Ep. 53 - Bob Nelsen on AI, China and the IRA

I think today, you have to give a nod to AI and the potential of AI, with lots of caveats. We live in a world that doesn't have a lot of data, has very low success rates, and very low ability to figure out how everything relates to each other in biology. The promise of AI to decode or even link some of the disparate types of data into some kind of approximation of what biology is doing is super interesting.

I think in both medicine and discovery, both of those things happen. In some sense, AI creates a set of tools that make existing ways of doing things more efficient, more effective, more precise. At the same time, the more interesting promise is: can you create a language of biology? Can you link all of these deterministic models of chemistry, human data, and what we propose about mechanism or function or genome together in some rational way to ask harder questions? Like, 'Make me a drug that treats what we call depression today.' Depression is defined on a 50-year-old questionnaire. So is there a way of looking at the data more precisely, whether it's transformer math or some other math that unifies these models? That's the big question.

I think all of the major diseases that are not precisely traceable to a point mutation are going to be approachable by these technologies. We haven't figured out how yet, but whether it's cardiovascular disease, metabolic disease, or anything in the brain, any of the complex multigenic or multifactorial diseases, clearly depression is not what we define by the questionnaire. Your depression and my depression may be fundamentally different diseases. You could say the same about metabolic disease and cardiovascular disease. And that's not even thinking about prevention. We only think of pharmaceuticals as reactive, but the realm of AI could easily take us into more preventive pharmaceuticals or intervening in pre-disease in a fundamentally different way. Our pharmaceutical industry is not built well to answer or ask those questions.

I think it's a start. It's better to have things approved and then proven wrong and not paid for than to have them not approved at all if there's a potential benefit. We need to lean into approving things, but we also need to lean into not paying for things that aren't effective. I actually think the current class of drugs should have been approved, and we should be rigorous about follow-up studies to make sure they're actually doing something. It's version 0.1 in the battle against Alzheimer's. Once we have a better mechanistic handle, it may be that there are completely orthogonal ways of getting ahead of neurodegeneration before it happens. Our payer model doesn't react well to that. We have a model that wastes trillions of dollars chasing disease, usually not curing or preventing it. Diseases like Alzheimer's will eventually be treated well before you have any phenotype. We need to figure out how to align the interests of society so that gets paid for. In the current system, those kinds of drugs would be really hard to develop. But there's promise in decoding the brain with AI and these complex networks that lead to diseases like Alzheimer's. Alzheimer's is probably many diseases, and in certain cases, it can be very precisely targeted. We just don't know how to reverse it or when.

I think it does create some perverse incentives to move away from small molecule drugs to biologics. In general, when you start going after innovation accidentally, whether you intend to or not, it will have unintended consequences. Right now, we're in an innovative environment that's going like this, so some of the impact gets lost in the noise. It's super hard to tease out the absolute effect, other than as an investor, you'd tend to lean more towards biologics over small molecules, all things equal, which probably isn't good for society. They're trying to go after abuse by pharma in the later years, but they're taking a sledgehammer to innovation at the same time. They start with nine years, and you can already see they want six years or four years, and then maybe the same for biologics. It's a bit like income tax: once it's in, it's not going away and will be abused. There's a reasonable chance that if there's a Democratic House, Senate, and president, they will really screw this up by pushing the number of years down, majorly impacting the US's competitive advantage in biotech and negatively impacting patients. I always thought it would be fun to sit down with Senator Sanders and say, 'How are we going to cure Alzheimer's together?' You said you can't even talk to companies. Companies make drugs; governments don't. If you don't talk to the companies and don't want to engage in the conversation, you're just screwing your constituents who are old. Why can't you engage in a fundamental conversation about how to cure Alzheimer's? That would save essentially the equivalent of the defense budget in outlays by 2050. A trillion dollars. An Alzheimer's cure would be worth a lot of money. Somebody could get paid a lot if they saved a trillion dollars. Senator Sanders, isn't that an okay thing? If you save a trillion, do you get a hundred billion? That's net positive 900 billion to the treasury. Maybe we can have a meeting of the minds, but he doesn't want to engage in that conversation at all.

Right now, we're playing defense. The only way to get people to play offense a little bit is by pointing out that the US is the leader and won't always be the leader if we decimate our innovative ecosystem. Other places will threaten our competitive advantage. There are some Democrats in Congress who really understand that. I don't think I'm going to change Senator Sanders' mind, but it would be interesting to sit down in a closed session and say, 'How could we cure Alzheimer's in the national interest?' Senator Sanders, can you take it upon yourself to spend a weekend off the record trying to figure out how not to bake in the current shitty system that wastes $3.5 trillion a year on reactive healthcare and incentivize a more proactive system? A pill that cures Alzheimer's is worth a trillion dollars. I don't know if that would go anywhere, but you have to try those things. You may be able to convince people on the margin that we're in the most innovative time in history, it's accelerating, and we can move to a system that is more preventive and curative, which is much cheaper than big hospitals. There are some elements of common ground between the innovative industries and the idea that the system is broken. Baking in the existing system is what I worry about.

We're still investing cautiously. Like everybody else, we look at the policy changes and it gives us pause in some areas. But clearly, it's a huge market with a lot of unmet healthcare needs. It's also a very important place when it comes to emerging diseases, as we've already seen. You don't need a lesson in the fact that you can't ignore it. Putting your head in the sand because there's strategic tension is probably a bad idea. I've always thought engagement is better than non-engagement. Even Armand Hammer was doing deals with Russia at the height of the Cold War. We do have common interests. At the same time, there's a lot of reason for concern, so we're watching and thinking more than investing and acting. But engagement is better than non-engagement on average. I also think the reason I helped create Resilience was to create our own national manufacturing infrastructure. If there are threats or competition, the response isn't always to play defense; sometimes it's to play really good offense. Frankly, you almost have to do both. I don't think we should be worrying as much about China on the defense; we should lean into playing offense and just winning. Winning means retaining your intellectual property. There's a huge concern about making sure that whether it's China or anywhere else, Europe and other places have been known to be interested in our IP. Defending and maintaining a strong patent system and enforcing the laws is really important. When I hear nationalism, I get nationalistic too and say, 'Yeah, the US should win.' We should win by making sure we have the best science, take advantage of it, invest in it, and don't hurt it. There's much more of a threat from overregulation than from China. China can screw us if we adopt Bernie Sanders' pricing policies and Ron Wyden's taxation policies. China could have written both of them. Frankly, we'll do much more damage to ourselves by bad policy than from being afraid. America's never really been afraid. We tend to play offense and try to win. So we should be cautious, play hard, and win.

I think we need to create our own reimagining of biomanufacturing, not just onshoring old technologies, but a reimagining of new technology to make a much more effective system, kind of like the semiconductor industry did. They invested tons of money in manufacturing. You need everybody's help: government's help, leverage, and some strong American companies to rethink what they're doing and use their purchasing power to support US companies building it up. Especially when you think about how good we are at manufacturing within some pharma companies, but that's not necessarily available to a bunch of biotech companies. It's not so obvious as just banning Chinese companies from the market. If you think about somebody like Wuxi, it's a different case than BGI. Wuxi is incredibly important to little biotech companies in the US and some pharmas in terms of getting stuff done. I'd like all of that to be done domestically, but we have to be really careful as we legislate not to have unintended effects on American competitiveness. That's going to be a really interesting conversation in the next few months.

Yeah, the law of unintended consequences could be extreme if it was somebody like Wuxi. BGI probably not, but they're very different companies. You have to be thoughtful about your approach. We should be encouraging US manufacturing and the reimagining of manufacturing in biology to make it more efficient and effective. There's an opportunity for the US to win in that; we're not always ahead already. Imagine if the costs of manufacturing cell therapies and gene therapies went down exponentially like they did for monoclonal antibodies a generation ago. That would have an enormous difference on the accessibility of those therapies to patients and the world. I think that will happen, but it may not be as linear as you'd like, and there are all kinds of policy impacts that could affect that. But you can see the trend, and those things are getting faster. Cell therapy is a great example. When we started Juno, we postulated that maybe someday you could get these things down to a day or two, and people thought we were crazy. That's already happening. I remember people thinking that was absolutely ridiculous, even the scientists. Sometimes it's good not to listen to scientists because they won't set the goals as high as they need to be to fundamentally change something.

We do both. It could be like a Prime or a Beam, or a company we're incubating now in RNA, where it's just an invention, a novel way of doing something better, or even a novel way of doing something nobody could think of doing. That's one way, more linear, even though they're still platforms. They might have an A, B, C, and an IPO. There are ways to make that go faster. Then there are things like Alkahest or Denali, or other things we're working on that go after large clinical problems. Some of the things we've been doing in Pumora may have multiple companies within them; we may pull two or three companies together into a bigger thing under a great management team. Then you have the ability of somebody like Ryan Watts to come into Denali and kill the lead project six months later, and kill the lead product 18 months later, and be just fine because he had a bunch of money. We did a very large Series A, and he could apply a higher bar than most biotechs because if you kill your lead product in most biotechs, you're in trouble. Those aren't really portfolios in the sense of trying to create a portfolio; they're trying to apply rigorous science. If there are multiple ways to do it, you do it multiple ways. Alkahest is very like that, where there's a hypothesis, but it's really to dig into the biology with multiple ideas and pursue all of it at the same time with enough resources. If you need to kill something, you just kill it, and there's no need to hang on to something just because it's your lead asset.

It's something we think about a lot. Christina Burroughs really led that effort within ARCH to be more systematic about finding the best people. We don't actually hire search firms or anything like that to do it within companies; we do, but as a firm, we've found the self-assembly model is a good way. We're in a place right now where people come to us when they have big crazy ideas or are ready to do their last company. The human capital aspect of what we do and getting better and better CEOs is probably the biggest thing that's changed over the last decade. As we've become more successful, we've been able to skip the two or three years of wasting money with a suboptimal CEO and go straight to great people. What we've also found is that people think we try to do big deals, but our biggest deals are usually created by cutting back a lot of people, leaving a lot of money on the table even though they're very large. It's more about getting the best human capital as a CEO or founding scientist together in one place with shareholders who are aligned with a long-term vision. Those shareholders happen to have a lot of money, so their minimum checks are pretty big. Turns out, CEOs who have just sold their company for $7 or $11 billion don't really want to hang out with venture capitalists pitching a $20 million Series A deal. They'd rather get $250 million to do what they want because they know what they want to do. Some of these deals are better when capitalized with $250 million in an A round or even more in a seed round, and some are better with $3 million or $5 million or $25 million in the conventional model to reduce risk. We have both flavors. The human capital piece is where we're getting better as a firm, and we're making better companies by applying better, more experienced, more rigorous people to these really cool pools of technology to figure out how to make drugs out of them without betting it all on one single molecule.

Yeah, and sometimes it's the same people, sometimes it's new people coming to us even if they've been funded by other venture funds because they want a lead or to create a company. We're creating most of our companies from scratch with others, thinking five or ten years down the road. The hardest thing for this industry to understand is that sometimes thinking long-term and ignoring the possibility of return, and not modeling stuff, is actually the best way to increase short-term optionality. But you can't fake it; you have to go for it with a lot of money and decide to nucleate a company that's going to change things. You get a great founder, and suddenly your short-term optionality increases because people want to do business with you, do deals, give you money or technology, and you can build the company more rapidly. But you can't start out by saying you want to build the company more rapidly because you'll let short-term goals get in the way of the long-term goal. The big problem with the industry is that especially in the late-stage VC crossover realm, they let the short-term goal get in the way of the long-term goal. You need to ignore short-term variations and always focus on the long-term, but when you get short-term opportunities, you don't ignore them. You can't start by saying you'll just focus on the short term and build a company.

Yeah, thank you. I really enjoyed it. Thanks a lot.