WEBVTT - D.A. Wallach Explains Why Biotech VC Is So Different 0:00:00.640 --> 0:00:03.320 Hey, they're aud Loots listeners. It is that time of 0:00:03.360 --> 0:00:06.600 the year again. We are going to be doing a 0:00:06.720 --> 0:00:09.879 call in show on the podcast. You can ask us 0:00:10.200 --> 0:00:13.119 any of your burning questions. 0:00:12.720 --> 0:00:15.600 That's right. You want to ask us about finance, markets 0:00:15.640 --> 0:00:18.759 and economics, go for it. You want to ask us 0:00:18.880 --> 0:00:22.400 about the year in podcasting, go for it. You want 0:00:22.400 --> 0:00:25.880 to ask about where Tracy is with growing chicken or 0:00:26.040 --> 0:00:31.000 raising chicken, growing chickens, raising chickens on her versioning firm 0:00:31.040 --> 0:00:33.360 in Connecticut, go for it. This is your chance to 0:00:33.400 --> 0:00:34.120 ask us anything. 0:00:34.360 --> 0:00:36.479 Joe's favorite cut of steak. 0:00:36.240 --> 0:00:38.760 Right, my favorite cut of steak, my favorite Chinese restaurant 0:00:38.760 --> 0:00:40.880 in the East Village. It's all fair game, all right. 0:00:41.280 --> 0:00:43.560 All you have to do is send a voice memo 0:00:43.760 --> 0:00:47.400 with your question, your name, your age, and your location 0:00:47.600 --> 0:00:50.720 to od Loots at Bloomberg dot net. 0:00:51.000 --> 0:00:54.120 Deadline to submit is December seventeenth, So get the men, 0:00:54.200 --> 0:00:54.880 assume as you can. 0:00:55.120 --> 0:00:57.800 We're looking forward to hearing what you have to ask 0:00:57.920 --> 0:00:59.920 and yeah, that's coming up. 0:01:02.720 --> 0:01:15.840 Bloomberg Audio Studios, Podcasts, Radio News. 0:01:18.040 --> 0:01:21.000 Hello and welcome to another episode of the Odd Lots podcast. 0:01:21.080 --> 0:01:22.440 I'm Tracy Alloway. 0:01:22.120 --> 0:01:23.800 And I'm Joe Why isn't they Joe. 0:01:23.800 --> 0:01:25.679 We were doing a Q and A this morning. 0:01:25.880 --> 0:01:26.319 That's right. 0:01:26.440 --> 0:01:28.240 It's a lot of fun go on a live Q 0:01:28.360 --> 0:01:28.560 and A. 0:01:29.160 --> 0:01:32.160 And someone asked a question about whether or not we're 0:01:32.200 --> 0:01:37.440 going to do more healthcare episodes, and no, we don't, 0:01:37.520 --> 0:01:41.880 and there's a reason for that. I personally am incredibly 0:01:41.920 --> 0:01:45.000 intimidated by the US healthcare system. I do not understand 0:01:45.040 --> 0:01:48.160 it at all. It is just a complete mystery to me. 0:01:48.920 --> 0:01:51.440 But I was very happy to say in response to 0:01:51.480 --> 0:01:55.520 that question that this same day, that's right, we're actually 0:01:55.520 --> 0:01:58.920 recording a healthcare episode with someone that we've wanted to 0:01:58.920 --> 0:02:00.480 speak to for a long time. 0:02:01.200 --> 0:02:03.720 I'm the same way in the sense that, first of all, yes, 0:02:03.880 --> 0:02:05.680 I'm the same way in the sense that I really 0:02:05.720 --> 0:02:08.760 do not know much about how the healthcare system works. 0:02:09.080 --> 0:02:11.480 I don't even know where to begin asking the right questions. 0:02:11.520 --> 0:02:13.160 There is athing you do. You have to just start 0:02:13.200 --> 0:02:15.799 a random episode, and that gives you the germs of 0:02:15.840 --> 0:02:18.400 the next question, the next episode, the next episode. But 0:02:18.440 --> 0:02:21.360 they're so it seems so big and sprawling, etc. That 0:02:21.880 --> 0:02:24.280 what is the first question to ask? So we just 0:02:24.320 --> 0:02:26.639 have to plunge right in and just pick one which 0:02:26.680 --> 0:02:29.160 we're doing now, and then maybe that will lead to 0:02:29.160 --> 0:02:32.000 the string of healthcare episodes, which we should have done 0:02:32.000 --> 0:02:32.360 a long time. 0:02:32.400 --> 0:02:34.959 Again, that's exactly right. There's also a lot of new 0:02:35.000 --> 0:02:37.639 stuff happening in healthcare at the moment, and we recorded 0:02:37.680 --> 0:02:40.720 an episode on Chinese biotechs a little while ago that 0:02:40.800 --> 0:02:44.400 was incredibly fascinating. Definitely, I'm very curious to see what's 0:02:44.440 --> 0:02:47.560 going on on the US side of biotech and nesting, 0:02:47.680 --> 0:02:50.120 and we do have another episode plan that's sort of 0:02:50.160 --> 0:02:53.280 tangentially related to that. But clearly there's a lot to 0:02:53.280 --> 0:02:57.200 talk about. The other very odd lotsy thing with this 0:02:57.280 --> 0:03:01.359 guest is we like people who have interest in career histories, right. 0:03:01.440 --> 0:03:04.320 That's right. How how they got to where they are 0:03:04.360 --> 0:03:08.640 today is often a very interesting question. You know, I 0:03:08.680 --> 0:03:10.840 have nothing against people who just took the normal path, 0:03:10.880 --> 0:03:13.440 people who just sort of, you know, went to college 0:03:13.560 --> 0:03:14.760 and they got their NBA. 0:03:14.800 --> 0:03:16.600 And okay, you forgive them, I forgive them. 0:03:16.600 --> 0:03:18.960 That's totally fine. But it's also interesting to hear about 0:03:19.040 --> 0:03:21.480 the people who maybe walked in through the side door, 0:03:21.520 --> 0:03:22.079 so to speak. 0:03:22.120 --> 0:03:24.440 Absolutely so, we do, in fact have the perfect guest. 0:03:24.480 --> 0:03:26.760 We have someone who has a lot of thoughts on 0:03:27.000 --> 0:03:30.560 US healthcare and who is also a biotech investor and 0:03:30.760 --> 0:03:37.040 also formerly the lead singer of Chester French. So Da Wallack, 0:03:37.120 --> 0:03:38.960 Welcome to the show. Thanks so much for coming on. 0:03:39.280 --> 0:03:43.080 Thanks for having me, guys. I'm I'm a odd lots junkie. 0:03:43.160 --> 0:03:45.920 So this is like going to the Grammys. 0:03:45.960 --> 0:03:51.040 Amazing, Thank you ever Grammy. No, don't sorry, sorry, sorry, 0:03:51.040 --> 0:03:52.040 I shouldn't have I shouldn't have it. 0:03:52.200 --> 0:03:53.680 Not yet is the right answer. 0:03:53.680 --> 0:03:55.120 Not yet, not yet, not yet. 0:03:55.280 --> 0:03:57.960 I guess my first question should be, can you talk 0:03:57.960 --> 0:04:01.160 to us about the through line between being musician and 0:04:01.400 --> 0:04:04.200 healthcare and how and biotech and how you got into 0:04:04.200 --> 0:04:06.360 the space, because I think you know, it's not a 0:04:06.480 --> 0:04:09.480 natural transition, to say the least. 0:04:10.360 --> 0:04:12.840 Yeah, well, I'll tell you how I ended up doing this, 0:04:13.080 --> 0:04:16.720 and then I'll try to connect them theoretically in some way. 0:04:16.760 --> 0:04:19.839 It might be a little tenuous. I've basically had three 0:04:19.839 --> 0:04:22.480 careers so far in my limited adult life. I was 0:04:22.520 --> 0:04:25.719 a professional rock musician with the band that you mentioned 0:04:25.720 --> 0:04:28.479 for several years, and then I kind of slipped into 0:04:28.480 --> 0:04:32.000 the venture capital world when I invested in Spotify about 0:04:32.560 --> 0:04:35.920 thirteen years ago, and that was pretty much the only 0:04:35.960 --> 0:04:38.120 company in the private markets I was well positioned to 0:04:38.200 --> 0:04:42.280 understand as a musician, and through the success of that, 0:04:42.360 --> 0:04:44.680 I got turned on to how exciting venture capital was. 0:04:45.120 --> 0:04:48.760 Started doing other types of investments across different industries. Was 0:04:48.800 --> 0:04:51.920 involved in SpaceX and Ripple and a bunch of interesting 0:04:51.960 --> 0:04:56.040 other startups, and then ultimately a guy I knew started 0:04:56.160 --> 0:04:59.360 an early stage a healthcare company. It was a telemedicine 0:04:59.360 --> 0:05:03.160 startup called doctor on Demand, and telemedicine at the time 0:05:03.240 --> 0:05:05.400 was not a hot topic because this is pre COVID, 0:05:05.960 --> 0:05:08.760 so we still primarily went to the doctor in person. 0:05:09.560 --> 0:05:12.919 And when I made that investment, I started to learn 0:05:12.960 --> 0:05:15.680 more and more about our healthcare system and was just 0:05:15.720 --> 0:05:18.040 blown away by how screwed up and stupid it was. 0:05:18.680 --> 0:05:22.880 And then that eventually evolved into learning more about biotechnology 0:05:23.400 --> 0:05:27.080 and the other sub sectors of healthcare that are critical 0:05:27.120 --> 0:05:30.000 to medicine, and it's ended up being what I do. 0:05:30.480 --> 0:05:32.640 In terms of the connection between music and any of 0:05:32.640 --> 0:05:34.880 this stuff, there are a couple of ways I can 0:05:34.920 --> 0:05:37.040 think about it. One is, I tell people, now my 0:05:37.200 --> 0:05:40.640 job is like being a record producer for scientists, so 0:05:40.680 --> 0:05:43.599 there's a little bit of a parallel there. But the 0:05:43.680 --> 0:05:47.960 other is that I think there's a unique challenge in 0:05:48.080 --> 0:05:52.520 music to combining art and commerce, And in healthcare there's 0:05:52.560 --> 0:05:55.040 a similar parallel challenge, which is how do you combine 0:05:55.080 --> 0:05:59.719 medicine and capitalism, which don't naturally go together very well? 0:06:00.120 --> 0:06:03.840 Producer, analogy makes a ton of sense. And you know, 0:06:03.880 --> 0:06:06.160 there are probably a lot of musicians who are really brilliant, 0:06:06.200 --> 0:06:09.320 they're really great musicians, but for whatever reason, the lightning 0:06:09.360 --> 0:06:11.520 doesn't strike where they are or doesn't strike nearby, and 0:06:11.520 --> 0:06:14.679 they don't take off. Probably many brilliant scientists, et cetera, 0:06:14.839 --> 0:06:18.840 but the path from brilliant science to commercial blockbuster can often, 0:06:18.920 --> 0:06:23.080 I assume, be tricky or dispiriting in many ways, et cetera. 0:06:23.320 --> 0:06:27.400 Biotech specifically, of all the things in investing, biotech strikes 0:06:27.480 --> 0:06:29.440 me is this whole different world than the rest of 0:06:29.520 --> 0:06:31.600 like investing. You know, when I think of like a 0:06:31.640 --> 0:06:36.440 software company, it's like, oh, okay, well they've accumulated these 0:06:36.440 --> 0:06:40.120 clients and their churn is low, et cetera. Yeah, this 0:06:40.200 --> 0:06:42.520 seems like a company that has traction is going to grow. 0:06:42.760 --> 0:06:46.160 When it comes to biotech, it's like, okay, here's some 0:06:46.839 --> 0:06:50.040 patent on a sequence and maybe ten years from now 0:06:50.080 --> 0:06:53.400 it'll get approved to something that'll be a therapy. It 0:06:53.480 --> 0:06:56.160 seems so much harder to figure out, like what are 0:06:56.200 --> 0:06:59.159 the heuristics that one would use to establish this is 0:06:59.200 --> 0:07:02.400 a likely this science is likely going to turn into 0:07:02.440 --> 0:07:02.799 a business. 0:07:02.800 --> 0:07:06.520 Oh, that's absolutely true. It's like a completely different paradigm. 0:07:06.560 --> 0:07:09.680 As an investor, I think the typical biotech company is 0:07:09.720 --> 0:07:12.920 like a bag of options, and each one of the 0:07:13.040 --> 0:07:16.200 drugs that the company is working on in success could 0:07:16.240 --> 0:07:19.360 be worth billions of dollars, but that's ten years away 0:07:19.600 --> 0:07:23.920 often minimum, and so you're trying to price things based 0:07:23.960 --> 0:07:28.320 on their ultimate potential scale times their probability of succeeding, 0:07:28.840 --> 0:07:31.760 and unfortunately, the base rates in terms of probability of 0:07:31.800 --> 0:07:35.800 success are very low. So if you take small molecules, 0:07:35.840 --> 0:07:40.360 which is one major area of drugs, the base case 0:07:40.640 --> 0:07:44.480 is like a five percent probability of success from the 0:07:44.600 --> 0:07:47.800 original idea to an FDA approval and a marketed drug. 0:07:48.280 --> 0:07:51.680 Now you get to a higher sort of prior probability 0:07:51.760 --> 0:07:57.480 with antibodies or so called biologics other classes of drugs 0:07:57.920 --> 0:08:01.800 that are intrinsically more likely to work than small molecules, 0:08:02.120 --> 0:08:04.840 but still in every case you're dealing with very low 0:08:04.880 --> 0:08:08.360 probabilities of success, and the entire challenge as a biotech 0:08:08.400 --> 0:08:11.800 investor is how do you manage those low probability events 0:08:12.200 --> 0:08:15.080 and build portfolios that are still likely to make money 0:08:15.400 --> 0:08:19.160 despite the fact that each individual project is relatively unlikely 0:08:19.200 --> 0:08:23.200 to work. I'd say, in tech, there's this well described 0:08:23.280 --> 0:08:26.760 kind of power law distribution of winners and losers, which 0:08:26.800 --> 0:08:29.440 is to say, a very small number of companies make 0:08:29.480 --> 0:08:32.040 all the money and pay for the huge number of losers. 0:08:32.640 --> 0:08:35.840 In biotech, that's still true to a degree, but the 0:08:35.880 --> 0:08:39.880 magnitudes of the winners are lower, and so a really 0:08:39.960 --> 0:08:44.000 good biotech investor probably has a lower, sorry a higher 0:08:44.080 --> 0:08:48.480 batting average than the typical tech investor, but the wins 0:08:48.520 --> 0:08:49.559 are not as big. 0:08:50.760 --> 0:08:54.040 So one thing I'm really curious about is how you 0:08:54.200 --> 0:08:58.880 source potential investments and how you find you use the 0:08:58.920 --> 0:09:02.640 analogy of the record, how you find talent in the space, 0:09:02.760 --> 0:09:05.720 or how the talent kind of finds you, and whether 0:09:05.840 --> 0:09:09.120 or not it's different from again, the sort of software 0:09:09.800 --> 0:09:12.360 or tech space that we usually talk about when it 0:09:12.360 --> 0:09:13.400 comes to venture capital. 0:09:14.360 --> 0:09:17.000 You know, when I started doing venture investing, it was, 0:09:17.040 --> 0:09:21.080 like I said, twelve thirteen years ago, it was obviously 0:09:21.120 --> 0:09:25.160 a well established part of the capital markets. But you know, 0:09:25.200 --> 0:09:28.199 I cold emailed Brian Armstrong from coinbase and was meeting 0:09:28.240 --> 0:09:33.400 with them two days later. And it's hard to overstate 0:09:34.240 --> 0:09:37.240 how much money has rushed in over the past decades. 0:09:37.320 --> 0:09:40.200 So what went from being an established but still kind 0:09:40.240 --> 0:09:44.320 of marginal part of the capital markets is now all 0:09:44.360 --> 0:09:47.960 anyone thinks or talks about. And so in biotech, what 0:09:48.040 --> 0:09:50.800 I found getting into this area was that it was 0:09:51.040 --> 0:09:53.800 more like that venture market I had encountered. There was 0:09:53.840 --> 0:09:57.640 a scarcity of capital relative to the to the caliber 0:09:57.640 --> 0:10:01.320 of ideas that were out there, and so I'd say 0:10:01.440 --> 0:10:05.319 deal sourcing is much easier in a sense because there's 0:10:05.400 --> 0:10:09.520 less money chasing a huge number of good ideas, and 0:10:09.800 --> 0:10:12.360 those ideas, by and large do come out of our 0:10:12.520 --> 0:10:17.040 university and research infrastructure here in America. The same is 0:10:17.080 --> 0:10:20.560 also true in other parts of the world in Europe, China, India, 0:10:20.600 --> 0:10:25.960 and so forth. But it's really the translation of those 0:10:26.080 --> 0:10:32.280 academic concepts into products that could make money that is 0:10:32.320 --> 0:10:34.840 the challenge. That's the so called valley of death that 0:10:34.880 --> 0:10:37.880 people sometimes talk about in our industry. There are just 0:10:37.920 --> 0:10:41.120 an immense number of cool ideas. If you go into 0:10:41.400 --> 0:10:45.200 any university in our country, but such a small number 0:10:45.240 --> 0:10:47.920 of them is ever going to cross that chasm. And 0:10:47.960 --> 0:10:51.000 part of that is that the expertise and the personnel 0:10:51.080 --> 0:10:55.080 required to do that translational work is not the same 0:10:55.120 --> 0:10:58.320 expertise that is required to do the inventing in the 0:10:58.320 --> 0:11:03.920 first place. So that is really what the large pharmaceutical 0:11:03.960 --> 0:11:08.480 companies have a specialized expertise and they train people in 0:11:08.600 --> 0:11:11.240 this translational work. How do you go from early science 0:11:11.600 --> 0:11:14.400 to real products? 0:11:29.160 --> 0:11:32.480 When I go to a typical venture capitalist website or 0:11:32.520 --> 0:11:34.800 I see their Twitter bio or something like that, it'll 0:11:34.840 --> 0:11:37.920 say like, we bet great founders, and I'm like, thanks, 0:11:37.920 --> 0:11:41.120 that's very helpful because that distinguishes you from the venture 0:11:41.160 --> 0:11:43.920 capitalists who back crappy founders. So I'm glad I'm gonna 0:11:43.920 --> 0:11:47.199 invest with you. And said, what's the biotech equivalent? What's 0:11:47.240 --> 0:11:51.120 the cliche in your industry that every VC says that 0:11:51.240 --> 0:11:54.319 ostensibly distinguishes them from all the others. 0:11:55.640 --> 0:11:57.800 Well, I'm not sure what the vcs say. I mean, 0:11:57.880 --> 0:12:00.640 they are kind of commoditized in the sense that most 0:12:00.679 --> 0:12:04.400 of the firms look pretty similar. They employ thirty PhDs 0:12:04.480 --> 0:12:07.880 and physicians, and the value of those people is that 0:12:07.920 --> 0:12:10.760 they can make sense of the information that you have 0:12:10.840 --> 0:12:16.000 to process to invest intelligently in this space. In terms 0:12:16.000 --> 0:12:19.800 of what distinguishes the founders that they like to look at, 0:12:20.360 --> 0:12:23.040 I'd say again, it's kind of the inverse of what 0:12:23.120 --> 0:12:26.679 you find in tech. There's a real premium on quote 0:12:26.720 --> 0:12:30.640 gray hair in the biotech industry because the only way 0:12:30.679 --> 0:12:33.120 to learn this stuff is to do it over and 0:12:33.200 --> 0:12:35.560 over again and to have had a lot of failures. 0:12:36.120 --> 0:12:40.400 And if you think about a software company, the tropes 0:12:40.480 --> 0:12:45.240 you are familiar with are you know, fail fast, pivot right. 0:12:45.320 --> 0:12:48.000 You know, like, you launch something, it doesn't work, you 0:12:48.040 --> 0:12:50.560 tweak the product design, you go into a different market. 0:12:50.880 --> 0:12:54.080 You can adapt very readily to the market. In biotech, 0:12:54.120 --> 0:12:57.280 if you choose to embark upon a clinical program, you're 0:12:57.320 --> 0:13:01.440 in for thirty or forty million bucks, an easy door 0:13:01.480 --> 0:13:04.800 to walk back out of. And so there's a real 0:13:04.840 --> 0:13:08.320 premium on people with experience who have done it multiple times. 0:13:08.920 --> 0:13:11.679 That is a little bit at odds in recent years 0:13:12.240 --> 0:13:16.199 with a movement that people have I think awkwardly dubbed 0:13:16.240 --> 0:13:20.840 tech bio instead of biotech. And really these are Silicon 0:13:20.960 --> 0:13:26.000 Valley tech investors, not totally unlike myself, who have gotten 0:13:26.000 --> 0:13:30.000 into biotech, and they think that what's about to change 0:13:30.600 --> 0:13:32.160 is it's going to go the way of the tech 0:13:32.200 --> 0:13:35.080 industry and the next big companies are going to be 0:13:35.120 --> 0:13:38.160 started by really clever twenty one year olds coming out 0:13:38.200 --> 0:13:42.640 of Stanford and that hypothesis people have been testing now 0:13:42.679 --> 0:13:44.839 for a few years. I'd say it's a little too 0:13:44.880 --> 0:13:49.559 early to issue a verdict, but that's never really been 0:13:49.600 --> 0:13:50.080 our theory. 0:13:50.400 --> 0:13:54.440 Is that hypothesis just predicated on AI coming in and 0:13:54.520 --> 0:13:56.480 making you know, drug development easier. 0:13:56.559 --> 0:13:57.319 Is that all it is. 0:13:58.320 --> 0:14:01.199 There's a lot of that. I'd say there are two 0:14:01.200 --> 0:14:04.280 parts of it. One of it is maybe more substantive 0:14:04.320 --> 0:14:07.000 than that, and this is a little nuanced, but I 0:14:07.000 --> 0:14:10.280 know odd lots of people like Nuance one of the 0:14:10.280 --> 0:14:14.679 big transformations that really gave rise to the biotech industry. 0:14:15.040 --> 0:14:17.400 And when I use that term biotech, I'm distinguishing it 0:14:17.440 --> 0:14:21.600 from big pharma. So biotech really just means small drug companies, 0:14:21.640 --> 0:14:25.440 many of them are public. What really gave rise to 0:14:25.480 --> 0:14:28.000 that industry was the big pharmas at the behest of 0:14:28.080 --> 0:14:35.520 Wall Street deprioritized early stage research because Wall Street said, 0:14:35.560 --> 0:14:38.680 you're wasting a lot of money on this really risky 0:14:38.840 --> 0:14:42.560 early stage discovery work. What we would rather you did 0:14:43.360 --> 0:14:47.400 was just let all these crazy guys like da finance 0:14:47.480 --> 0:14:51.240 startups and once they work, just buy them. You know, 0:14:51.280 --> 0:14:53.120 you're going to pay a higher price, but you won't 0:14:53.120 --> 0:14:56.200 be burning all this money on early stuff. What that 0:14:56.280 --> 0:15:01.160 led to was an exodus a very specialized technical experts 0:15:01.200 --> 0:15:04.360 from the pharma companies, and it created the so called 0:15:04.440 --> 0:15:09.760 cro or contract research organization ecosystem. So you now, as 0:15:09.800 --> 0:15:12.760 a consequence of that, for the past twenty years, have 0:15:12.960 --> 0:15:20.240 had a very proficient environment full of contract organizations that 0:15:20.280 --> 0:15:23.200 you can hire as a little company to outsource a 0:15:23.200 --> 0:15:25.400 lot of work that you couldn't in the past. So 0:15:25.800 --> 0:15:28.400 the best analogy to to tech would be sort of 0:15:28.400 --> 0:15:33.000 like virtual servers or cloud infrastructure, Like you know, to 0:15:33.080 --> 0:15:35.359 have a startup, you used to have all these servers 0:15:35.600 --> 0:15:38.000 in your office, and then at some point you didn't 0:15:38.000 --> 0:15:41.000 need that, so the cost of new company formation went 0:15:41.080 --> 0:15:45.280 way down. So part of the argument for younger, more 0:15:45.320 --> 0:15:48.520 agile founders has been, look, we got this whole new 0:15:48.960 --> 0:15:52.440 kind of infrastructure through which they can build companies in 0:15:52.480 --> 0:15:56.960 a really agile way. The other argument, you know exactly 0:15:57.040 --> 0:16:01.600 your question, is around AI, and that is basically, look 0:16:01.640 --> 0:16:05.240 these old people don't understand AI. Let's get some young 0:16:05.800 --> 0:16:08.520 Silicon Valley computer science he types to do this, and 0:16:08.560 --> 0:16:10.000 they're gonna show them how it's done. 0:16:10.880 --> 0:16:15.720 I feel like that's probably a phenomenon that goes beyond biotech, 0:16:16.040 --> 0:16:19.320 where there's this fantasy, and maybe in some cases it's 0:16:19.360 --> 0:16:23.320 even correct, but there is this fantasy that every industry 0:16:23.400 --> 0:16:26.760 out there must be dominated by old dinosaurs who don't 0:16:26.800 --> 0:16:29.320 know how to use tech and who have been doing 0:16:29.360 --> 0:16:32.600 something the same way forever. And so you're twenty twenty five, 0:16:32.640 --> 0:16:34.600 it must be out of date by now and they 0:16:34.600 --> 0:16:35.600 haven't figured this out. 0:16:35.720 --> 0:16:39.320 And if we could just cough cough, journalism. 0:16:38.960 --> 0:16:42.120 Yeah right, if we could just hire wiz kids, then 0:16:42.200 --> 0:16:45.520 we could reinvent the industry from first principles and just 0:16:45.560 --> 0:16:47.920 do a much better job than the legacy of things. 0:16:48.080 --> 0:16:51.720 And I think, whether it's healthcare or whether it's industrial 0:16:51.800 --> 0:16:55.080 stuff that we see Silicon Valley getting excited about right now, 0:16:55.200 --> 0:16:58.360 it just feels like the default assumption must be that 0:16:58.400 --> 0:17:01.680 the veterans are doing something wrong, and with pure brain power, 0:17:01.720 --> 0:17:03.120 we could figure out what that thing is. 0:17:04.040 --> 0:17:07.439 I think that is a reasonable characterization or what people 0:17:07.520 --> 0:17:10.879 say today in a lot of different places, and I 0:17:10.960 --> 0:17:16.320 don't think it's true in my sector. But as with 0:17:16.720 --> 0:17:23.240 every conversation about AI, the challenge is balancing two ideas 0:17:23.280 --> 0:17:25.640 that can be true at the same time but seem contradictory. 0:17:25.640 --> 0:17:28.960 And one is that this stuff is amazing, and it is, 0:17:29.320 --> 0:17:34.520 particularly in life sciences, responsible for some true breakthroughs, like 0:17:34.600 --> 0:17:37.960 the breakthrough that won Demisisabus at deep Mind the Nobel 0:17:38.000 --> 0:17:43.200 Prize last year with alpha fold, which was this amazing 0:17:43.240 --> 0:17:47.040 discovery they made that using machine learning models you could 0:17:47.119 --> 0:17:50.520 solve a problem that had gone unsolved for decades, which 0:17:50.600 --> 0:17:53.840 was can you predict from the sequence of a protein's 0:17:53.880 --> 0:17:58.960 amino acids what three dimensional shape a protein is going 0:17:59.000 --> 0:18:04.520 to take in a physical environment. And I just threw 0:18:04.560 --> 0:18:07.919 around a bunch of terms of art. But this is 0:18:08.119 --> 0:18:11.879 fundamental to drug development and drug discovery. So it's like, 0:18:11.920 --> 0:18:15.560 on the one hand, you can't deny these breakthroughs that 0:18:15.560 --> 0:18:19.359 we're experiencing. You can't deny that when you talk to Gemini, 0:18:19.960 --> 0:18:22.800 it's staggering what this thing can do. I mean, I'm 0:18:22.840 --> 0:18:25.560 sitting there all day having it teach me about asset 0:18:25.600 --> 0:18:28.399 pricing models or whatever else I'm interested in. But at 0:18:28.440 --> 0:18:32.840 the same time, the religious movement that is powering all 0:18:32.880 --> 0:18:35.440 of the investment and a lot of the entrepreneurship here 0:18:35.840 --> 0:18:39.520 across industries is full of hot air and is making 0:18:40.040 --> 0:18:43.360 claims that are preposterous unless you are a zealot. 0:18:44.240 --> 0:18:47.000 Just real quickly, if we had been having this conversation 0:18:47.359 --> 0:18:49.600 in a month ago, would you have said Gemini or 0:18:49.600 --> 0:18:52.119 would you have said CHADJPT Because I switched from chad 0:18:52.200 --> 0:18:54.439 JPT to Gemini in the last month, and I'm just 0:18:54.480 --> 0:18:57.800 curious whether you're what you would have said a month ago. 0:18:57.920 --> 0:18:59.960 A month ago, I was using all of them. Now 0:19:00.040 --> 0:19:01.119 I'm only using Gemen. 0:19:01.160 --> 0:19:03.080 It's interesting, all right, good data plant. 0:19:03.119 --> 0:19:06.239 Okay, talk to us about the choke points when it 0:19:06.280 --> 0:19:09.760 comes to new drug development, because I imagine, okay, maybe 0:19:09.800 --> 0:19:13.040 AI machine learning can speed up some of the research 0:19:13.160 --> 0:19:16.199 or discovery process, but even after that, you have to 0:19:16.240 --> 0:19:20.000 go through these really long clinical trials that in some 0:19:20.119 --> 0:19:24.719 cases take decades. What are the major I guess, like 0:19:24.880 --> 0:19:27.959 stumbling blocks to getting something to the market. 0:19:29.359 --> 0:19:33.639 Your question held the answer. The process of taking a 0:19:33.680 --> 0:19:36.399 drug from idea to the market. You can think of 0:19:36.480 --> 0:19:41.119 as a funnel. To just use a visual analogy and 0:19:41.200 --> 0:19:43.840 into the top of the funnel, go all the millions 0:19:43.840 --> 0:19:46.359 of ideas that people have, and then as you go 0:19:46.400 --> 0:19:49.400 down the funnel, you are spending progressively more and more 0:19:49.440 --> 0:19:54.040 and more money to prove two things. The first is 0:19:54.080 --> 0:19:57.480 that the drug is safe and won't harm or kill people, 0:19:57.960 --> 0:20:00.960 and the second is that the drug works and actually 0:20:01.040 --> 0:20:05.800 modifies the disease that you're trying to treat. And the 0:20:05.880 --> 0:20:10.080 tragedy of our moment is that the only way to 0:20:10.119 --> 0:20:13.000 figure out if drugs are safe and effective is to 0:20:13.160 --> 0:20:17.440 try them in human beings, living, breathing human beings, and 0:20:17.480 --> 0:20:23.600 that is extraordinarily time consuming and incredibly expensive financially. So 0:20:24.680 --> 0:20:27.959 I wish for the day when AI is able to 0:20:28.280 --> 0:20:33.080 fully simulate an accurate human in the computer and we 0:20:33.280 --> 0:20:36.159 don't need to do clinical trials on real people. But 0:20:36.359 --> 0:20:40.280 until that moment, the vast majority of the cost and 0:20:40.400 --> 0:20:45.720 expense and time that is involved in drug discovery remains 0:20:45.760 --> 0:20:49.800 with us. So most of the AI technologies that people 0:20:49.800 --> 0:20:54.640 are excited about really would have the effect of putting 0:20:54.720 --> 0:20:58.240 more good ideas into the top of the funnel, But 0:20:58.520 --> 0:21:02.080 unfortunately that doesn't solve a problem that we have. We 0:21:02.240 --> 0:21:05.960 already are drowning in good ideas, and the issue is 0:21:06.040 --> 0:21:08.399 exactly the choke point or bottleneck that you're referring to. 0:21:08.480 --> 0:21:11.200 This is really there's actually two questions. First of all, 0:21:12.040 --> 0:21:14.760 is there low hanging fruit from a regulatory side to 0:21:14.840 --> 0:21:17.679 accelerate that process. People like to fathom, oh, the FDA 0:21:17.840 --> 0:21:20.000 must be super There's another area people will say, oh, 0:21:20.000 --> 0:21:22.439 the FDA must be super slow and do things one 0:21:22.480 --> 0:21:24.679 way we could expede this up. I don't know. Is 0:21:24.720 --> 0:21:27.760 there somewhere along the process where like from a regulatory 0:21:27.760 --> 0:21:31.080 standpoint or some other thing, that the either the cost 0:21:31.119 --> 0:21:33.840 of the timelines could shrink or is it mostly still 0:21:33.880 --> 0:21:36.800 just the reality of we have to test these things 0:21:36.800 --> 0:21:38.680 on humans and that's costly going, it takes time. 0:21:39.000 --> 0:21:40.800 Well, we don't need to do anything. We could have 0:21:40.880 --> 0:21:43.760 no FDA and anyone who has a good drug idea 0:21:43.920 --> 0:21:46.760 just launches it commercially and if some people die from 0:21:46.840 --> 0:21:49.080 that and it doesn't do anything, that's fine. By the way. 0:21:49.160 --> 0:21:52.840 That's kind of like the supplement yeah, time and the 0:21:52.880 --> 0:21:56.560 way we deal with it. Milton Friedman famously thought that 0:21:56.600 --> 0:22:00.159 the FDA should only assess the safety of drugs, and 0:22:00.200 --> 0:22:02.199 if a drug was proven safe, put it on the 0:22:02.240 --> 0:22:04.959 market and let the market dictate whether people determine they 0:22:04.960 --> 0:22:07.520 should pay for it based on their lived experience with 0:22:07.600 --> 0:22:12.199 whether it works or not. Now, I just personally prefer 0:22:12.280 --> 0:22:15.600 to live in a world where if I've got something 0:22:15.800 --> 0:22:19.760 that's going wrong, I can more or less trust that 0:22:19.880 --> 0:22:22.680 the product my doctor gives me has been proven safe 0:22:22.720 --> 0:22:27.280 and effective. And that reflects that we have today a 0:22:27.320 --> 0:22:30.960 pretty high bar for approving drugs. But we could certainly 0:22:31.040 --> 0:22:33.120 lower that bar. We could change the type of data 0:22:33.160 --> 0:22:36.160 that the FDA requires, And that's what's happening in China. 0:22:36.200 --> 0:22:38.440 By the way, I know you mentioned this other episode 0:22:38.480 --> 0:22:41.679 you did with my friend Tim. In China, the regulatory 0:22:41.760 --> 0:22:45.080 environment has been moving pretty rapidly, and they've done that 0:22:45.119 --> 0:22:48.080