$100 Trillion Future of Biotech: AI, China & Next Industrial Revolution | CEO Ginkgo Bioworks

Engineers coming into biology. That was like the origin of symbio.

I think a lot of the ideas we had back then were kind of stupid. Anything in biotech that was not a drug, like a therapeutic, has really gotten wiped in the last few years.

So the root problem is it's alien technology, like literally four billion years advanced over ours. We can write code for it, but we have like no design specs, no nothing. Everything is reverse engineering. GLP-1s, they're a trillion-dollar product. I mean, they're crazy. It's like the biggest consumer product since the iPhone.

You mentioned China. We're in danger of losing even therapeutic R&D drug discovery to China right now.

We're... we got to just start. And the first step is kill the bench.

Thanks, man. Good to be on.

Doesn't seem like it. Yeah.

Sounds great.

Yeah, I mean, so the origins of it was back at MIT, like in the early, kind of late 90s, early 2000s. So one of my co-founders is this fellow Tom Knight, who was a professor in electrical engineering and computer science at MIT. And since like, I think Tom started on the faculty in like 1972. So like, I have this great black and white photo of Tom with his master's thesis, which is this like refrigerator-sized mini computer, right? So Tom was that generation that basically went from mainframes to mini computers, and then he was a computer architect. He architected a famous mini computer called the LISP machine, and then when it went to chips, Tom taught the semiconductor design course for years at MIT. And so that was his background till like the mid-90s, when Tom is like, oh, you know, forget computers, the interesting thing to program in the future is going to be cells. And the core idea is, hey, DNA is digital code. You can read it with DNA sequencing, write it with DNA synthesis, like DNA printing. And if you can read and write code, could you bring in like a lot of the learnings from computer science and electrical engineering? A different type of code, you know, right? Sure, like digital bits, but could some of the analogies, some of the learnings hold over as we moved into biology? And so, you know, it was really like engineers kind of coming into biology that was like the origin of symbio, and in particular like computer science and electrical engineers. All right. There had been like chemical engineers that sort of came in even around the early days, like in the Genentech era, that were like doing fermentation and all that sort of stuff. But what was really unique about synthetic biology was like the electrical engineers and computer scientists. And it was not like the easiest interaction between that group and the biologists out of the gate at MIT. Like I remember I would go to like the biology retreat because I worked for this guy Drew Endy and he was in the biology department, and I would show my poster, you know, highlighting my grand acts of genius of like these genetic logic circuits and all this stuff, and the biologists were like, this is garbage, you know. And so there was a real tension there. And so anyway, I can talk a little bit more about like where that went and what I think a lot of the ideas we had back then were kind of stupid. But some of the ideas were good, and one of those ideas in particular became the basis for Ginkgo. And so I can talk about that too. But that was kind of how it got started. It was a little bit like computer scientists and electrical engineers recognizing DNA as a coded language and saying like, well, what could we bring over from our world?

Yeah. So let me tell you, like, okay, it's now 20 years later, like what the hell has worked and what has not, and what do I think like would be cool for like a new generation of people to work on, right? So, um, all right. So if I rewind the clock back, you know, like '03, um, we had a bunch of ideas around, uh, basically trying to like almost like standardize, um, components within biology in order to make them, um, uh, more reusable, right? Um, and then we even had ideas around like logic circuits and things like that. Like if I use a repressor to turn this gene off, that's a little bit like a, you know, like an off switch or something, or an OR gate or something. And so we would have these elaborate things, and you know, a little bit of that is a good idea. Like, um, uh, you know, like antibody-drug conjugate in pharmaceuticals, where an antibody has a toxin stuck on it and the antibody delivers it to the tumor, so the toxin goes only to where you want it to go. That's a little bit of logic, right? Like, am I in the right place? If yes, deliver drug, right? If not, don't. So that level of logic is valuable. Uh, we were trying to build like 8-bit adders out of 17 repressors. It was like trying to make biology act like a computer when biology is fundamentally not a computer. So those ideas were stupid, uh, and I think like not worth it. Um, what I think ultimately was a very good idea, um, that Tom taught me, uh, was this concept of abstraction. Uh, and so in Tom's era, if you wanted to be a computer scientist, you also had to be an electrical engineer. Okay? Because like, how in the world could you program like this elaborate, you know, giant thing with all the wires and everything else if you didn't know like how it worked? Right? Now, of course, like today, you know, my 10-year-old can like draw boxes on an iPad or something and program a computer. So, so like what happened in the intervening, you know, 50 years? And what happened was abstraction, right? We went from working directly at the hardware level. Uh, we added things like assembly code, operating systems like Windows, programming languages, Java, graphical programming languages. Now, ChatGPT, you can talk to it and it can write code, right? So we basically separated computer science and electrical engineering into two different disciplines. And the unappreciated, like, powerful thing about that is then they each got to develop their own cultures. Okay. So computer science went in the direction of like math, basically, uh, and electrical engineering went and ran out like physics, right? Like lithography, TSMC, like automation, right? Like this was the atoms one and this was like the information one. Okay. And then, but the whole time they kept talking to each other. So every time you had more powerful chips, the software people could just invent more powerful software and it would magically work. Okay? They didn't have to have like a big powwow about how to make it all work together. All right? Because they had clean abstraction. All right? So Tom Knight, my co-founder, he gets into biology. He's like, okay, I want to program DNA. It's like programming computer code, and the first thing I got to learn how to do is pick up a pipette. Because how in the world can you program biology if you don't know how to do a polymerase chain reaction, a PCR reaction? If you don't know how to culture cells or transform a cell with new DNA, right? If you don't know what a restriction enzyme is and how to do that at the lab edge, how in the world could you have the right to write DNA code?

Yeah. Tom's like, what the hell is this? Like, he's like, it's electrical engineering, computer science all over again. Like, you shouldn't have to know how to do tedious lab work to get to program DNA. Like, those are two different things. And like this one, the lab one, is much closer to electrical engineering, right? Like, it's in the world of atoms. It needs automation. It needs standardization. It needs scalability. You want to push the limits of the physics on what you could do. And then designing the DNA, well, that's a lot more like computer science, right? Like, that's a theory of the mind. Like, do we understand how biology works? Like, what is our versions of, you know, object-oriented programming in DNA or whatever the hell else? Like, they are just very different disciplines, but we're still in the era where it's the EECS department, electrical engineering and computer science all together, right? And we need to split them, right? We need to split like laboratory biotechnology from like genomic design. And that Tom was just deeply right about. And we are still now, it's 20 years later, and I swear to God, if you go to get a PhD at MIT or you want to be a bioengineering undergrad degree... It's like we have not, we've not done it. Like, we have not actually split the two. And at Ginkgo, we're trying to, like, we have been banging on this problem with funding for now for the last 10 years to try to rip the two apart. But like that to me is the core issue of biotechnology, is until we get those two apart, we're kind of in like the 50s of computers, and it's brutal. You know, like only the high priests get to program because they're the only ones who knew how to put all the electronics together and knew how to program. Like, it was just too big of a lift. And that's what biotech's like today. There's tons of young people, cool ideas, things they'd like to do with biotechnology, but they got to go learn to become bench molecular biology lab experts, and they're like, what the hell is that? I don't want to learn that. You know, and so anyway, that's the... Does that make sense? That idea of like abstraction. Um, because that to me is the lesson of the last 20 years, is we haven't solved that still, and so it's still very... the field repels people because a lot of them want to mainly do the programming, they don't want to do, you know, the electrical engineering equivalent, if that makes sense.

So it's not that it doesn't work, it just works like only so well, basically, right? Um, and the short form of it is, um, I think, and by the way, we were like, we computer scientists and engineers, I was a chemical engineer, like co-founders, computer scientists and so on, like engineers coming into biology, I think intrinsically don't, like, they don't have like an intrinsic respect for like the complexity of cells. And like the best way I can explain it to you, like so you get a sense of like how crazy the complexity of cells is, is like how did we get them? Like how did we get to modern biology? Like you look out the window, you see a tree or something, right? Like how did it get there? And the way it got there was a genetic algorithm has been running on planet Earth starting four billion years ago when we first had like the first replicating organisms. And the way the algorithm works is you get mutations, changes in the code, like any, like a genetic algorithm if you ran on a computer. We make a new generation. If it's better, it stays. If it's worse, it dies. And we repeat that. And we use the incident energy of the sun on planet Earth every day times four billion years is what's powering that genetic algorithm.

It's pretty amazing. And when you look out the window today, you see the integrated effort of all that energy. Okay? That's why biology is a freak show technology. Okay? Right? Like everything we've invented, we've just burned, you know, fuels over the last 300 years to like make our staggering acts of genius as humans, but biology had four billion years to bake in like just molecular complexity at a level that is bananas. Bananas. Okay. Right. Like it's just nuts. Like it's nuts microscopically. Like when you just drill into it, it's like how in the world is it doing it, you know? Like how's it all hanging together? Like why is it not... why do you have a tree last a hundred years, you know? Right? Like it's repairing itself. It's crazy, you know? Like it's collecting energy. It's producing nanotechnology. It's wild, right? Like...

Yeah. And by the way, like, you know, no offense to Sam and OpenAI, but like still the only remaining, you know, the only form of general intelligence is still biological. So, so like, so what the hell is going on, right? You know, like, and so, but the reason we have the conceit of, hey, we ought to be able to design it and stuff is because I swear to God it runs on code. Like, and that's just wild, right? Like it's wild. Like in the same century that we basically built modern computing, we discovered that information was encoded across generations digitally in biology.

Amazing, right? Yeah. Yeah. It turns out if you want to transmit information, it's translated less lossy in a digital format, right? And so like biology figured that out four billion years ago and like has been piling up learning since then. And it's amazing. And so, so, so like that's the root problem. So the root problem is it's alien technology, wildly, like literally four billion years advanced over ours. Uh, and we can write code for it, but we have like no design specs, no nothing. Everything is reverse engineering. Uh, and so like that, and that's your engineering puzzle, and it looks nothing like any other engineering field, right? Because every other engineering field we took like basic scientific learnings and then ratcheted up complexity, all that's human understandable. In fact, the first field that reminds me at all of biology is AI neural nets, because now you suddenly have these things with a trillion parameters, you're like, oh, I don't know how it works, and I'm like, let me introduce you to systems biology, you know, like we should start studying these things like systems biologists study cells. Like I think Anthropic had a paper out about this, like that's totally the right answer, you know, like these things are like basically little pieces of alien tech and you're going to have to reverse engineer them, like go call some biologists, you know, right? So anyway, but until then, every other engineering field is very much like just layered, layered logic.

Yeah. So, okay, so how do we try to do that at Ginkgo? Right. So what I basically tried to do was to, um, like build out essentially automation, like laboratory automation to do the lab work that was being done traditionally by like someone with my training, like a PhD at the lab bench. Okay. Um, and so we started from the very beginning. We were always building on automation. Um, and there's like different challenges within that that we don't have to get into today. We can, um, but there's like moving liquids, there's moving across multiple different pieces of lab equipment. There's a whole like basket of stuff, um, that you need to address if you're going to like automate things. So that was like one big investment area we made. Um, the other thing we invested in was trying to improve like the design tools, you know, like that the computer science half of the equation, like could we make more standard genetic elements, could we make more standard strains? And I would say the automation side of the house was more successful. And that's if you look, you fast forward 10 years, what I basically did was all these research partnerships, we've done 250 R&D partnerships, Novo Nordisk, Pfizer, Merck, Corteva, uh, industrial biotechnology companies like Zymergen, like, uh, you know, uh, uh, like chemical companies, Givaudan, flavor and fragrance company, all customers of ours where they were basically asking Ginkgo to program an organism for them to meet a spec, and then I would get royalties and milestones on the back end, and that was the business model. And we did like, you know, decent, we built a decent business doing that. It just takes a long time to get to the royalties and milestones. So that like flywheel, like that kind of app store has taken longer than I would like. And then additionally, a lot of the, and this is I think an interesting topic for this podcast and the people watching this, like the non-therapeutic uses of biotech. There was like a wave of them from like, we did YC. We were the first biotech in YC 2014. In the next like five or six years, there was like a ton of like, remember all these animal-free meat companies, you know, like the Impossible Foods and things like that. There was like a whole bunch of those. There was a bunch of like...

Oh yeah. None of this. So this is my point. None of these have done well. Cannabis companies, fragrance, cosmetics, anything in biotech that was not a drug, like a therapeutic, has like really gotten wiped in the last few years, basically like since interest rates went up and like risk came out of venture in a lot of areas. That's now coming back. But like, so that whole crop of what was called symbio on the application layer, that really dried up.

Yeah. I mean, so this again comes back to your, like, what you want to call symbio, right? So like my definition of symbio would be around basically the tooling that you use to design organisms. A lot of people like just think of synthetic biology as everything that's not a drug. Okay. Like that's, they kind of, there was like two sort of like worldviews on this or something. Um, and so of the stuff that's like not a drug, nothing has been like a major blockbuster. No, I would say nothing, nothing anywhere close to what a drug is. There is a couple companies that have done like things in the chemical industry. It's kind of called Genomatica. Their thing has gone to decent scale, but it's not like a crazy commercial success. It's just, it just like Cargill has built a giant facility. There's a million dollar, or not million dollar, billion dollar plant being built in, I don't know, like an Asia, some country in Asia that's also going to do that. And this is to make like basically like petrochemicals with microbes. So it's huge scale, but it wasn't a money maker, right, in the way that a drug is a money maker. And so like, so there's been like things like that. Uh, the animal, I mean, now the animal meat stuff has been like not great. I would say like the best winners are like, I guess Impossible, but like even they have not done well. So like, um, cannabis also, cannabis fell off. And so like there's that. These are people basically trying to make like cannabinoids with fermentation rather than extraction from the plants. Um, materials. There was like a bunch, you know, spider silk leathers. There was a bunch of leathers. Um, that stuff hasn't done well, right? Like so these were all like no one knew if they could be good applications and they just didn't pan. And there's a little bit of a chicken and egg, where is it because we're not good enough at engineering the cells or is it because they're not like ripe markets for disruption, right? You know, like, you know, unclear, right? Um, it's like, why did electric vehicles happen when they did? Was it batteries? Was it Musk could market it to a certain scale to like get over the hump that everyone else had died on before? Could you have had electric vehicles like 20 years earlier, but you needed to market them better? You know, like it's not clear, but like there was a moment where it felt like you could have flipped meat, and like that's a giant ass market, right? And like, and by the way, chickens are like almost impossible. Like they're super good at converting like feed into mass, but cows are not great. So, so, so like just in terms of like dollar for dollar, like you ought to be able to beat cows. Now whether you like steaks, like whatever, like, you know, like, but like just on a pure basis of like, can you do the thermodynamics, like can you make it cheaper, you ought to be able to do beef, and you certainly can do like high-end things like certain fish, chicken stuff. And like, but I think all that's kind of off the table right now. Like, you know, like, you know, there's not enough consumer demand. It's not the vibes right now, you know, right? Like, you know, the vibes, the vibes are no longer Soylent, you know, right? The vibes are meat, man, or whatever, you know, like steak face or whatever this. So, like, you know, I don't know.

I don't know. It's different. So, so like the, you know, like that stuff matters. These are consumer products. So, so like I would say like there was not like a crazy winner out of that, which is a bummer, right? Um, because we could really use in the industry something other... I mean, Ginkgo's business, we thank God, like we're a platform, we were able to move into therapeutics, but like a lot of these other companies were just long a certain application area and it just didn't work, right? Like, and so that was, that was just killer. And so that's kind of the story, my take on the story.

I mean, the use case I like that's coming up now that I'm like the most excited about is basically like kind of the human enhancement or like the consumer biotech or cosmetic biotech or whatever you want to call it, but like non-therapeutic but used with humans. Uh, which surprised me actually. Like I kind of assumed all the non-human stuff would happen first, but then like I just, I think it was like two years ago or whatever. Like, um, so Brian Johnson was like an early investor in Ginkgo. I'm friends with Brian, but like he had gone and like done that stuff in Próspera where he got like the, um, like the gene therapy or whatever. And then I was like looking on Twitter and there was like all these other people also doing it, like random guy, like here's me like also getting my like, you know, my whatever unregulated gene therapy from Minicircle. And like I was like, holy crap, like, you know, like people are like very willing to do this, you know. Um, and so like that was, and then now you see it like exploding with all the peptide stuff. So, so there's sort of an interesting thing. And then, by the way, that's just like the off-market peptide stuff I find amazing that people are doing as much of that as they are. But like the GLP-1s are like, they're a trillion-dollar product. I mean, they're crazy. I think they're like a $10 trillion product. Like it's like the biggest consumer product since the iPhone. Like it's wild. Like they're incredible. So, so I think that's opening the door to a whole new category that weirdly, like the kind of like the other thing about the synthetic biology people, they were also a little bit like on the edge. Like the mainline of biotech has always been therapeutics. So if you weren't doing therapeutics, you were also just like a weirdo. So, so there's like a weirdo community of biotech people and they're like, I think they should go all do the consumer stuff because like I just think that that's going to work, like that's going to be a big market. Uh, you got to watch the regs, you got to be safe, but like it's, uh, I think there's just so much potential there. It's crazy. So, so on the app layer, that's the one I like the best coming up. I can talk about the tech layer, but for the apps, that's the one I think is cool.

Yeah, anything. I mean, muscle mass I think is a big one. Like, yeah, like any... I mean, I think the easy way to think about it is like regardless of how long, like there's going to be some, you know, whatever, like things break down. There's going to be some irreducible complexity. Probably the brain is like completely impossible, you know, to just have last forever or something. So, so like there's going to be some upper limit, but like along the way to that point, like just be like you're 25 years old. Like, you know, I'm 44 now, right? Like, let me tell you, like I feel different than when I was 25. It was awesome when I was 25. So, so like that, like who is not going to want that? Everybody's going to want that, right? Like that is like market size 8 billion, you know, lifetime TAM infinity, you know, right? Like so, and then exceeding that, like sure, I don't know, what are other things you want to see better? You want to think smarter? I don't know, like there might be other edges, but even just that is like superhuman, as like even not even like in 80, like that is like superhuman at 40, right? Like to be able to have those effects. And so like I think that's just a monster market, like it's incredible.