WEBVTT - Henry Blodget on the Software Selloff Hysteria and the Problem for OpenAI 0:00:00.280 --> 0:00:08.119 Bloomberg Audio Studios, Podcasts, radio news. 0:00:08.600 --> 0:00:13.840 Hello, Hello, Hello, welcome to on Air Fast twenty twenty six. 0:00:14.720 --> 0:00:17.240 Thank you so much. I'm Kristin Meinzer. I'm your MC 0:00:17.560 --> 0:00:20.360 for the day. And in case you don't know who 0:00:20.360 --> 0:00:22.800 I am, I'm a podcaster, I'm a royal watcher. I'm 0:00:22.840 --> 0:00:25.000 a culture critic. I've been very busy over the last 0:00:25.000 --> 0:00:26.920 week with some royal news you may have heard of, 0:00:27.320 --> 0:00:30.680 so you may know me from those things. But today 0:00:30.760 --> 0:00:33.360 is not about me. It is about the fantastic talent 0:00:33.440 --> 0:00:35.680 that is going to be on the stage and at 0:00:35.680 --> 0:00:39.400 this festival. Let's get to our first live taping of 0:00:39.520 --> 0:00:44.040 the day, Bloomberg's Odd Lots podcast. The hosts of Odd 0:00:44.080 --> 0:00:48.040 Lots are Joe Wisenthal and Tracy Alloway. Please welcome them 0:00:48.040 --> 0:00:53.160 to the stage, along with their very special guest, Henry Blodgett. 0:00:53.400 --> 0:01:03.000 Thank you so much. Welcome to the stage. 0:01:03.560 --> 0:01:04.399 Thank you so much. 0:01:04.440 --> 0:01:07.240 And hello and welcome to a live episode of the 0:01:07.280 --> 0:01:08.200 Odd Lots podcast. 0:01:08.240 --> 0:01:10.280 I'm Joe Wisenthal and I'm Tracy Alloway. 0:01:10.680 --> 0:01:14.200 Tracy, We're definitely not going to do this, but it 0:01:14.280 --> 0:01:16.000 definitely feels like almost. 0:01:15.800 --> 0:01:18.120 Every episode could be about AI these days. Right, Like 0:01:18.160 --> 0:01:18.759 it's just touching. 0:01:18.840 --> 0:01:21.199 So all the things we talk about are solely getting 0:01:21.400 --> 0:01:24.160 subsumed into this topic that we're not going to do it, 0:01:24.160 --> 0:01:24.920 but we probably could. 0:01:25.040 --> 0:01:27.559 It's not just that every episode could be on AI, 0:01:27.680 --> 0:01:30.160 it's that every episode could probably be made by AI 0:01:30.240 --> 0:01:32.120 at this point, right. And in fact, we've had a 0:01:32.160 --> 0:01:34.320 few people last year. I remember there were a few 0:01:34.319 --> 0:01:38.000 people who made AI generated episodes of the podcast and 0:01:38.040 --> 0:01:40.040 they weren't that bad. And I have to think if 0:01:40.040 --> 0:01:42.360 they did it today, they would be even better. 0:01:42.600 --> 0:01:45.479 It would definitely be better anyway, not better than us, 0:01:45.600 --> 0:01:47.840 but they would be better than they were. I'm not 0:01:48.040 --> 0:01:50.160 sure they'd be better than they were a year ago anyway. 0:01:50.320 --> 0:01:52.760 So we have to talk about AI and what it 0:01:52.840 --> 0:01:55.520 is doing to the market, finance, economics, et cetera. And 0:01:55.560 --> 0:01:57.440 of course we have to talk about what it means 0:01:57.480 --> 0:01:59.880 for us in the media industry. And I'm really excited 0:01:59.880 --> 0:02:02.080 to say we do have the perfect guest. 0:02:02.120 --> 0:02:02.600 We're going to be. 0:02:02.600 --> 0:02:05.720 Speaking here with Henry Blodgett, the CEO of Regenerator. He 0:02:05.840 --> 0:02:07.360 used to be my boss when I was at a 0:02:07.480 --> 0:02:11.720 Business Insider, someone who has a background in both media 0:02:11.880 --> 0:02:14.519 and of course Wall Street and finance, so the perfect 0:02:14.560 --> 0:02:17.079 person to talk about all the different dimensions. 0:02:17.120 --> 0:02:18.120 Truly the perfect guest. 0:02:18.160 --> 0:02:20.400 All right, Henry, thank you so much for coming back 0:02:20.440 --> 0:02:21.040 on odlots. 0:02:21.160 --> 0:02:22.680 Thank you for having me so. 0:02:24.400 --> 0:02:26.360 The last time we talked to was in May of 0:02:26.440 --> 0:02:29.600 last year, and the theme then was everyone was talking 0:02:29.600 --> 0:02:33.160 about here are these huge evaluations for AI companies, how 0:02:33.200 --> 0:02:35.520 could they possibly be justified? And that was an interesting 0:02:35.520 --> 0:02:39.600 conversation And it feels like the narrative has now flipped 0:02:39.600 --> 0:02:42.000 one hundred and eighty degrees where it's like these companies 0:02:42.040 --> 0:02:44.280 are so powerful they're going to destroy every other company 0:02:44.320 --> 0:02:47.200 they're wake and every software company, a payments provider, et cetera. 0:02:47.440 --> 0:02:49.680 Has been like plunging really since the start of the year. 0:02:50.280 --> 0:02:52.280 How have you like updated your views? Where are you 0:02:52.320 --> 0:02:54.040 now versus where you were in same May. 0:02:54.400 --> 0:02:56.560 We are early in the development of the industry. This 0:02:56.639 --> 0:02:59.320 looks a lot like the Internet in the nineteen nineties 0:02:59.639 --> 0:03:02.560 for the if you remember, basically, when you have a 0:03:02.639 --> 0:03:06.560 huge opportunity like this early on, you can have a 0:03:06.680 --> 0:03:10.880 huge range of reasonable predictions about what is going to happen. 0:03:11.120 --> 0:03:14.240 And here we one day it's catastrophe. None of us 0:03:14.240 --> 0:03:16.120 are ever going to have a job. Again, everything's going 0:03:16.160 --> 0:03:18.320 to create her other days, Hey, it's just a fancy 0:03:18.320 --> 0:03:21.560 word process or no worries. Open ai is way overvalued. 0:03:21.760 --> 0:03:24.519 I will say that since we talked last I believe 0:03:24.560 --> 0:03:28.400 that we were aghast at the open ai three hundred 0:03:28.400 --> 0:03:31.200 billion dollar valuation. I believe the latest one is eight 0:03:31.280 --> 0:03:34.399 hundred billion, So we are climbing. But there are other 0:03:34.440 --> 0:03:36.760 companies out there that are worth many times that. So 0:03:36.800 --> 0:03:39.440 if one does think that AI is going to take 0:03:39.480 --> 0:03:41.720 over the world and open Ai is the Google of 0:03:41.760 --> 0:03:44.680 the AI era, which is the thesis which I think 0:03:44.760 --> 0:03:48.000 is very misguided, we can get into, then one can 0:03:48.080 --> 0:03:51.200 imagine that a company might be worth many multiples of 0:03:51.360 --> 0:03:52.720 some of the companies that are out there. 0:03:53.040 --> 0:03:55.520 So one of the crazy things that happened this week 0:03:55.600 --> 0:03:58.880 speaking of AI and media, is there's a guy. We've 0:03:58.880 --> 0:04:02.000 had him on the show, James Galen aka Satriny. He 0:04:02.080 --> 0:04:05.520 runs a hedge fund also publishes an investment newsletter. He 0:04:05.600 --> 0:04:08.920 basically wrote a doom case scenario of the year twenty 0:04:09.000 --> 0:04:12.800 twenty eight where everything had become AI and because everything 0:04:12.880 --> 0:04:15.360 was AI, no one had any jobs, no one was 0:04:15.400 --> 0:04:19.240 actually buying anything in the economy, and we basically had 0:04:19.279 --> 0:04:23.560 an economic and financial crisis when I read that piece 0:04:23.640 --> 0:04:26.919 and then saw the market reaction to it, so stocks 0:04:26.960 --> 0:04:30.520 fell pretty dramatically, and people were attributing it to basically 0:04:30.560 --> 0:04:33.760 a newsletter that was a thought piece. I thought that 0:04:33.800 --> 0:04:36.159 was kind of insane, and it speaks to what you 0:04:36.320 --> 0:04:38.800 just said, which is, when we're in this really uncertain time, 0:04:39.120 --> 0:04:42.880 you can go from sort of euphoria to doomerism very 0:04:42.960 --> 0:04:43.520 very quickly. 0:04:43.720 --> 0:04:46.400 Exactly because it's all predictions. We're all looking into the 0:04:46.400 --> 0:04:48.640 hazy future saying how big is it going to be? 0:04:49.200 --> 0:04:52.719 These small changes in your assumptions make a huge difference 0:04:52.760 --> 0:04:55.480 in the current value. I will say, because that research 0:04:55.480 --> 0:05:00.279 report was both heralded and pillarid, this is exactly fully 0:05:00.360 --> 0:05:02.800 what you want analysts to do. You want them to 0:05:02.839 --> 0:05:05.400 look out and take a strong position and say here's 0:05:05.400 --> 0:05:07.440 what it can look like. And I thought it was 0:05:07.440 --> 0:05:10.960 a fascinating report. I will point out that even though 0:05:11.000 --> 0:05:14.280 as you just characterized it, that this is economic armageddon, 0:05:14.560 --> 0:05:17.000 one of the first sentences of the report says ten 0:05:17.040 --> 0:05:21.400 percent unemployment, which means ninety percent employment. Most of us 0:05:21.440 --> 0:05:24.760 still have a job, and yes there are changes in industries, 0:05:24.800 --> 0:05:27.000 but most of us are still working. So I was 0:05:27.040 --> 0:05:32.600 surprised after hearing about this predicts catastrophe that I picked 0:05:32.600 --> 0:05:34.800 it up. It's like, well, no, it predicts a big 0:05:34.839 --> 0:05:38.240 technology change that sometimes takes the economy awhile to digest, 0:05:38.560 --> 0:05:39.800 but not armageddon. 0:05:40.240 --> 0:05:42.200 What do you think though, it says like about the 0:05:42.240 --> 0:05:45.080 market right now? I mean, and it's kind of an 0:05:45.120 --> 0:05:48.120 interaction of a market story in a media story, because 0:05:48.200 --> 0:05:51.040 here's a substack piece that really did seem to move 0:05:51.120 --> 0:05:51.720 the market. 0:05:51.560 --> 0:05:53.279 Quite a bit on Monday, right yeah. 0:05:53.360 --> 0:05:56.640 What does it say about the market environment that, because 0:05:56.640 --> 0:05:59.599 it's all your right, the software stocks have been getting 0:05:59.600 --> 0:06:03.040 clovered all year, and our AI agents, I mean, we'll 0:06:03.040 --> 0:06:05.039 get into that, gonna somehow make it so we don't 0:06:05.040 --> 0:06:07.640 need to buy regular business software anymore. I don't totally 0:06:07.760 --> 0:06:10.080 understand that, but this has been the fear all year. 0:06:10.200 --> 0:06:11.920 What does it sort of say about the market slash 0:06:11.960 --> 0:06:15.000 media environment that a substack piece like this could just 0:06:15.120 --> 0:06:17.640 go absolutely megaviral like that and move the market. 0:06:17.880 --> 0:06:22.840 Everybody is twitchy. Valuations were high to begin with. We're 0:06:22.880 --> 0:06:27.040 all reevaluating our views every day saying, am I crazy 0:06:27.080 --> 0:06:28.960 thinking it could be this? Maybe I'm crazy? I have 0:06:29.000 --> 0:06:32.719 to panic a little bit so we've seen stocks roll over. 0:06:32.800 --> 0:06:36.640 The multiples are compressing a little bit again. So far 0:06:36.680 --> 0:06:39.400 from armageddon. It's people saying, okay, there may be some 0:06:39.600 --> 0:06:42.880 change here. And to your point, the big Bayer case 0:06:43.120 --> 0:06:46.159 now is that, oh my goodness, claud Code is so good, 0:06:46.520 --> 0:06:49.279 no one will ever buy software anymore. You'll just say, hey, 0:06:49.320 --> 0:06:50.200 make me the software. 0:06:50.279 --> 0:06:51.640 Yeah, I've done that. 0:06:51.720 --> 0:06:54.680 Yeah you have. You're a big experimenter. Yeah, but I 0:06:54.720 --> 0:06:57.239 would say you were probably not spending billions of dollars 0:06:57.279 --> 0:06:58.960 in enterprise software before I do. 0:06:59.520 --> 0:07:01.479 I don't know a thousands of people out in my 0:07:01.560 --> 0:07:04.159 payroll who they need to get it at Fried you 0:07:04.200 --> 0:07:04.960 know exactly. 0:07:05.000 --> 0:07:07.080 You depend on this. And one of the nice things 0:07:07.120 --> 0:07:09.720 about having a company standing behind a product is you 0:07:09.760 --> 0:07:12.440 actually can hold them accountable for it. And maybe claud 0:07:12.520 --> 0:07:15.200 Code doesn't make your enterprise software the you why you want, 0:07:15.360 --> 0:07:18.240 somebody doesn't get paid. So as I hear that, to me, 0:07:18.320 --> 0:07:22.640 that is hysteria. I don't understand it. Software companies are 0:07:22.720 --> 0:07:26.040 not going away. Companies that are not in the software 0:07:26.080 --> 0:07:29.360 business are not going to suddenly say, oh, let's have 0:07:29.760 --> 0:07:33.000 a junior person who understands AI build all our software 0:07:33.040 --> 0:07:36.360 for us. It is not going to happen, but overall, 0:07:36.720 --> 0:07:40.160 big changes coming, lots of disruption. Do we know for 0:07:40.280 --> 0:07:42.600 sure what that's going to do to profits of these 0:07:42.640 --> 0:07:45.800 companies and change in the economy. We do not, And 0:07:45.840 --> 0:07:49.160 that is called the discount rate getting higher. It feels riskier. 0:07:49.440 --> 0:07:54.280 And on that day, Monday, everybody was feeling suddenly risk averse. 0:07:54.440 --> 0:07:58.040 Today I'll point out as we tape, everyone's saying, no problem. 0:07:58.400 --> 0:07:59.440 Off to the races again. 0:08:00.200 --> 0:08:03.840 You've obviously lived and been very active and influential in 0:08:03.880 --> 0:08:08.240 a previous technological I guess cycle of disruption. What do 0:08:08.280 --> 0:08:11.280 you think people are getting wrong here, especially on Wall 0:08:11.280 --> 0:08:14.160 Street when it comes to their approach on this particular 0:08:14.200 --> 0:08:15.800 one AI versus dot Com. 0:08:16.200 --> 0:08:19.400 I think this goes way beyond Wall Street. I think 0:08:19.400 --> 0:08:22.880 there really is a big camp still, and the case 0:08:22.880 --> 0:08:24.680 gets made and then it gets tamped down, and then 0:08:24.720 --> 0:08:27.200 it comes back even more strongly that there really will 0:08:27.240 --> 0:08:30.720 be no jobs for humans anymore. And what I would 0:08:30.720 --> 0:08:32.559 say to that is, I think the bigger risk with 0:08:32.679 --> 0:08:38.440 AI is actually that somebody invents an agenttic system that 0:08:38.600 --> 0:08:40.520 actually gets out in the world and does a lot 0:08:40.559 --> 0:08:43.120 of damage. I think that is a bigger risk. If 0:08:43.160 --> 0:08:46.920 you look at technology job transitions over time, like the 0:08:47.000 --> 0:08:51.760 big one, which is agriculture to industrial. Two hundred years ago, 0:08:51.960 --> 0:08:54.720 ninety three percent of us worked on farms. Now it's 0:08:54.720 --> 0:08:58.400 two percent. That's a big transition, and yet all through 0:08:58.440 --> 0:09:01.560 that the number of jobs grow grew. And same thing 0:09:01.640 --> 0:09:05.360 with things like telephone operators with steam engines. You look 0:09:05.400 --> 0:09:08.559 at any piece of it, there is disruption and pain 0:09:08.800 --> 0:09:11.480 and I'm not minimizing that for the people whose jobs 0:09:11.520 --> 0:09:15.240 are changed or disrupted by it. But the economy overall 0:09:15.520 --> 0:09:18.440 so far in history has gone on to create a 0:09:18.440 --> 0:09:21.480 lot more jobs. And so when you see the lay 0:09:21.520 --> 0:09:25.000 waste to jobs predictions, you don't also often get the 0:09:25.040 --> 0:09:26.760 other side, which is how many new jobs are going 0:09:26.800 --> 0:09:27.320 to be created? 0:09:27.640 --> 0:09:29.680 Well, this is exactly what I want to ask you, 0:09:29.720 --> 0:09:33.960 because we hear that all the time during technological advancement cycles, 0:09:34.440 --> 0:09:36.640 new jobs are created. But I think the difficulty with 0:09:36.720 --> 0:09:39.640 AI is that it's hard to see what those new 0:09:39.720 --> 0:09:40.960 jobs are we better are. 0:09:40.760 --> 0:09:41.080 Going to be? 0:09:41.160 --> 0:09:43.520 Yeah, like, what are we better at than a supercomputer 0:09:43.679 --> 0:09:47.000 that can code a program or write an article in 0:09:47.080 --> 0:09:48.079 less than a minute. 0:09:47.880 --> 0:09:49.679 Or mimical voice and do a podcast. 0:09:49.800 --> 0:09:53.679 Okay, so how where are we three years in at 0:09:53.679 --> 0:09:57.680 this point. These job doom predictions started twenty minutes after 0:09:57.760 --> 0:10:02.800 chat GPT came out. Where are we maybe a little 0:10:02.800 --> 0:10:05.640 bit at the margin, we're seeing something maybe, but maybe 0:10:05.640 --> 0:10:08.440 it's still that companies way over hired after the COVID 0:10:08.480 --> 0:10:12.240 boom or now sort of right sizing themselves. As you say, 0:10:12.600 --> 0:10:15.560 I can go into LA, I can go to Google. 0:10:15.640 --> 0:10:18.080 I think it is drop in a topic and say 0:10:18.160 --> 0:10:20.600 make this into a great Odd Lots podcast. 0:10:21.880 --> 0:10:23.000 Some people have said that Google. 0:10:23.480 --> 0:10:26.600 Multiple people think that Google's notebook LM, which does that 0:10:27.040 --> 0:10:28.560 creates a podcast out of a document. 0:10:28.640 --> 0:10:30.120 Multiple people have said they. 0:10:30.000 --> 0:10:32.400 Trained this on Jo and Tracy like this sounds like 0:10:32.960 --> 0:10:35.680 other people are claiming that too, but yes they could have. 0:10:35.920 --> 0:10:39.240 I'm just saying, you guys continue to grow. All I 0:10:39.280 --> 0:10:41.160 hear more and more about is oh, you gotta listen 0:10:41.160 --> 0:10:43.400 to Odd Lots grow and growing, growing, growing, growing. And 0:10:43.440 --> 0:10:46.600 I don't think we were talking earlier that you've radically changed, 0:10:46.640 --> 0:10:48.640 you made the way you make the show. We're still 0:10:48.679 --> 0:10:52.960 recording it here. You didn't just queue up lams. Who 0:10:53.280 --> 0:10:57.680 is going to rush to do the big interview with Google. 0:10:58.240 --> 0:11:00.439 They're not. They're gonna come do it with you guys, 0:11:00.480 --> 0:11:03.360 because they love you, and your audience loves you. So 0:11:03.400 --> 0:11:07.600 I'm just saying, even in something that seemingly looks so 0:11:07.960 --> 0:11:11.120 ripe to be destroyed, you guys cost money. Why is 0:11:11.120 --> 0:11:19.439 Bloomberg spending it away right to defend I'm just saying optimistic. 0:11:19.559 --> 0:11:21.520 So I mean I hear both ways, and thank you 0:11:21.559 --> 0:11:22.120 for saying that. 0:11:22.160 --> 0:11:24.000 On the other hand, you know, people might not have 0:11:24.040 --> 0:11:27.560 the same affinity for someone who does claims adjusted and 0:11:27.559 --> 0:11:29.280 at insurance. They don't really care if they have the 0:11:29.360 --> 0:11:32.240 human touch for that or tech support. I want to 0:11:32.240 --> 0:11:35.320 ask a related question. It also relates to the piece 0:11:35.360 --> 0:11:37.320 that came out on one day, and it relates to 0:11:37.360 --> 0:11:39.520 this question of whether open ai is could be. 0:11:39.440 --> 0:11:40.960 The Google of the next generation. 0:11:41.520 --> 0:11:44.440 One of the arguments that the piece stipulates, and I 0:11:44.440 --> 0:11:46.760 think it sort of fits with what I have observed, 0:11:47.240 --> 0:11:49.719 is that ai doesn't seem to have network. 0:11:49.480 --> 0:11:50.680 Effects in the same way. 0:11:50.720 --> 0:11:53.240 So it's like you could switch from O chat, GPT 0:11:53.840 --> 0:11:56.560 to claude with almost no issues and it's just not 0:11:56.720 --> 0:11:59.520 an issue. Whereas after the first time I used Google 0:11:59.559 --> 0:12:03.040 in two though I never used Yahoo again, that was 0:12:03.080 --> 0:12:04.440 the last time I'd ever. 0:12:04.440 --> 0:12:05.680 Used another search engine. 0:12:06.320 --> 0:12:08.280 This seems this is one of the arguments in the 0:12:08.280 --> 0:12:10.400 piece that there isn't gonna be the same stickiness and 0:12:10.480 --> 0:12:13.480 network effects in the AI era in the way that 0:12:13.600 --> 0:12:16.760 defined the last twenty five years of the Internet. And 0:12:16.840 --> 0:12:19.960 so when you think about like open ai versus Google 0:12:20.000 --> 0:12:22.560 this year versus the dot com era, does that resonate 0:12:22.600 --> 0:12:23.800 to I. 0:12:23.720 --> 0:12:27.040 Think that what folks who would say that open ai 0:12:27.240 --> 0:12:29.160 is the Google of the era would say that more 0:12:29.200 --> 0:12:31.640 and more, you're gonna build more of your life into it. 0:12:31.640 --> 0:12:33.640 It's going to know everything about you, it's going to 0:12:33.720 --> 0:12:36.040 custom so it's gonna be the switching costs are going 0:12:36.080 --> 0:12:38.320 to grow. But to this point they have not. And 0:12:38.360 --> 0:12:40.640 I'll tell you the thing that really made me think 0:12:40.679 --> 0:12:43.320 they were in trouble, and I still think that given 0:12:43.360 --> 0:12:46.640 the valuation and given the general consensus that they've already won, 0:12:47.440 --> 0:12:49.880 was when Google Gemini came out and people said, oh 0:12:50.200 --> 0:12:54.199 it's better. Yeah, they passed them, and all the bulls said, oh, 0:12:54.440 --> 0:12:57.120 just wait till the next version of open ai comes out. 0:12:57.360 --> 0:12:59.480 I will say, if you go back into the nineteen nineties, 0:13:00.360 --> 0:13:02.920 Amazon was very controversial. A lot of people thought it 0:13:02.920 --> 0:13:06.120 would go bankrupt. It's just a bookstore. Barnes and Noble 0:13:06.200 --> 0:13:07.880 is gonna put them out of business as soon as 0:13:07.880 --> 0:13:11.280 they have a website or Walmart or whatever. Walmart and 0:13:11.320 --> 0:13:16.000 Barnes and Noble never got anywhere near Amazon. Amazon was 0:13:16.120 --> 0:13:19.840 so far out ahead always and still almost went bankrupt 0:13:20.000 --> 0:13:23.200 by the way, but nobody was ever close. And in 0:13:23.240 --> 0:13:26.400 this case, less than two years after the product comes out, 0:13:26.679 --> 0:13:31.280 the incumbent has caught up to me. That's when I said, like, whoa, Okay, 0:13:31.640 --> 0:13:33.800 maybe this actually is going to be like the Internet, 0:13:34.240 --> 0:13:37.760 which is the first five years all of us, including me, 0:13:38.080 --> 0:13:41.680 were saying like, oh, just buy the leaders, no problem, 0:13:41.880 --> 0:13:44.320 like they're the ones that are gonna survive. You know 0:13:44.360 --> 0:13:47.720 how many survived. Of that first five hundred companies that 0:13:47.760 --> 0:13:52.559 went public in the nineteen nineties, one went on to 0:13:52.800 --> 0:13:56.000 actually make investors a lot of money after the crash. 0:13:56.080 --> 0:13:59.840 That was Amazon. Two others that I know of, Cisco, 0:14:00.280 --> 0:14:04.480 after twenty five years, finally got back to its stock prize. 0:14:04.960 --> 0:14:08.240 eBay did okay for a while and then kind of 0:14:08.240 --> 0:14:10.600 sputtered out a little bit. That's out of hundreds and 0:14:10.679 --> 0:14:14.840 hundreds of companies. So I think it's very possible that yes, 0:14:15.160 --> 0:14:18.600 AI is huge. It's a great idea. We all want it, 0:14:18.720 --> 0:14:21.480 and somebody is going to come along soon and due 0:14:21.480 --> 0:14:24.120 to open ai what Google did to Yahoo, which by 0:14:24.160 --> 0:14:27.120 the way, Yahoo was the big search engine winner and 0:14:27.160 --> 0:14:30.200 then Google destroyed it. So I just think it's way 0:14:30.240 --> 0:14:45.760 too early to say that we open a eyes the Google. 0:14:49.080 --> 0:14:52.760 We started by mentioning the open ai valuation, which everyone 0:14:53.440 --> 0:14:56.560 was a gas stat last year. Now maybe not so much. 0:14:56.600 --> 0:14:59.760 But even if you assume a future where open ai 0:14:59.880 --> 0:15:03.440 or whoever comes in and disrupts every single industry, and 0:15:03.560 --> 0:15:07.120 there are thousands of employees and companies that are paying 0:15:07.120 --> 0:15:10.640 two hundred dollars a month or whatever to access the 0:15:10.680 --> 0:15:14.520 new LM whatever the new model is, open ai, at 0:15:14.600 --> 0:15:17.040 least for right now, is still losing money on every 0:15:17.080 --> 0:15:19.720 customer that's like a power user. So how does that 0:15:19.760 --> 0:15:22.000 actually translate into a working business model. 0:15:22.360 --> 0:15:25.280 It's a good question. I think what the bulls would 0:15:25.280 --> 0:15:28.640 say is the cost of what they do is going 0:15:28.680 --> 0:15:31.520 to plummet, and so the lines are going to cross 0:15:31.520 --> 0:15:34.080 here pretty soon. If they own the whole market, suddenly 0:15:34.160 --> 0:15:37.520 they start to become incredibly profitable. We are not there yet. 0:15:37.600 --> 0:15:39.760 And I think when we talked last time at a 0:15:39.840 --> 0:15:43.200 three hundred billion dollar valuation, which everybody was horrified by 0:15:43.200 --> 0:15:45.440 at that point. But if you looked at the projections, 0:15:45.560 --> 0:15:48.640 the projections were three years from now, they'll do one 0:15:48.760 --> 0:15:52.120 hundred billion of revenue. And if you at that point 0:15:52.160 --> 0:15:54.520 were looking at it saying, okay, if I believe that, 0:15:54.720 --> 0:15:58.080 then it's three times revenue, the economics will probably change. 0:15:58.200 --> 0:16:00.320 It starts to look a little more reasonable. Now we're 0:16:00.320 --> 0:16:03.120 at eight hundred billion, so it's a much bigger multiple, 0:16:03.280 --> 0:16:06.320 and you've got to actually have a much bigger revenue 0:16:06.360 --> 0:16:09.640 stream and profits. But that's the argument. I don't see 0:16:09.680 --> 0:16:13.680 it yet. I gotta say the economics are incredibly difficult 0:16:13.880 --> 0:16:14.240 for them. 0:16:14.440 --> 0:16:16.520 They seem brutal. And there's another element too. 0:16:16.560 --> 0:16:20.760 So obviously, unlike during the Web two point zero era 0:16:20.840 --> 0:16:23.480 where there was like cheap, plentiful compute, obviously, the capital 0:16:23.480 --> 0:16:24.720 expenditure budgets. 0:16:24.360 --> 0:16:27.480 To these companies are just eyewatering. They're spending like crazy. 0:16:27.600 --> 0:16:30.080 There's also this other element that I think is interesting, 0:16:30.320 --> 0:16:33.760 which is at least of a few of these labs, 0:16:34.200 --> 0:16:36.680 like the people found to them are true believers, Like 0:16:36.720 --> 0:16:39.280 they think they're like they think they're on a mission 0:16:39.640 --> 0:16:42.680 to build AGI is going to change everything. Build God, 0:16:42.800 --> 0:16:45.200 Build God, right, build a digital God, and it's like 0:16:45.240 --> 0:16:47.320 they're not going to stop at anything. They're not gonna 0:16:47.320 --> 0:16:49.280 be like, you know what, it's pretty good, let's tap 0:16:49.320 --> 0:16:51.680 the brakes. We're pretty happy with how this model works. 0:16:52.160 --> 0:16:53.640 You know, I don't, or at least I don't get 0:16:53.640 --> 0:16:54.240 that a privy. 0:16:54.400 --> 0:16:57.920 Also, they've couched it as this like existential race, like 0:16:57.920 --> 0:16:59.880 there's only gonna be one company that dominates in the 0:17:00.080 --> 0:17:03.160 and so the upper limit on spending to become that 0:17:03.240 --> 0:17:05.520 company is infinity. 0:17:06.000 --> 0:17:10.360 Yes, However, there is a limit to money. And when 0:17:10.359 --> 0:17:13.359 you look at when people are raising money, you watch 0:17:13.440 --> 0:17:16.880 where they're going for it. Sure, the vcs, that's one 0:17:16.920 --> 0:17:19.080 pool of money, and then the big strategics, and then 0:17:19.119 --> 0:17:21.160 the vcs who missed the first time and are feeling 0:17:21.160 --> 0:17:22.960 so embarrassed about it they'll pay up to get in. 0:17:23.240 --> 0:17:26.320 Then it is the international funds that are watching all 0:17:26.359 --> 0:17:29.320 this stuff. Yeah, exactly. So it's some point you run 0:17:29.359 --> 0:17:31.359 out of money, and we are talking about a company 0:17:31.400 --> 0:17:35.240 that is raising hundreds of billions of dollars just to 0:17:35.320 --> 0:17:38.560 keep the engine running and just to keep pace. This 0:17:38.640 --> 0:17:44.120 is the other problem. Google, Facebook, Microsoft are generating tens 0:17:44.160 --> 0:17:46.680 of billions of dollars of free cash flow every year 0:17:46.800 --> 0:17:49.600 that they can use to buy chips open AI has 0:17:49.600 --> 0:17:51.880 to raise it in the open market. So if those 0:17:51.920 --> 0:17:55.919 economics don't start to change soon, we're gonna run out 0:17:55.960 --> 0:17:56.280 of money. 0:17:56.720 --> 0:17:58.480 Joe and I were talking about this before we came 0:17:58.520 --> 0:18:01.440 on stage. But another thing happened recently was there is 0:18:01.480 --> 0:18:05.440 a karaoke machine making company that announced it was now 0:18:05.560 --> 0:18:09.119 an AI company, which when I see headlines like that, 0:18:09.480 --> 0:18:12.000 remind me a lot of the crypto boom, but also 0:18:12.040 --> 0:18:14.000 remind me a lot of the dot com boom, and 0:18:14.040 --> 0:18:15.640 we are starting to see more and more of those. 0:18:15.720 --> 0:18:17.360 Wasn't there a toilet company recently? 0:18:17.720 --> 0:18:20.520 The toilet owner is legit, so like apparently, no, no, no, 0:18:20.600 --> 0:18:23.120 for real, So you said it was not like, oh, 0:18:23.119 --> 0:18:26.040 we're an AI company. Apparently it's the Japanese toilet company 0:18:26.040 --> 0:18:28.239 and they have some like porcelain or something like that. 0:18:28.320 --> 0:18:31.480 It's very important for like semiconductors or something. So everyone's like, oh, 0:18:31.520 --> 0:18:33.680 get out of the toilet business. You have this access 0:18:33.680 --> 0:18:34.640 to this material. 0:18:34.920 --> 0:18:38.760 But so that one was kind of legit. The karaoke one. 0:18:39.000 --> 0:18:42.840 Okay, okay, so yeah, the toilet one. Maybe the karaoke one. 0:18:42.880 --> 0:18:44.840 When you see headlines like that, what do you think 0:18:44.880 --> 0:18:47.440 do you think like this is a rational pivot into 0:18:47.480 --> 0:18:50.840 a booming market and everyone should be experimenting with AI 0:18:50.920 --> 0:18:52.560 at the moment, or do you think like, nah, this 0:18:52.680 --> 0:18:53.399 is getting crazy. 0:18:53.760 --> 0:18:58.040 I think we are in a euphoric bubble period where 0:18:58.040 --> 0:19:01.879 we just discovered this amazing new thing. Nobody knows what 0:19:01.920 --> 0:19:04.520 the future is, nobody knows what's gonna work, and we 0:19:04.560 --> 0:19:09.320 are effectively creating this enormous R and D lab, which is, hey, 0:19:09.600 --> 0:19:12.240 what you can do something, here's some money, go out 0:19:12.240 --> 0:19:14.120 and experiment. I hope you'll be one of the winners. 0:19:14.560 --> 0:19:17.199 Fifty to one hundred years ago, all of that stuff 0:19:17.280 --> 0:19:21.960 happened in labs like Edison's lab or Bell Labs inside 0:19:21.960 --> 0:19:25.600 these big corporations. You never saw it. Now everything is 0:19:25.600 --> 0:19:28.600 a startup. You see it. It's happening in front of you. 0:19:28.680 --> 0:19:31.199 But there's no question that most of the companies are 0:19:31.240 --> 0:19:33.600 gonna fail. Happens aget in again and again. And I 0:19:33.640 --> 0:19:36.280 think what investors are saying when they see that is okay, 0:19:36.520 --> 0:19:39.000 you know that is a quick flip. It's hey, I'll 0:19:39.040 --> 0:19:41.240 get it. Other people. A lot of other people love AI. 0:19:41.280 --> 0:19:43.000 They'll buy it after I do, and it'll go up 0:19:43.040 --> 0:19:43.760