00:16:27.560 | So I do think that's a downside of this,

00:16:29.960 | but net-net, I think it's a positive development

00:16:32.320 | for entrepreneurs and innovation.

00:16:34.040 | Yeah, I could certainly express some things

00:16:37.760 | that I worry about that are not necessarily

00:16:40.880 | in the best interest of the entrepreneur or the investor.

00:16:45.040 | So one, just with this much money being pushed upon you,

00:16:50.040 | and if you don't take it, your competitor will take it.

00:16:54.120 | You're forced into the game.

00:16:55.720 | Like you're not allowed to not play.

00:16:57.320 | Is that what you feel like happened between Uber and Lyft?

00:17:01.400 | Sure, sure, absolutely.

00:17:03.000 | But there were many examples.

00:17:04.480 | I mean, I think that a company back on the board of Zillow

00:17:09.480 | was forced into this home purchasing market

00:17:14.120 | because the open door team raised so much money

00:17:17.640 | and was gonna tell the world

00:17:19.720 | that they're gonna be obsolete.

00:17:22.560 | And so if you don't engage,

00:17:25.880 | you could have multiple compression on your head.

00:17:28.720 | It's an interesting dynamic.

00:17:32.800 | Anyway, you're gonna have higher burn rates.

00:17:35.280 | And so you're gonna know less about your unit economics

00:17:38.160 | just by a natural fact 'cause you get further away

00:17:40.800 | when you're operating that way.

00:17:42.440 | And you can't raise $400 million

00:17:44.120 | and not have a high burn rate.

00:17:45.320 | Like there's no point in it

00:17:47.240 | unless you just like interest income.

00:17:49.160 | The staying private longer thing

00:17:54.680 | becomes both the dog and the tail.

00:17:57.360 | Like it's hard to know what's causing which, right?

00:18:01.000 | Because once the investors want,

00:18:03.400 | and this is true of founder liquidity

00:18:05.880 | and employees secondaries also,

00:18:08.360 | once the people want to bring the money to you,

00:18:13.360 | especially preemptive rounds, this kind of thing,

00:18:16.480 | you start searching for ways

00:18:19.280 | to make it easy for that round to take place.

00:18:21.960 | And so you start encouraging staying private longer,

00:18:24.880 | you start encouraging the secondaries.

00:18:27.520 | And that can create a misalignment of interest.

00:18:30.840 | Let us not forget that the Byrd founder

00:18:32.880 | took out 50 million in a private round,

00:18:36.240 | and that company's now bankrupt, right?

00:18:39.680 | Well, I definitely, there are two things,

00:18:42.160 | two features of this that worry me.

00:18:44.120 | There is absolutely no doubt in my mind

00:18:47.640 | that excess capital distorts company behavior, okay?

00:18:52.640 | And in particularly these early phases,

00:18:56.240 | it's very difficult, it's not impossible,

00:18:58.680 | but it's very difficult to stay fit and efficient

00:19:02.600 | when you have a buffet of all options sitting in front of you

00:19:06.920 | and you can fund all of them.

00:19:08.360 | Scarcity breeds necessity, scarcity breeds innovation.

00:19:12.840 | So I think if you're on the board of a company

00:19:14.520 | or a founder of a company or CEO of a company,

00:19:16.520 | you have to think long and hard about the negative cultural

00:19:20.320 | and negative fundamental effects to your business

00:19:23.440 | of taking too much capital.

00:19:24.560 | I mean, we saw this song and verse

00:19:27.120 | over the period 2018 to 2022.

00:19:30.720 | And frankly, we're still moving

00:19:32.200 | through the hangover of it, Bill.

00:19:33.720 | Yeah, and stay private longer

00:19:36.200 | can become stay private forever,

00:19:38.280 | which has a negative impact on IRR and eventual liquidity.

00:19:43.240 | And if you've taken, there's another element of this

00:19:47.200 | that's gonna bring the regulators in

00:19:48.880 | and Lord knows what they'll do,

00:19:51.240 | but it'll probably mess it up,

00:19:52.760 | is that if this sector of growth

00:19:57.040 | has been stolen from the public markets

00:20:00.160 | by the quasi public market, as you call it,

00:20:03.160 | then the people in Congress will cry foul

00:20:06.560 | that the individual investor

00:20:08.800 | doesn't have an opportunity to play.

00:20:10.800 | I'm sympathetic.

00:20:13.880 | And they'll start doing stuff

00:20:15.560 | that will cause more chaos, I'm sure.

00:20:20.000 | Well, I'm sympathetic to the argument.

00:20:21.680 | You and I've discussed this many times,

00:20:23.280 | the number of public companies in the US has collapsed.

00:20:26.200 | The whole idea of accredited investor status

00:20:28.440 | that you have to have a certain amount of money

00:20:29.800 | in order to be deemed worthy to invest in private companies.

00:20:34.400 | And if they stay private until they're worth $100 billion

00:20:37.600 | or take open AI, like there are a lot of retail investors

00:20:40.560 | that would love to buy open AI today.

00:20:42.800 | And the only ones who can fundamentally,

00:20:45.080 | or who can really access it are accredited investors,

00:20:47.640 | high net worth investors.

00:20:49.160 | So it is ironic-

00:20:50.000 | And then you have to know some, like there's no,

00:20:52.280 | and it doesn't matter, but we're making the same point.

00:20:55.520 | But I would say, again,

00:20:59.080 | at least the companies that we're involved in,

00:21:01.240 | we have these conversations very openly with the founders

00:21:04.880 | about the risk reward and the trade-offs.

00:21:06.760 | I do think there is a lot of these pressures

00:21:09.400 | that you rightfully acknowledge.

00:21:14.360 | But ultimately it seems to me like, as to this question,

00:21:18.320 | whether or not this is going to yield good results, right?

00:21:21.960 | I think it's a net, net,

00:21:23.200 | I think it's generally deeper liquid, more liquid markets,

00:21:25.880 | generally a positive for the entrepreneurs and the founders.

00:21:30.400 | And I think for the firms, listen,

00:21:32.600 | late stage private deals,

00:21:36.320 | like let's go back to think Groupon.

00:21:38.960 | I think the last private round there was 19 billion

00:21:41.720 | and two years later, it's worth 1 billion.

00:21:44.320 | This comes down to stock selection.

00:21:46.240 | This comes down to investing.

00:21:47.560 | It comes down to your underwriting.

00:21:48.840 | It comes down to risk reward.

00:21:50.600 | I might think that open AI at $90 billion

00:21:53.680 | is an incredible risk reward

00:21:55.440 | and wish that I had more money in open AI at that valuation.

00:21:59.480 | Others may think that's ludicrous.

00:22:01.320 | That's called a market, right?

00:22:03.320 | And you're going to be proven right or wrong

00:22:05.520 | in the fullness of time.

00:22:06.960 | And frankly, Bill, to your point about two and 20,

00:22:09.840 | LPs get to decide.

00:22:11.520 | If you go out and you want to raise money from them,

00:22:13.640 | you want to raise billions

00:22:14.760 | in order to put billions in open AI,

00:22:16.800 | they get to ultimately decide what they're willing to pay.

00:22:20.080 | And, you know, again,

00:22:21.640 | they're going to probably be held accountable

00:22:23.320 | at some point in time for their returns as well.

00:22:24.960 | Although that can take a very, very long time.

00:22:27.560 | As you know, that could be a 10 to 15 year window before,

00:22:31.600 | like corrections in the LP market

00:22:33.800 | are one of the slowest things that can possibly happen.

00:22:37.320 | And I'll close this part

00:22:39.360 | by acknowledging something that you've said before,

00:22:41.600 | which is, and I've talked about in other places,

00:22:45.040 | but some of this and the reason it's systematic

00:22:48.600 | and not cyclical is just a recognition

00:22:51.240 | that when technology companies gain a foothold

00:22:54.640 | and have positive momentum,

00:22:56.240 | they often go much further and longer and higher

00:22:59.120 | and they have network effects.

00:23:00.960 | And so this is, some of this at least

00:23:04.080 | is the market coming to grips with that

00:23:07.560 | and recognizing that you can pay 30 or 40 times revenue

00:23:12.000 | for something if it's going to hyper growth

00:23:13.840 | for four or five years because of systematic advantages

00:23:17.000 | and adjusting their game on the field as a result.

00:23:20.080 | - I blame a fair bit of that on you.

00:23:21.840 | I mean, you educating everybody on power law

00:23:25.760 | and network effects over the last 15 years

00:23:28.200 | hasn't helped our plight at all.

00:23:30.000 | I mean, the competition is stiff.

00:23:31.720 | I think that the efficiency

00:23:33.280 | of the late stage private market, you know,

00:23:36.000 | it's no longer five or six players in that market.

00:23:38.800 | As you know, you're talking 20, 30, 40 players, you know,

00:23:42.760 | that will show up to these things.

00:23:44.760 | So it's difficult.

00:23:46.520 | And I know, I stress about this all the time.

00:23:49.440 | We're going to be judged ultimately

00:23:52.000 | by the returns on that capital.

00:23:54.240 | And I think the folks who are deploying that capital

00:23:57.480 | for the most part are pretty extraordinary

00:24:01.120 | and worthy competitors.

00:24:02.200 | And we end up, frankly, these big rounds

00:24:04.080 | oftentimes are collaborations between many firms,

00:24:06.640 | much as you've seen in private equity develop

00:24:08.640 | over the years.

00:24:09.480 | But one of the things I did want to touch on,

00:24:11.240 | you raised this question, like how much of this

00:24:13.440 | has gone into, you know, generative AI

00:24:15.920 | and how is this skewing it?

00:24:17.960 | And so we have a couple of charts here.

00:24:19.240 | I think Sapphire Ventures, you know,

00:24:21.160 | put these charts together and it just shows like, you know,

00:24:24.960 | the vast majority of money, I think AI funding in 2023,

00:24:29.320 | Forexed, you know, 28 billion and over 700 deals.

00:24:33.800 | But what was interesting is about 65% of that,

00:24:36.800 | talking about power law, 65% of that,

00:24:39.000 | I think went into five or six companies,

00:24:42.320 | you know, that you know of.

00:24:43.760 | And by the way, those companies need big checks

00:24:46.400 | because they're consuming voracious amounts of capital.

00:24:50.240 | And you know, perhaps that's a jumping off point.

00:24:53.560 | Although some of this may involve these credit deals,

00:24:56.800 | some of this count.

00:24:58.080 | Oh, for sure, for sure.

00:24:59.600 | And listen, that's another thing one has to take into account

00:25:02.720 | when thinking about these valuations.

00:25:04.520 | But I know that you have a little bit of angst

00:25:06.840 | based on prior pattern recognition,

00:25:08.960 | just about, you know, whether this is developing

00:25:11.880 | into a bit of a hype cycle.

00:25:13.920 | So why don't you talk to us about your thoughts

00:25:15.640 | about what's going on.

00:25:16.480 | Look, I think it unquestionably is in a hype cycle.

00:25:20.400 | And by hype, I don't mean negative necessarily,

00:25:23.680 | just that everyone's talking about it.

00:25:26.640 | There's not a CEO or a CIO in the US

00:25:31.640 | or probably around the world that hasn't asked the question,

00:25:35.200 | what are we doing in AI?

00:25:36.760 | You know, how can it impact our business?

00:25:39.280 | Like it is on the tips of everybody's tongues, right?

00:25:43.920 | And so when you get in these situations,

00:25:47.440 | I'm always like, I don't know why,

00:25:49.800 | but I'm always fascinated with what's reality

00:25:53.040 | and whether we might go too far

00:25:55.080 | and what we're promising versus what can actually be done.

00:25:59.640 | I think in this case, because some of the leaders

00:26:05.200 | and I'd say OpenA is probably the strongest,

00:26:08.880 | are willing to spout hyperbole.

00:26:11.400 | And what I mean by hyperbole is they're willing

00:26:14.320 | to just state really vague, broad, big ideas

00:26:18.360 | without much meat on them.

00:26:20.260 | And I think that then creates a situation

00:26:24.800 | where I think you end up having tension in the system

00:26:28.680 | where some people feel like their own business model

00:26:32.400 | is threatened by this notion,

00:26:34.200 | especially if the notion goes too far.

00:26:36.220 | There was a super interesting comment made

00:26:38.280 | by the new chairman of TSMC where he said,

00:26:41.160 | "OpenAI Sam Altman is too aggressive for me to believe."

00:26:45.480 | Now, why would he go out and say that out loud?

00:26:48.280 | Like, you know, what's his incentive?

00:26:51.260 | And I would say his incentive is he's running a business.

00:26:54.520 | He's got people asking him questions all the time

00:26:57.280 | about where this business is going, how's it gonna be.

00:27:00.240 | And there's this other person running around the globe

00:27:04.000 | telling everyone everything, including spreading rumors

00:27:07.200 | that he's gonna spend 100 billion of Microsoft's money

00:27:10.280 | and he's gonna build his own chips and build his own fabs.

00:27:13.720 | And that then starts to be distracting

00:27:17.600 | to a company like TSMC.

00:27:18.860 | Yeah, but hey, you know,

00:27:20.520 | like let me take the other side of it, right?

00:27:22.560 | I'm sure you would.

00:27:24.400 | No, I mean, I'm just saying, you know,

00:27:27.280 | Elon set out a vision for rockets that could land themselves

00:27:30.600 | and auto fleets that would be replaced by electric cars

00:27:33.760 | at the time he said these things,

00:27:35.360 | they sounded totally outlandish.

00:27:37.160 | And by the way, the time took longer

00:27:39.680 | to get to most of these points than Elon envisioned,

00:27:42.320 | but we ultimately got there

00:27:43.960 | and it ultimately blew our socks off.

00:27:45.800 | And I think two of the things you have to do, right?

00:27:48.520 | You have to will the future.

00:27:49.920 | You have to manifest the future.

00:27:51.320 | You have to describe the future.

00:27:52.800 | You have to motivate your employees.

00:27:54.580 | And importantly, you have to motivate sources of capital.

00:27:58.420 | Right?

00:27:59.260 | And so I think in the case of Sam, you know,

00:28:02.120 | he's out there saying these things.

00:28:03.480 | And again, I don't need to come to his defense.

00:28:06.680 | I think he's incredibly thoughtful.

00:28:08.000 | I think he is a true believer.

00:28:09.760 | As I am a believer,

00:28:11.040 | I don't know what the time series is going to be.

00:28:13.320 | And I'm certain all of these experiments will not work.

00:28:16.560 | With that said, if I am the CEO of TSMC,

00:28:19.920 | I would say exactly what he did.

00:28:21.680 | And the reason I would say it

00:28:23.280 | is because if Sam's going to emerge as competition,

00:28:26.800 | build his own fab, build his own chip, et cetera,

00:28:29.320 | what would I want to do?

00:28:30.200 | I would want to undermine the sources of capital

00:28:32.720 | so that he can't become competition.

00:28:34.880 | I would say to the sovereigns

00:28:37.520 | who are thinking about funding him,

00:28:39.000 | whoa, you better be careful

00:28:40.760 | because this guy is too aggressive.

00:28:42.960 | This is part of the war of words

00:28:44.880 | that we see out there all the time.

00:28:46.440 | I mean, Databricks does this and incredibly,

00:28:48.960 | Ali is amazing against Snowflake.

00:28:52.240 | I see this happen in VC land all the time

00:28:55.240 | where people will say something

00:28:56.440 | to try to freeze the markets

00:28:58.400 | so a company doesn't get funded.

00:28:59.920 | - Rarely works, yes, yes, yes.

00:29:01.760 | A company doesn't get funded.

00:29:03.360 | So all I would say is that it may in fact be very shrewd

00:29:08.360 | by Sam to be doing exactly what he's doing.

00:29:12.360 | I don't think that necessarily undermines

00:29:14.880 | the advances that we're making.

00:29:17.800 | Although I would say that we're in the fog of war right now

00:29:20.840 | and it's very hard to know the timescale

00:29:23.560 | that a lot of these things will unfold.

00:29:25.160 | - I would, so you use maybe like Elon and the Tesla example.

00:29:30.040 | I think another example is crypto, right?

00:29:33.440 | And so we went through a phase

00:29:36.400 | where there were very, very smart people

00:29:39.040 | on podcasts like this and around the globe

00:29:44.200 | saying that crypto and blockchain

00:29:46.680 | would replace the corporate entity

00:29:48.600 | and that marketplace companies like Uber wouldn't exist

00:29:53.200 | and all this stuff.

00:29:54.600 | And that just didn't play out

00:29:56.960 | and I don't think it's going to.

00:30:00.080 | - Yeah.

00:30:00.920 | - I'll go out on a limb.

00:30:03.400 | And so, but for a while, we all believed it.

00:30:06.280 | There was a moment in time where the scooter companies

00:30:11.080 | were claiming that they were gonna take 75% of Uber's rides.

00:30:15.480 | That did not happen.

00:30:16.840 | There hardly anyone riding here in Austin.

00:30:20.000 | And so sometimes you're right.

00:30:23.440 | Sometimes you are drawing a picture of the future

00:30:27.000 | that we actually get to.

00:30:28.640 | And sometimes that doesn't play out.

00:30:30.560 | So, you know, there's more meat on the bone here to be fair.

00:30:35.040 | But when you say this stuff will do anything and everything,

00:30:38.400 | when you say-- - Oh, for sure.

00:30:39.440 | - My computer, when it's not summarizing my email,

00:30:42.760 | will be curing cancer.

00:30:44.480 | And when you start talking about UBI

00:30:47.200 | and how no one's gonna work,

00:30:48.840 | like, I just think that's like la la land stuff.

00:30:53.080 | And lots of people are saying it.

00:30:55.040 | Lots of people are saying it.

00:30:56.640 | - No, no, listen, I think here's where I think

00:30:59.440 | your admonition is smart.

00:31:02.120 | And as you know, we've looked at a lot of stuff

00:31:04.080 | at Altimeter in the AI landscape.

00:31:06.000 | You know, enabling technologies, infrastructure,

00:31:08.200 | picks and shovels.

00:31:10.000 | For all the reasons about uncertainty and high valuations,

00:31:14.360 | a lot of valuations reflect or discount

00:31:17.480 | or underwrite high levels of certainty, right?

00:31:20.240 | That I think are hard to peg at this moment in time.

00:31:22.920 | This is what I think happens in the fog of war, right?

00:31:26.440 | And the goal of an analyst is to develop deep conviction

00:31:29.960 | at these moments in time

00:31:31.680 | and be right before everybody else does, right?

00:31:34.560 | I think about this in the internet.

00:31:37.160 | People had deep conviction it was going to be huge,

00:31:39.520 | that search was gonna be important.

00:31:41.520 | But a lot of people ran out and invested in Ash Jeeves

00:31:44.280 | and Alta Vista and Lycos and Excite

00:31:47.320 | and name all the companies

00:31:48.640 | that a few years later would go to zero.

00:31:50.480 | And it's almost certain that that will also happen in AI.

00:31:54.560 | That didn't mean that the internet wasn't going to be huge

00:31:57.320 | or that Google wasn't going to be

00:31:58.760 | a multi-trillion dollar company,

00:32:01.120 | which I also think will happen here.

00:32:03.920 | But the, go ahead.

00:32:05.560 | Well, I just think this is,

00:32:06.520 | so I wanted to do this one before the software one

00:32:09.760 | because I think they're relevant to one another.

00:32:11.840 | So there are people that believe that AI,

00:32:16.840 | and when they say that, I can't,

00:32:19.280 | I never know whether they're talking about AI or LLMs,

00:32:22.880 | which I view as a very small subset of AI,

00:32:26.040 | but we'll just do everything.

00:32:30.280 | It's just gonna do everything.

00:32:32.120 | And so this, our last topic about,

00:32:36.160 | these people have come out and said,

00:32:37.880 | well, because of what you saw in the earnings period

00:32:40.280 | and how the stocks reacted, software is dead.

00:32:43.600 | And so there are people that believe one day

00:32:46.120 | you'll just tell your LLM what you want it to do

00:32:49.200 | and it'll do everything that software did.

00:32:52.480 | - Right.

00:32:53.520 | - That's a pretty strong form of it.

00:32:55.480 | I think there's a lesser strong form of it

00:32:57.720 | that the UI around LLMs are gonna enable

00:33:02.720 | a type of interaction with what we used to think of

00:33:07.680 | as a SaaS application,

00:33:09.680 | that's gonna make the older apps feel tedious

00:33:13.320 | and therefore you end up with a replacement cycle,

00:33:16.440 | maybe as big as SaaS replacing on-prem,

00:33:21.520 | maybe as big as SaaS replacing client server

00:33:25.120 | or whatever came before or many in mainframe.

00:33:28.240 | And if you believe that,

00:33:30.480 | how much of this gets rewritten?

00:33:33.080 | So you've invested in so many software companies,

00:33:36.600 | you guys are deep in your analytics.

00:33:38.640 | What did you see in the past three weeks from earnings

00:33:42.840 | and how do you think about AI as a risk

00:33:46.200 | for multiple compression and disruption

00:33:50.040 | for the software industry writ large?

00:33:53.200 | - Yeah, no, I think it's super important question.

00:33:55.720 | By way of transition to that,

00:33:58.520 | I did wanna say this thing

00:34:00.120 | that not everybody is all in on AI.

00:34:02.960 | I'm going to WWDC at Apple tomorrow

00:34:06.760 | and Apple Intelligence is out this morning

00:34:10.240 | talking about AI service that they're gonna unveil.

00:34:14.080 | They have not spent tens of billions of dollars

00:34:16.360 | building LLMs or frontier models

00:34:19.600 | and there are some people who are critical of that bill.

00:34:22.240 | And I imagine they're looking at it and saying,

00:34:24.400 | "We don't see the industrial logic.

00:34:26.200 | We don't see the return."

00:34:27.400 | It's known to be a very financially

00:34:30.680 | and fiscally conservative company.

00:34:32.480 | And so they'll probably announce a partnership with OpenAI.

00:34:35.440 | My sense is they probably look at that and say,

00:34:38.080 | "Oh, that's a transition for us.

00:34:39.920 | Let them spend all the huge early dollars.

00:34:42.760 | We'll come in and be a late mover.

00:34:44.800 | We'll spend a fraction of the money."

00:34:46.680 | And ultimately we control the platform.

00:34:48.880 | We control the device.

00:34:51.160 | And so we're not at risk of getting disintermediated.

00:34:54.560 | My point is that I think there are different choices

00:34:56.840 | being made by different companies.

00:34:58.640 | But that said, using the Apple example,

00:35:02.520 | I do think that the heat is so loud.

00:35:05.160 | Even on this very podcast, we go on, "What are they doing?

00:35:08.560 | Why haven't they made Siri better?"

00:35:10.560 | And like the drumbeat gets to the point.

00:35:13.200 | If they were to hold that tomorrow and not mention AI,

00:35:16.480 | which they would never do,

00:35:19.000 | it would raise immense questions.

00:35:21.560 | No, that's for sure.

00:35:22.560 | Of course not.

00:35:23.400 | Why would they?

00:35:24.240 | But I think they're going to have a series of announcements

00:35:27.680 | in terms of integrating with OpenAI.

00:35:29.560 | They're gonna make the phone better.

00:35:31.440 | They're gonna say you can only get it on 15 plus or 16.

00:35:34.320 | They'll drive a replacement cycle that begins in 2025,

00:35:38.840 | but they won't spend the dollars to have their own solution

00:35:42.040 | probably until you get in to the release cycle

00:35:45.880 | at the end of '25.

00:35:47.560 | And I think that's a perfectly acceptable solution.

00:35:49.840 | Listen, if everybody was critical of them, Bill,

00:35:51.840 | for not building their own LLM,

00:35:53.200 | everybody knows they have not yet.

00:35:55.160 | And so if people were critical of them,

00:35:56.560 | the stock wouldn't be at 196 bucks a share, right?

00:35:59.480 | People are voting with their wallets and saying,

00:36:01.280 | "Listen, it seems like a pretty good balance

00:36:03.760 | between the choices that you've had."

00:36:05.840 | In fact, I think if people are gonna be critical

00:36:08.000 | of anything, a lot of people are looking

00:36:09.760 | at the total CapEx of the hyperscalers,

00:36:12.040 | now at $200 billion, and saying,

00:36:14.560 | "When are you gonna get a return

00:36:15.880 | on the dollars that you're spending?

00:36:17.680 | And how on earth can you go up from there

00:36:20.560 | and have the industrial logic to earn a return?"

00:36:23.160 | Since you mentioned the hyperscalers,

00:36:25.280 | there was a series of layoffs announced in two of them.

00:36:30.040 | I think it was Google and Azure, was it?

00:36:34.400 | Or Amazon? Yes.

00:36:35.600 | I think they were small targeted.

00:36:37.320 | If you're in just hyper growth mode,

00:36:41.800 | why are you doing layoffs?

00:36:42.800 | I don't understand.

00:36:43.640 | I mean, listen, look at the most recently reported

00:36:47.920 | number of employees at Meta.

00:36:49.720 | Everybody knows that Zuck's in beast mode around AI.

00:36:53.120 | And yet, when he reduced the headcount

00:36:55.200 | from 86,000 to 69,000 a couple years ago,

00:36:59.000 | I think at the end of the most recent quarter,

00:37:00.640 | he still only had 69,400.

00:37:03.400 | There's massive slack in these businesses, Bill,

00:37:05.880 | if they weren't firing people

00:37:07.560 | or encouraging them to turn over.

00:37:08.400 | Yeah, it was just within the unit specific to this,

00:37:11.320 | which was confusing to me.

00:37:12.600 | Anyway, I think it's a good sign out of those guys,

00:37:15.920 | but let's unpack the software stuff that you led us to.

00:37:19.440 | And what I want to start off by saying

00:37:21.880 | is there's a lot to unpack,

00:37:25.840 | but let's start with the obvious, right?

00:37:27.400 | When the future gets less predictable, right?

00:37:30.760 | For any reason, whether it's macro, whether it's micro,

00:37:35.160 | then you have to increase the discount rate

00:37:37.320 | in your free cash flow and your DCF, right?

00:37:39.880 | And that means that multiples go down.

00:37:42.360 | Slowing growth also reduces multiples.

00:37:45.840 | And then we've talked a lot about how high interest rates,

00:37:48.960 | higher than expected, reduces multiples, right?

00:37:51.760 | So this is the triple whammy.

00:37:53.800 | The perfect storm for software multiples is right now

00:37:57.240 | because we've had slowing growth.

00:37:58.920 | We have a lot more uncertainty about the future,

00:38:01.200 | irrespective of what side you're on.

00:38:03.760 | And on top of that,

00:38:05.240 | interest rates have remained higher than expected this year.

00:38:08.840 | And we were starting from historical highs during ZURP.

00:38:11.680 | So let's take a look, right?

00:38:13.480 | This is the chart a lot of people have seen,

00:38:16.080 | and this is made by our team.

00:38:18.760 | And it just shows you where we are

00:38:20.520 | in the historical context of forward revenue multiples, right?

00:38:24.920 | And so we're trading, you know,

00:38:26.440 | about 20% below the 10-year average ex-COVID.

00:38:30.600 | And some people are starting to view that as an opportunity

00:38:33.560 | because when you see headlines that all software is dead,

00:38:37.080 | if you don't believe that to be true

00:38:39.120 | and you see these valuations,

00:38:41.080 | then you say to yourself, "I need to go hunting."

00:38:43.640 | And I'll tell you, there is smart money

00:38:45.960 | that's starting to buy software again.

00:38:47.960 | You see this next chart, which was by Goldman Sachs,

00:38:51.560 | and it just shows us a multiple of free cash flow.

00:38:54.800 | So similar to the chart above.

00:38:57.120 | But what did we hear in the quarter, Bill?

00:38:59.480 | Okay, so UiPath came out

00:39:02.160 | and said their growth slowed down to 6%.

00:39:05.160 | Salesforce came out,

00:39:06.440 | said their growth slowed down to 7%.

00:39:08.840 | And I think for a company like Snowflake,

00:39:10.880 | it came in at 26%.

00:39:13.440 | And Databricks is rumored to still be growing

00:39:16.120 | well in excess of 50%, okay?

00:39:18.680 | But I pulled some snippets

00:39:20.320 | that we could throw up on the screen here

00:39:23.080 | that we got from the commentary, right?

00:39:25.520 | Workday said we saw probably a bit more scrutiny

00:39:29.080 | than we've seen this time last year.

00:39:31.000 | I just think people have taken a little bit of a pause.

00:39:34.200 | Salesforce, the momentum we saw in Q4 moderated in Q1,

00:39:39.240 | and we saw elongated deal cycles, right?

00:39:42.600 | So deal compression and high levels of budget scrutiny.

00:39:46.440 | UiPath, in mid-March,

00:39:47.920 | we began seeing increased deal scrutiny

00:39:50.160 | and longer sales cycle with large multi-year deals, right?

00:39:54.520 | So multiples in all of them have compressed.

00:39:58.320 | And one of the things I think about is public markets,

00:40:02.120 | when they hear things like that,

00:40:03.360 | they shoot now and ask questions later,

00:40:05.720 | or as Druckenmiller likes to say,

00:40:07.480 | invest and then investigate.

00:40:09.400 | And so if you look at

00:40:11.320 | the forward free cashflow multiples on these business,

00:40:14.800 | UiPath is now trading at something like 20 times

00:40:18.360 | and Snowflake at 36 times.

00:40:20.800 | If you look at their revenue multiples,

00:40:22.520 | 4X and 9X respectively on 2025.

00:40:26.280 | So those are a hell of a lot lower

00:40:27.960 | than what we saw over the past few years.

00:40:30.520 | But the question really is,

00:40:32.760 | what does this all mean for the future of growth,

00:40:35.400 | for the future of profitability of these businesses?

00:40:38.440 | And can you plausibly see,

00:40:40.640 | is the core business gonna be attacked

00:40:42.840 | or do you see use cases where AI

00:40:45.320 | is actually going to be an accelerant to the business?

00:40:49.400 | Let's take those one at a time.

00:40:50.600 | So I would make the argument that the pressure,

00:40:55.600 | and I say this without judging it as positive or negative,

00:41:01.080 | the pressure to be completely focused on AI

00:41:05.960 | at the CEO level and the CIO level

00:41:08.640 | is so high right now, from everywhere,

00:41:12.000 | picking up the Wall Street Journal and reading it,

00:41:13.880 | watching CNBC, listening to us, whatever,

00:41:17.120 | that they have to spend or they feel they have to spend.

00:41:20.600 | And we'll post a link to the CIO survey

00:41:25.600 | that Battery Ventures published.

00:41:28.120 | But what you see is, I think 8% of CIOs

00:41:32.760 | have a budget increase over 10%.

00:41:35.200 | So 92% don't.

00:41:37.200 | So think near fixed budget.

00:41:40.520 | But 85% of them say they're aggressively increasing

00:41:43.600 | their spend on AI.

00:41:44.600 | - Right.

00:41:45.440 | By definition, it's gotta come out of something.

00:41:49.440 | - It's gotta come from something, right?

00:41:51.040 | And so I think we're at a point where if you're not AI,

00:41:55.520 | you're budget's somewhat at risk in selling into a CIO.

00:42:00.120 | And by the way, one category that almost never gets reduced

00:42:04.040 | is security spend.

00:42:05.520 | So if security spend's not going down and AI's going up,

00:42:10.520 | it's even, you're more, if you're not security or not AI,

00:42:16.400 | I think you're even more at risk

00:42:17.960 | of having trouble with expansion dollars.

00:42:21.520 | - No, there's no doubt about it.

00:42:24.240 | That's what we've seen.

00:42:25.880 | That's why I think you see some of this course slowing.

00:42:28.640 | Then one thing you didn't point out

00:42:30.680 | is when people have questions about the future,

00:42:34.160 | which they do around AI.

00:42:35.960 | So let's say they're spending a bunch of money on Workday,

00:42:38.400 | spending a bunch of money on Salesforce,

00:42:40.760 | spending a bunch of money on Snowflake or Databricks

00:42:43.160 | or you name it.

00:42:44.480 | And now they go in to present their case for this year.

00:42:47.720 | And the first question that's gonna get asked to them

00:42:49.880 | is, well, how are these guys doing with AI?

00:42:52.200 | What are they doing with AI?

00:42:53.680 | Is this the multi-year bet we wanna make?

00:42:55.960 | Or should we be betting on Google?

00:42:57.360 | Or should we be betting on Microsoft Azure?

00:42:59.640 | It just freezes.

00:43:00.840 | It says, go back, do more analysis,

00:43:03.080 | and then come back to me.

00:43:04.400 | And I think that's probably the number one thing

00:43:06.920 | you're seeing here, Bill.

00:43:08.120 | Rather than budget pressures,

00:43:09.760 | what I really think you're seeing

00:43:11.400 | is the slowing down and the elongation of the big commits

00:43:15.320 | because people really need to make sure

00:43:17.480 | that they're betting on the future, not on the past.

00:43:19.680 | - Well, and that becomes especially true in two areas.

00:43:24.680 | One is any area where LLMs are proving to be effective.

00:43:31.360 | And so in the enterprise,

00:43:33.520 | customer support is the one everyone's talking about, right?

00:43:37.680 | So if you sell a system in that space,

00:43:41.040 | and it's gonna cause this freezing

00:43:43.760 | you're talking about writ large,

00:43:45.640 | because this is the area where enterprises

00:43:48.680 | are experimenting the most.

00:43:50.320 | It's where there's the most number

00:43:52.000 | of application-based AI startups.

00:43:56.040 | And it's the one where you're getting

00:43:57.600 | the most reinforcement in the public discourse

00:44:00.400 | about success.

00:44:01.400 | You're hearing people,

00:44:03.000 | I don't know if the clarinet thing's real

00:44:04.520 | where you fire 70% of your workers,

00:44:07.240 | but there's plenty of people that echo something

00:44:09.240 | similar to what you heard about copilot for programmers,

00:44:12.720 | 20% gains of efficiency for your workforce,

00:44:16.520 | that kind of thing.

00:44:17.600 | And so those areas are especially true

00:44:20.760 | in what you're talking about.

00:44:21.600 | Another area that you just mentioned in HR,

00:44:25.960 | there are a lot of applications in the hiring process.

00:44:30.320 | I've seen a lot of AI apps in that area.

00:44:33.040 | And so you're gonna freeze,

00:44:35.400 | you're naturally gonna freeze and say,

00:44:37.080 | "Oh shit, I gotta figure out what this is gonna mean."

00:44:40.240 | So there are repercussions.

00:44:42.160 | Now, there's a question like that kind of thing

00:44:44.920 | can go too far.

00:44:45.840 | Like in the example of scooters and Uber,

00:44:48.960 | where everyone thinks that it's gonna be disruptive

00:44:52.160 | and it won't.

00:44:53.080 | And these are hard things to figure out.

00:44:55.200 | The second category where this can happen

00:44:57.520 | is just where LLMs are really good.

00:44:59.760 | UiPath was a company that took a particularly big fall.

00:45:04.760 | And if you study the different uses of RPA,

00:45:07.960 | some of them are form-based,

00:45:09.400 | some of them are ingesting invoices,

00:45:11.640 | some of them, some of those automation processes

00:45:15.000 | are things that LLMs are very, very good at.

00:45:18.120 | Right.

00:45:18.960 | And then that puts you more in the crosshair, right?

00:45:21.240 | Yeah, I think you nailed it.

00:45:23.280 | And listen, lots of people were short UiPath,

00:45:26.000 | lots of people are short these call center

00:45:28.280 | software businesses for all the reason

00:45:30.240 | that you're talking about.

00:45:31.240 | I think our good friend, Aaron Levy,

00:45:34.200 | laid this out pretty well in this tweet here,

00:45:38.400 | where he talks about these three major axes,

00:45:40.760 | the things that are most likely to be replaced.

00:45:45.040 | I mean, it's similar to what you just talked about.

00:45:47.400 | What's the level of automation being applied to the work?

00:45:50.240 | What's the cost of the work that's being automated?

00:45:52.560 | What's the volume or frequency of the work

00:45:54.480 | that's being automated?

00:45:55.840 | So like in the case of UiPath, to your point,

00:45:59.000 | here's a business, think about the setup here, Bill.

00:46:01.960 | It grew only 6% in the quarter.

00:46:04.320 | Most of its free cash flow gets eaten up

00:46:06.520 | by stock-based compensation.

00:46:08.120 | So one might argue that it's not even real free cash flow

00:46:11.400 | on a per share basis.

00:46:13.080 | And it's right in the center of the bullseye

00:46:15.600 | how AI can automate this stuff, which

00:46:17.880 | at a minimum causes a lot of churn, a lot of delay,

00:46:20.560 | and a lot of pricing pressure.

00:46:22.560 | So I think the market's reaction to some of these things

00:46:25.920 | is pretty rational.

00:46:27.640 | Now, take something that's, I think,

00:46:29.960 | a hotter topic among a lot of our friends, which

00:46:32.240 | would be Salesforce.

00:46:33.560 | An incredible founder, CEO, and Mark Benioff.

00:46:37.280 | Mark has been early to get on trends,

00:46:40.400 | whether they're social, whether they're mobile, et cetera.

00:46:43.120 | And he's been all over AI.

00:46:45.880 | But you have a debate.

00:46:47.640 | On the one hand, you have folks like Chamath

00:46:51.720 | saying his company 80/90 can really disrupt them

00:46:57.080 | because you can get 90% of the benefits for 80% of the cost.

00:47:02.000 | Aaron Levy would argue, no way, you can't do that.

00:47:05.000 | People don't want a constellation of services.

00:47:07.680 | All these things exist because this

00:47:09.480 | is what Salesforce customers are demanding of them.

00:47:13.480 | But I think part of the reason Salesforce has recovered well

00:47:16.720 | here, Bill, is this is a company that has gotten fit.

00:47:20.840 | This is a company that is running efficient.

00:47:23.400 | So out of their $13 billion of free cash flow, they convert--

00:47:27.560 | I think they have $2 or $3 billion in SBC.

00:47:30.400 | And so they convert over $10 billion in free cash flow.

00:47:33.880 | So if you look at it for a market leader,

00:47:35.800 | it's not that expensive, even though its growth rate

00:47:38.720 | has slowed way down.

00:47:40.880 | And then I think, listen, you've asked me

00:47:44.240 | a fair bit about Snowflake.

00:47:46.520 | And folks, because we were early investors in that company,

00:47:51.240 | I get questions about Snowflake and Databricks

00:47:53.440 | and these data platforms all the time.

00:47:55.720 | Like, what does it mean for the database?

00:47:57.680 | What does it mean for these data platforms?

00:47:59.880 | So maybe just a second on that.

00:48:03.880 | I think one of the things people question when--

00:48:07.320 | in the case of Snowflake, growth decelerates to 26%.

00:48:12.560 | People are like, OK, yes, the multiple on free cash flow

00:48:16.160 | has come down a lot.

00:48:17.040 | But they still have a lot of SBC.

00:48:20.920 | All that free cash flow gets eaten up by SBC.

00:48:23.360 | So a lot more scrutiny gets put on those free cash flow.

00:48:27.120 | And I think they're going to have to demonstrate

00:48:29.480 | how the growth rate will remain higher for longer

00:48:33.000 | and how they can get more fit around SBC

00:48:35.960 | if they want to maintain or re-expand their multiple.

00:48:40.600 | But when I evaluate them or Databricks

00:48:44.160 | across the three axes that Aaron laid out,

00:48:48.560 | they have a lot of things that they can expand

00:48:51.720 | and are upsides, I think, from AI.

00:48:54.000 | First, I just think the core business of data--

00:48:57.240 | data is a primitive to AI.

00:48:59.600 | You need structured data.

00:49:00.960 | You need unstructured data.

00:49:02.560 | I happened to be at their event last week.

00:49:05.480 | I see the head of data from ExxonMobil

00:49:07.560 | talking about how they're going all in on data engineering

00:49:12.600 | and the data platform there.

00:49:13.960 | And there's a lot more workloads that they're

00:49:16.440 | going to bring to bear.

00:49:17.440 | So I think just the stickiness of enterprise relationships

00:49:21.440 | outside of Silicon Valley, it's deep and it's broad.

00:49:24.600 | And these companies have made long-term commitments

00:49:27.240 | to these platforms.

00:49:28.160 | They're not just going to shift them in a second,

00:49:30.680 | though they may slow down expansions

00:49:33.240 | to the point we just made.

00:49:34.800 | But what are a couple of the easy places

00:49:37.320 | I think that Databricks and Snowflake

00:49:39.320 | can go to automate functions?

00:49:41.000 | One would be transforming data.

00:49:43.240 | So if you have 100 different sources of disparate data,

00:49:46.280 | just think you've got to dedupe that data.

00:49:49.000 | You've got to cleanse that data.

00:49:50.280 | All the things that there used to be a lot of manual

00:49:52.440 | interventions, a lot of workloads in order to do,

00:49:55.040 | I think AI can do that particularly well.

00:49:57.560 | AI infrastructure, training models, building chatbots,

00:50:00.960 | fine tuning.

00:50:01.960 | Again, I think that they can build--

00:50:03.840 | all those guys are building the AI infrastructure to do that.

00:50:06.640 | And then just basic thing, like how do we

00:50:08.920 | extract signal from the data?

00:50:10.920 | So text to SQL, right?

00:50:14.280 | Being allow-- allow somebody to talk to the data

00:50:18.160 | and spin up UIs that really used to be big businesses

00:50:23.280 | unto themselves, business intelligence companies.

00:50:25.680 | So I think that there is an opportunity for that.

00:50:28.680 | And in fact, there was this video,

00:50:32.320 | I think, that one of my analysts posted.

00:50:34.000 | I saw it posted by a few people on Twitter, which is Jensen.

00:50:37.360 | And we'll spin it up here.

00:50:38.840 | So that Snowflake is no longer just a data company,

00:50:42.240 | but they're also a computing company, running Cortex AI.

00:50:46.760 | Big opportunity for Snowflake.

00:50:48.400 | If you guys are watching, if you guys are following Snowflake,

00:50:51.000 | Snowflake just added a new business to themselves.

00:50:55.400 | Not just computing, not just data processing, but computing.

00:51:00.560 | Accelerated computing.

00:51:02.720 | Jensen talking about Snowflake over in Taipei

00:51:05.720 | this week in the context of Cortex AI.

00:51:08.920 | And he said, listen, this is a huge new business

00:51:11.400 | that we've done in partnership with them that is totally

00:51:14.320 | upside to their core business.

00:51:16.120 | Now, whether or not that, in fact, shows up

00:51:18.040 | and the revenue shows up, that's at the heart of the debate

00:51:21.120 | that everybody's having about every one of these platforms

00:51:24.600 | today.

00:51:25.480 | It's interesting.

00:51:26.560 | And ironically, it gets at one of the same things

00:51:29.640 | we've talked about on the consumer side with memory.

00:51:32.800 | And today, the LLM, because it's so text heavy in how it works,

00:51:38.920 | how it was built--

00:51:40.440 | I mean, text is the cornerstone to it, and language.

00:51:45.040 | And as a result, when people use the word hallucination

00:51:50.400 | or whatever, they're talking about errors.

00:51:53.000 | And you can't really rely on a system

00:51:56.200 | to be numeric if it hallucinates.

00:51:59.080 | And so you're not going to run your accounting on an LLM

00:52:04.520 | today.

00:52:05.440 | And I think the interesting--

00:52:08.720 | so today, in the enterprise, you're

00:52:10.480 | seeing a lot of-- we already talked about customer support.

00:52:13.320 | In the database area, we're talking

00:52:15.000 | about having basically language translate

00:52:19.680 | into very complex queries and have

00:52:22.600 | the LLM live between the questioner and the data source

00:52:26.800 | and therefore provide value by being a UI of sorts for input

00:52:33.880 | and output.

00:52:34.840 | Let me give you the example of this.

00:52:36.360 | I mean, how many times have you wished

00:52:39.120 | that a CEO running a business, they

00:52:43.280 | don't want to go to their data analyst

00:52:45.360 | to try to spin something up.

00:52:46.520 | They have a question.

00:52:47.360 | They just want to ask their computer.

00:52:48.900 | Ask Google, and it gives you the freaking answer.

00:52:50.960 | Yes, yes, yes, totally.

00:52:52.480 | I totally get it.

00:52:53.200 | And that's what's happening, and that's what people are doing.

00:52:57.200 | There's a company called Glean that

00:52:58.880 | has a bit of momentum that props up

00:53:01.840 | kind of a universal corporate AI query

00:53:06.480 | against a bunch of different data sources in your business.

00:53:09.200 | And there's authentication and security risk

00:53:11.960 | and all this stuff, and they help you manage all that.

00:53:15.240 | But I think the long-term question about how disruptive

00:53:19.120 | this will be for the app companies

00:53:21.800 | comes down to whether any of the foundational model

00:53:25.160 | companies eventually build in a data store

00:53:28.280 | that is easy to query and is holistic and not lossy.

00:53:32.960 | And that hasn't happened to date,

00:53:35.360 | but I bet you it's something they're thinking about

00:53:37.800 | from a multimodal standpoint, such that a developer--

00:53:41.160 | today, any developer using a foundational model

00:53:44.400 | is using a separate data store, right?

00:53:47.000 | And they're using it for UI.

00:53:49.240 | They're using it potentially for data cleansing or anything

00:53:53.560 | like that, but they're not storing

00:53:55.720 | numeric data inside of it.

00:53:57.600 | So I think that would be something

00:53:58.920 | to watch over a very long time frame

00:54:01.240 | to see if anyone tries to build that into the API, if you will,

00:54:06.440 | on how you use one of these models.

00:54:08.520 | And this really, Bill, is, I think,

00:54:12.880 | what happens at the start of these phase shifts, right?

00:54:16.760 | It is the fog of war, and we get these headlines--

00:54:19.680 | the end of software.

00:54:21.240 | Software is dead.

00:54:22.880 | I will tell you, in technology, you probably

00:54:28.520 | have better opportunities listening

00:54:31.200 | to the wise words of Warren Buffett, which

00:54:33.840 | is, you buy when there's blood in the streets,

00:54:36.840 | and you sell when there are trumpets in the air.

00:54:39.720 | And trumpets were the air in software during 2021, right?

00:54:44.600 | Because people said, these are annuities.

00:54:46.920 | The discount rate should be really low.

00:54:48.640 | They're going to last forever.

00:54:49.880 | They're going to grow to--

00:54:50.920 | They got comfortable with 22x revenue multiples.

00:54:52.840 | Correct, correct.

00:54:54.400 | And now, in a very short period of time,

00:54:56.800 | you had a disruptive force.

00:54:58.480 | We tend to overshoot.

00:55:00.120 | And I suspect that these headlines, the end of software,

00:55:02.920 | the death of software, in, I would argue, less than 24

00:55:07.520 | months, will appear to be silly.

00:55:09.480 | I think there are some really interesting investments

00:55:12.520 | to be made, particularly in the public markets,

00:55:15.840 | because a lot of these incumbents

00:55:17.880 | have incumbent advantages.

00:55:19.720 | And they're going to accelerate when all of this inference

00:55:23.560 | starts coming online in Q3 and Q4.

00:55:26.440 | And remember, they had to invest ahead

00:55:29.240 | of the revenues coming online.

00:55:30.800 | You have to buy that capability and build that capability

00:55:33.520 | and hire those teams now.

00:55:35.400 | My suspicion is you'll start seeing

00:55:37.440 | some of this acceleration late in Q3, Q4 of this year,

00:55:42.480 | heading into next year.

00:55:44.080 | And when that happens, people will say, oh, no.

00:55:46.640 | Software is not dead.

00:55:48.280 | It, in fact, is an accelerant.

00:55:50.240 | But it's not going to be equally good for all companies.

00:55:54.080 | I do think there are companies like UiPath,

00:55:56.720 | where the fundamental value proposition is challenged.

00:55:59.480 | They have to reinvent the model.

00:56:01.080 | And I think there are other businesses that's

00:56:02.960 | going to be an accelerant to the core business.

00:56:04.960 | I will take the other side of this, what you just said,

00:56:10.600 | to a certain extent, which is I don't think you can

00:56:13.480 | simultaneously have people get re-optimistic about software

00:56:18.600 | and the hype cycle of AI to continue at the pace it's been

00:56:22.040 | on, because I think they're at odds with one another.

00:56:24.520 | Because the most glorious statements about what AI is

00:56:30.320 | and can do say it replaces everything.

00:56:33.680 | And as long as that's being trumpeted and believed

00:56:37.320 | by enough people, I think the onus is on--

00:56:43.400 | everything moves from half full to half empty for anyone

00:56:46.720 | that's not holistically AI.

00:56:51.240 | Yeah, we shall see.

00:56:52.880 | That's what makes a market.

00:56:53.960 | That's what makes a podcast.

00:56:56.160 | That's what makes the basis for urinized debates

00:57:00.480 | over the last 20 years.

00:57:01.800 | I mean, if it wasn't AI disrupting something,

00:57:04.600 | it was the mobile phone disrupting desktop search

00:57:07.320 | and whether we could monetize that.

00:57:10.320 | It was the internet disrupting what came before it.

00:57:14.200 | By the way, two quick things before we leave.

00:57:17.320 | One, I look forward to tomorrow.

00:57:19.480 | It's cool you're going to be there.

00:57:20.920 | So you can let us know exactly how it goes.

00:57:22.920 | I think that'll be really interesting.

00:57:24.880 | And then I was really kind of positively moved by this.

00:57:30.360 | We can put it in the thing, but Google, Jim and I

00:57:33.720 | did an ad with Mark Cuban highlighting their enterprise

00:57:39.280 | apps, Docs, Sheets.

00:57:42.320 | And I was compelled.

00:57:48.440 | It's an area where they have not invested a lot.

00:57:51.320 | They've always had Sheets and Docs.

00:57:53.200 | And a lot of the startups in our community live on that stuff.

00:57:57.080 | But it hasn't really competed with the Microsoft stack.

00:58:00.560 | And with Satya's comeback, everyone's

00:58:03.040 | been super excited about what this means for Microsoft.

00:58:07.040 | And if they play their cards right at Google

00:58:10.360 | and tie this into Android even, it

00:58:13.920 | could be a huge win for them.

00:58:15.240 | It's fun to see this thing.

00:58:16.800 | I recommend people check it out.

00:58:19.320 | It's a-- you've got to pay for the price of admission here.

00:58:23.160 | This is an incredible time to be alive.

00:58:25.720 | All this innovation, I certainly know coming out the other end,

00:58:29.120 | it's going to yield a lot of prosperity.

00:58:32.240 | And so who the particular winners or losers are,

00:58:35.240 | what the timescale is, that's what

00:58:37.080 | we get paid to figure out.

00:58:39.000 | But I have no doubt that this is good for all of us.

00:58:42.240 | Bill, I'll see you soon.

00:58:43.240 | Take care, Michael.

00:58:44.040 | [MUSIC PLAYING]

00:58:52.440 | As a reminder to everybody, just our opinions, not

00:58:55.240 | investment advice.

Ep10. Pre-IPO Market, AI Hype Cycle, Is Software Dead? | BG2 with Bill Gurley & Brad Gerstner

Chapters