Building Flagship Pioneering With Noubar Afeyan: Paranoid Optimism, “What If?” Questions, and Defying Convention

0:00If you're developing a drug, it's a very, very slow process, largely because we actually don't know much about the human body. Despite the hundreds of billions that I've spent, you're literally kind of putting a molecule in a body that's supposed to get to a particular cell type and find a particular protein and do nothing other than that. and in doing that actually alter the state of that protein enough to affect your whole body sometimes with a disease. The fact that it works is remarkable.

0:40Tomer Cohen:Most companies start with an idea. The most innovative companies start with a question, a what-if question. For example, what if you can inject something in somebody's body and the body make its own medicine? In 2010, that question sounded unreasonable, but it quickly became the exploration that led to Moderna. My guest today is Neubar Afayan, the founder and CEO of Flagship Pioneering and the co-founder of Moderna. At age 13, Neubar fled Lebanon during the civil war and rebuilt his life from scratch in Canada. He later came to describe innovation as some sort of intellectual immigration, leaving the familiar behind but then moving into unknown territory where the map is still incomplete.

1:25Tomer Cohen:That idea sits at the heart of his company. Each year, his teams ask roughly 100 what-if questions and the most promising of those become prototype companies. And then they try to kill them. Not protect them, not justify them, try to kill those ideas. because Nuber believes that breakthrough always comes from going beyond what's already known. And doing that requires what's called paranoid optimism. On one hand, enough optimism to attend something unreasonable, starting with a what-if question, but then have enough paranoia to keep asking whether you're wrong or not. In this conversation, we explore what changes when you build with biology, not software.

2:14Tomer Cohen:How flagship systematically creates companies from scratch. What Moderna really proved about platform thinking in biology. And how can AI fundamentally reshape scientific discovery. Newbar describes innovation as taking the fiction out of science fiction. And this conversation is exactly about how you do that. I'm Tomer Cohen, and this is Building One.

2:43Tomer Cohen:lubar it's a pleasure to have you uh on this podcast thank you so much for joining me pleasure to be here thank you tomer so i want to start on a personal note um i saw you once describing innovation as some kind of um intellectual immigration and you've all you yourself experienced real displacement early in life when you were fleeing the middle east as a refugee at 13 and rebuilding your life from scratch with your family in Canada and then the U.S. And looking back, I was trying to think about that relationship with uncertainty. Was it that instability early on kind of shaped your relationship with risk?

3:20Tomer Cohen:My parents are also refugees. They fleed North Africa to come to Israel. But for them, it was all about stability. They were looking for stability. So yours felt a bit different with your kind of entrepreneurship routes. I was curious how you thought about that. You know, growing up, I didn't really think about these things. So I moved in at 13 years old. I'd not seen snow before. You know, the kind of culture was completely different. We moved to Montreal. It was not kind of a safe route, but I found it that I think I felt more comfortable with my relative competitive advantage because I felt like the people who were locals had no advantage over me because they'd never go there.

4:00And so I kind of felt that if you want to be at the center of the mass that exists, you better have some innate advantages, your last name, how you look, who you know. I had none of those things. And so when I found myself meandering at the edges comfortably, after a while I thought, well, what is causing me to be comfortable there? And I realized it's my lack of confidence in being relevant in the middle of the pack.

4:27Tomer Cohen:What drew you into science and biology early? I was good in school, so I was lucky in that regard. And so initially I thought I'll do electrical engineering. Then I kind of thought, okay, I'll do chemical engineering. I loved it in the sense that it was kind of really challenging and it was very applied. I hated it because I was working on old, old ideas, you know, petrochemicals and kind of things that were, you know, 50, 100-year-old industries. But then I realized there's a cutting edge of that field. And in the early 80s, that was biotechnology. I look back and I say, okay, so why did I go to the extreme edge of what any engineers were doing?

5:05I really do think it was I was running away from where I thought I had no advantage and therefore running to where I was hoping I'd have some advantage of only by being the only one there. When I, you know, have described innovation as just intellectual immigration, you know, a lot of immigrants do extremely well in this country for the same reason.

5:27Tomer Cohen:Is it like immigrants are pushed to the edges so they innovate within the edges because that's kind of where it's like a pool. It sounds like free it was a pool. You felt like, hey, I can shine here. So when I started my first company, I was 24 years old. I was a graduate student at MIT. I was just graduating. And I had, like, this was not what people did in 1987. Inexperienced 24-year-olds were not starting companies. Now, 15 years later with the internet, that began to change. And today, it looks like five-year-olds are starting companies. I kind of thought, okay, so graduating MIT, I was meant to be a professor.

6:00And then the other choice was go to a large company. At the time, big pharmaceutical companies, big tech companies of the day. And I found that kind of going to startup, everybody has a founding story of their life as a founder. And for me, it was a chance encounter with somebody in 1985 who told me how he and a friend of his 30 years earlier had started a company. and so I had no idea who this guy was and he told me his whole story one-on-one in a conference and it turned out to be David Packard. But he literally, a meeting with him caused me to say, you know what, this is what I should do. So that was a pull.

6:35It was a pull in that he described that mere mortals could propose ideas and that the legal system, the financial world allowed for that even back in the 80s, let alone today. As an immigrant, by the way, you kind of feel like you've got a huge disadvantage and that's your advantage.

6:53Tomer Cohen:Now that you have this amazing experience, both with biotech, but also looking at other companies, what do you think outsiders most underestimate or underappreciate about biotech? Let me just say that the very first company I started, we were making enabling technologies by which people could make drugs. So these were things like mass spectrometry technologies, chromatography technologies, things that in the late 80s, early 90s, when my first company was being built, were breakthrough products, but they took a year, year and a half to develop and take to the market. I spent probably 10 years of my life building what became about a 100 million revenue company in the old fashioned way.

7:35That is prototype, test, launch. These were products that ranged from hundreds of dollars per unit to million and a half, which was the most expensive instruments we made. It kind of gave me perspective when I got into the downstream side of this industry, which is drug discovery, drug development, vaccines, diagnostics, which do have this longer and more science-based kind of journey to them and uncertainty. If you're developing a drug, it's a very, very slow process, largely because we actually don't know much about the human body. Despite the hundreds of billions that have been spent at a molecular level, I think people listening who aren't in the biotech sector may be surprised at just how little we understand.

8:20You're literally kind of, you know, putting a molecule in a body that's supposed to get to a particular cell type and find a particular protein, ideally, and do nothing other than that. And in doing that, actually alter the state of that protein enough to affect your whole body, sometimes with a disease. I mean, the fact that it works is remarkable. so you can imagine that there's a huge amount of trial and error, a huge amount of surprise and disappointment because you test these things in laboratories, in cells, in dishes and stuff. Then you test them in animal models that look nothing like a human except we convince ourselves that mice are like humans in some basic ways because we need something to test it in.

9:06Then we convince ourselves based on mouse data that we should go try in chimps and other non-human primates. They look a little more like humans physiologically, but not really. Their immune systems have key differences, et cetera. And then from there, we wait for the regulator. We got to prove to them that this thing's not going to be unsafe and it's not toxic. And then as best we can, then we go into humans for a prolonged period of time. We got to show that it's okay, it's safe, and we're not going to kind of hurt people. And then we start increasing dosage, and then we see how it's doing. Then we start treating patients.

9:39Initially, you often start with volunteers. So I'm describing to you kind of a fairly esoteric process. So you might say, like, again, why do you do this? The answer is, you know, if you can make a vaccine that saves people's lives, or you can make a therapeutic for cancer, or whether you can make something for obesity, not only is this financially rewarding, it is just like super impactful to the world.

10:03Tomer Cohen:I feel like sometimes, at least in kind of the kind of classic technology software, where there's that kind of sense of urgency to just move fast. But what I'm hearing from you is patience and just almost like a different approach to it. Whether it's biotech or, you know, we've also started companies in the energy space, material space, a lot in AI, both early on and more recently. I'm probably most known for, you know, giving talks on what I call paranoid optimism, which is kind of the key balance to me that it takes to dare to do completely unreasonable things and live to tell about them. The counterintuitive thing is that when you want to be more optimistic, you shouldn't be less paranoid.

10:46You should be more paranoid.

10:48Tomer Cohen:I think most people familiar with your name would be with Moderna, at least on the kind of broader, outside of biotech crowd. But some people don't understand. Moderna actually started as an incubation that you started within a much larger company that, honestly, for me, learning about it was extremely inspiring. That's called flagship pioneering. And flagship is a very unique model. You flip the model, you actually generate the hypothesis internally, you build the prototype companies, you create companies from scratch. After your first biotic company, and as you were coming into the space, what did you see missing?

11:27So what happened is I started my My first company was called Perceptive Biosystems, ran it through to about 1997 when it was merged with another, the then, you know, other large company in the space. So we built a bigger one together. And so this notion of serial entrepreneurs was a beginning of a thing already in the early nineties. And I, for whatever reason, maybe it's my engineering background, always thought, well, why would you do something serially when you could do it in parallel? And then people told me, you can't do this in parallel. And I said, why can't you do this in parallel? And they said, well, because investors only want you to do it serially because they want to diversify, which means you can't diversify.

12:08You have to have all your eggs in one basket. And I thought, what's wrong with that picture? And I was brand new in the space and I met a venture capitalist. And there I saw just how interconnected the financial markets were. But importantly, every venture capitalist I met since was really, really, really smart seeming. And I thought, well, how could all these really smart people be doing just this one thing? They get to see hundreds of people with hundreds of ideas, and you can't but become smarter because you just have a massive amount thrown at you, which entrepreneurs don't. So in the mid-90s, I tried on my own to create some new companies with others, even while I was running this first one.

12:48And so I ended up getting fortunate to create four other companies. One was a cancer vaccine company. One was a diagnostics company. One was a drug discovery company. You know, this is where I was a co-founder. I was not investing per se, but I was basically helping create the companies. I care about words. That I wanted the word entrepreneurship to be converted to the word entrepreneuring. So it turns out the act of doing something in the English language ends with the letters I-N-G. So that, so entrepreneuring became kind of for me the right mindset and an institutional entrepreneuring. That's what I got interested in.

13:24And flagship was the result of that. Now, we didn't know what we didn't know, as is usually the case, so we've learned a lot in 26 years. But now I would say that the fundamentals, the processes, the way in which we've learned across literally 127 companies we founded solely and probably thousands that we considered founding and didn't found, that knowledge base accrues to the very next company that we conceive and create. But the thing we've done, which probably is interesting to your audience, is because we work in science, we've also started saying, like, can we really rely on science that walks in the door, not just the entrepreneurs?

14:07Or should we have our own ability to invent and create science that would otherwise not exist, at least not now? And that's what we've also worked on.

14:17Tomer Cohen:How do you establish this kind of firm? Like, is this like you bring, you have like a collection of experts that each have their own field. And then like there is a brainstorming done. We did like an episode with the CEO of Google X recently. And, you know, they bring diverse, but it's a very large organization, Google X. And they bring a very diverse couple of people together. But yours is unique, right? It's like, how big is the company? How do you think about hiring for this? So when we started, it was largely a small group of people I had worked with that came together and said, hey, instead of starting one thing, why don't we start a few?

14:52Instead of capitalizing with each one at a time, going around raising money, let's put a pool of capital together that just supports at least getting them jointly created and advanced. And then we'll see what happens. Innovation happens in a space that I would call an adjacency. So if you draw a circle around what exists and what's known, and then you draw another circle outside of it, that is essentially what's going to exist and what's going to be known over the next months and years, then you could imagine that that second circle perimeter is something that essentially represents at any given time how far before experts tell you I have no idea if it's going to work or I have no idea if there's a need there.

15:34It turns out that if you go outside enough from the current reality, experts are completely counterindicated because their expertise is a function of the here and now and maybe the immediate future. So we operate knowably outside adjacencies in everything we do and have for many, many years. So we have a core team of scientists, engineers, MDs, who for a living, about 100, 150 times a year, essentially launch what we call explorations that are truly just what-if questions. So I'll give you an example. In 2010, one of the 40, 50 things we were doing in the summer was an exploration that said, what if you could inject a molecule in a person and have the body make any protein drug you want, any biotech drug you want?

16:21Well, you could imagine you could stick in a piece of DNA or a virus or a bacterial cell or an mRNA or whatever. And in the case, as you start thinking about how, you start realizing the pluses or minuses of things people have tried, and then you say, okay, well, now I get it. People hadn't tried mRNA in animals because they caused an immune response. Okay, check. And then you keep doing that, serve. But then you go to people and say, hey, if I can do that, what good is that? Well, people say, well, geez, you don't have to build a billion-dollar plant to make every single new drug, which you had to, or wait 10 years to get the process.

16:53So you start iterating on the need and the solution. Neither of them are real in the beginning to try to match with it. That kind of future backwards innovation just isn't practiced anywhere because who's going to fund you?

17:07Tomer Cohen:What does platform mean in biology and why do you care so much about it for flagship pioneering? For us, a platform is a product generation engine. So a platform is a common new understanding, new capability set from which you can derive many, many products. So mRNA is a product generation engine because we were the first to figure out how to design, how to overcome an immune response, how to deliver, how to make it safe in the body. We took our time to come up with how do you do these things so that you don't get one product, but eventually you get 20 products. So by 2020, what people don't realize is that Moderna had a pipeline of 20 distinct vaccines and therapeutics, 10 each, that we were advancing in, some in human trials, some not yet in human trials.

17:57And people said, how could it be that 10 years later they don't have a drug? The answer is an approved drug. The answer is we had 20 just not approved. And we were building a whole new market, a whole new capability. And to do that, you have to diverge and not converge.

18:15Tomer Cohen:Can we take Moderna as an example? So Moderna, the what if was a mRNA? No, the what if was what if you could inject a molecule in a body and make anything you want. The what if was not a solution. When it became mRNA, then it became mRNA in lipid particles, then it became lmRNA and lipid particles that could be used more soon for vaccines and on and on and on. So there's layers of the onion that get added on. But the initial what if actually did not suggest a solution. We started a company several years ago called Profound Therapeutics, where the what if was, what if there's a bunch of proteins in the human body that we are not aware are there?

18:52If you ask that question as a scientist, you'll get laughed out of the room. It's funny how insistent they are that whatever exists, we know. So in the 2020-ish timeframe, whenever it was 2021, we started asking, what if there's proteins we're missing? We don't know. We went, so, okay, you say, all right, well, what good is it? Well, the answer is we've got drugs against the proteins we know, some of them. What if there was a whole bunch more we could find and now we can go after diseases? That's a use case. What ended up happening, this little company discovered thousands and thousands, it turns out, since then validated, since then accepted proteins that nobody knew existed.

19:28And you might say, how could it be? We did the human genome sequence, and we had all these databases, like how could it be just a few years ago? The answer is because it turns out that the way we define a protein, and therefore where we look for it, assumes the central dogma of biology, which is DNA makes mRNA, mRNA makes proteins. So every other way that the cell might have invented to make a protein, which doesn't follow the dogma doesn't exist as far as we're concerned except when you can measure it and prove it and catch it in the act of being made and then now you redraw the maps of of disease biologies all that so it's just one instance of a what if so we do about a hundred of these a year about 20 of them become prototype companies what we call proto companies what do you think is

20:14Tomer Cohen:fundamentally different about this wave in biotech like where do you see the ai making Obviously, AI has been there for years, computational biology. But is it the hypothesis generation? Is it the target discovery? Is it the notion of molecular design, experimentation design? I think the entire act of scientific discovery will be replaced by an AI-centric approach. We can generate hypotheses, scientific hypotheses, because we've trained basic degenerative systems to come up with, knowing what a hypothesis is, coming up with thousands of them in any given area. And that, generating hypotheses, then down-selecting using certain agents to ones that are testable, coming up with the experimental design to falsify them, automatically running the experiments, easier said than done, but within narrow fields you can already with robotics, and then interpreting the data using more machine learning tools in order to alter the hypothesis, that is the wheel of science.

21:20Arguably, for me, the most valuable wheel that humans have invented, even more than the round one. That process, augmented by automation and literally doing things that human minds are frankly quite limited, no matter what IQ people wield and no matter how many of them you put together, is unfathomable. And we're seeing that. This is not a dream. We have had an ongoing project now for three and a half years, number 97, that has now become known as Lila Sciences. Lila Sciences is a 400-person team that has a kind of demonstration of each of the parts of that circle. And in some instances, the whole circle closing.

22:00We've generated massive tokens, numbers of tokens in the trillions, but of scientific data, not of human language, internet-derived data. We're training models that don't necessarily abide by what people thought. Biology is the rules that are being followed. I think the other thing is what we call polyintelligence. The general notion being that we as humans have lived up until probably, well, five, ten years ago in the broad human psyche as the only intelligent life form on Earth. and then we created a machine intelligence, which since the 50s has been around, but really barely intelligence, lots of machine, but barely intelligence.

22:46And now we've kind of made room for the fact that maybe we've got similarly intelligent, whether more or less things than us. So now we've made room. The interesting thing to me is that alongside human intelligence, machine intelligence, we now need to go back and look, and everywhere I look, I find intelligence, that all of nature is just a bunch of forms of intelligence. That's not how we were taught biology. Again, I don't mean by that mouse intelligence. I mean by that a plant cell, an immune cell in your body, a virus. These are all just forms of intelligence. Why do I say that? Well, they have a central code.

23:22They are able to sense and act. They are able to adapt. And they're undergoing truly a form of reinforcement learning. It's called Darwinian evolution. They're every bit as much, in my view, a learning system. So now you start treating nature as a forms of intelligence, and now you get a triangle. That's what I call polyintelligence. Nature's intelligence, human intelligence, and machine intelligence. And for the first time, and I'm very optimistic about this, although I also need to be paranoid, as I said earlier, that machine intelligence might be able to allow us to emulate nature's intelligence forms in ways that our brains, the way we use memory, the way we do compute, do not allow ourselves to.