Her Multi-Billion Dollar Quest to End Disease - EP 25 Priscilla Chan

I am a tennis player, so we have one thing in common at least, which is that we were both the captains of our tennis team.

Oh my God, you were?

Yes. And you were the co-captain?

I think I was the co-captain. Um, the, uh, I played tennis in high school, um, partially because the tennis coach was my science teacher. I saw that, and he was like, you have to do that. I was like, I don't know how. Um, and so he's like, I'll teach you, it's no problem. And, um, I was, uh, my senior year co-captain with someone who's far better than— I would say co-captain denotes no tennis skill on my, on, on my part.

Yeah. Well, I was looking at this because everything I could read, it seemed like you, you started playing your freshman year. It's a really hard sport to just like pick up.

It was really hard. I, um, I, uh, I— it was really hard. I worked hard at it. I was never that good. I will tell you, the lasting legacy of my tennis career is I'm and now a mediocre pickleball player.

Okay, okay. Yes. So the pickleball people are taking over all our courts.

It's true.

This is like the side part of my podcast where I hopelessly fight against this because I'm not gonna win. Um, yeah. So, so you didn't— the tennis, it didn't stick with you?

I still played, but it's, it's the thing. I still do play tennis, but like with tennis you really need someone who's like pretty well matched to you. Whereas pickleball, like, you're— it's like just giant ping— I'm sorry, I don't know if I'm supposed to say that. It's giant ping pong.

Yeah.

And you can play with your friends and it's a great time. And so I do both.

Okay, we're going to talk science now. Okay. Priscilla Chan, most recent title and current title is co-CEO and co-founder of CZI, the Chan Zuckerberg initiative. You're a doctor, pediatrician, and obviously I read an academic all-star throughout, throughout your life. So super impressive. As always, this podcast is made possible by the fine people at E1 Ventures. Thank you, E1 Ventures. So you started CZI about a decade— well, a decade ago.

Yeah, almost exactly.

With huge, like, goals. Yeah. Um, I mean, the, the— I don't know if this has always been the same tagline, but, but supports the science and technology to make it possible to cure, prevent, and manage all disease in the century.

Was it—

was that from— that's us.

That's from day one.

Yeah. So this is a lofty— this is an ambitious, ambitious goal. Although in this, like, AI era, I feel like people say things like this quite seriously. Yeah, yeah, yeah.

Um, it has always been our goal— we want to cure, prevent, or manage all disease. And, um, 10 years ago, you know, this almost exactly from now, uh, 10 years from today, Mark and I were deep in talking to scientists, like, what's possible, what should we be working on, what's feasible. And we're like, well, Well, can we— we have— we wanna set a big audacious goal. Can we just say we're gonna cure all disease? And they were like, well, no. And we were like, well, why not? And they were like, well, okay, but what about all these other things? And like, what is even the pathway to doing that? And we sort of pushed through. And I will tell you, a lot of the scientists that we work with could not keep a straight face. They were like, ugh, I have like, I have academic credibility to uphold.

When you wanted to use This, that, to have that as our tagline.

And they're like, and we're like, I, we had this, we had an announcement, um, a public announcement. And like, literally what some of the coaching was, well, don't smirk when you're saying it.

Yeah.

Um, and then what's totally crazy and unbelievable is exactly what you just said. Today people are like, yeah, and probably not till the— people are saying that it's not even going to take the end of our century. We have given ourselves too much runway. And I think that's how we know that we are in just an incredibly exciting moment in science and in medicine.

Yeah, yeah. And so is this good or bad that now everybody's piggybacking on this? I mean, I think—

It's great.

Yeah.

You think we're going to do this ourselves? Of course not. Yeah. Like, we depend on so many others to be able to do this work. It's fantastic.

Yeah. And I mean, we can— we'll get more into sort of CZI and what it's doing. But I mean, obviously a lot of the motivations for this came from you being a doctor and, and what I— in an interest in science to begin with. Yeah.

So I, um, oh gosh, where do I start? I'll start where I actually, uh, I like to share this actually at our new hire orientation. Um, I, um, am from Quincy, Massachusetts, and I'm the child of immigrants. But I've also like, how do I explain this? I'm 40 years old, I've lived a lot of life, but the most important part of my life was actually before I was born. My grandparents, my two sets of my grandparents were from Saigon. And during the Vietnam War, they needed to flee Saigon and they couldn't get their kids out. And there's 6 on one side, 9 on the other, and they were buddies. And they were like, okay, we got to get our kids out of this country. Um, there's a war. The only way they could get their kids out were these little boats that would float out into the South China Sea. And they, um, but these boats would sink. They were just like crap boats. They would sink. And so they would hear stories of whole families sinking and dying. And what they decided to do was pair up their buddies. Um, uh, like my mom was paired with my aunt and we're like, okay, you guys take care of each other. And I mean, this sounds crazy to say, but if a boat sinks, we would only lose one or two kids.

Yeah.

And what happened, um, is that— and I try to hope that this is like a little spiritual, a little scientific— I hope that's written into my epigenetics, the belief that there's something better out there, and it's audacious, it's big, it's scary, you take a risk and you do it. Like, that's in me because of my grandparents, my parents, and what they chose to do.

Yeah.

So I ended up in Quincy, Massachusetts, and we were refugees, we were immigrants, we didn't know the system. And so many people stepped up to make a difference in my life. And that's when I realized, like, I want to help kids. And I've taught, I went to medical school, and I ended up at UCSF where I trained as a pediatrician. And if you train as a pediatrician at UCSF, it's like a little bit cute and cuddly, but not that cute and cuddly.

I read some crazy stories of things you were doing. Yes.

It's really about taking care of really sick kids with rare diseases. And most of the time you're like, "I don't know what that is." Yeah. Or a parent, like, slips you a PDF and they're like, "This is everything that's known about our disease." and you're like scouring the paper and you're like, like, what do we even know? What can I do to help this child and this family? And like, that is how I learned that basic science is where hope comes from for medicine and for patients. And if we don't invest in basic science, like, there is no sort of hope for better in medicine. Yeah. And sort of that's sort of like The really short version of sort of how I got to really making CZI focused on basic science.

I mean, I read— well, okay, so totally appreciate that sentiment. And then, yeah, I mean, I read— I mean, this is more specific, but a story about two premature children that you got in the parking lot and that, yeah, they weren't breathing and you had to—

What internet are you on? Uh, this is true. Uh, you know, we would train at San Francisco General Hospital and UCSF. And, you know, in San Francisco General Hospital, you would take care of anything sort of— we call it like that comes off the street life. And at UCSF, you did the super high-end sort of like bone marrow transplant, specific genetic diagnoses. And at the General, When you were on call, you would take care of any kid that sort of came off the street. And one of the most memorable cases I ever took care of was a mom who was really sick and pregnant, and we couldn't even get her past the emergency room. They sort of did an emergency C-section right there in the middle of the emergency room, and I had to grab a baby not knowing how old the baby was and revive the baby. Right. Everything, all of my skills had to come into play.

Yeah. Yeah. I'm making you teary at the beginning. I feel bad.

It's okay.

Would you— and I know I'm going to take us forward for a second. There was, there was something just from when you were growing up that I was so curious about. Yeah. I mean, could you paint a picture of— because I read some of these stories that you were talking about, like what your house looked like. You have two sisters?

I have two younger sisters.

Two younger sisters. And so there's the grandparents, the parents in the house. You're— people are speaking Cantonese in the house, and you're this kind of bridge between— yeah, the, the outside world and the house. But yeah, I was trying to picture what it looked like in my head.

Okay, I grew up in a yellow house. Um, it had three bedrooms and one and a half baths. Okay. And it— my grandparents shared a room, my 3 sisters and I shared a room, and my parents shared a room. And we only had one bathroom. We took turns. And I— all of my cousins on my dad's side of the family lived within a mile. And my grandparents have 11 grandchildren. So that little house— and I think I've looked up recently, it's something like 1,800 square feet— would packed with all my aunts and uncles and all my cousins. And so we didn't have much, um, but we, uh, had a big extended family, um, and my parents each worked 2+ jobs my entire childhood.

Yeah. Were you there from grade school through high school?

I was born there, um, and then my— we moved when I was like a sophomore or a junior.

Okay.

But there was something awesome about this house. I don't know how this happened, 'cause it's kind of on a big street, but the bookmobile came every, like, now this is new information, I think. I don't think you found this on the internet.

I never see this.

But the bookmobile came every, like, Tuesday or Wednesday, and that was my access to books and the world. And it literally stopped right in front of my house.

And what were you, I mean, were you fiction? Were you nonfiction? Were you?

I would read everything. But mainly, mainly nonfiction. And then I went to a vo-tech high school, and about half the kids in my high school went to college. But you could also be a plumber, carpenter, an electrician. I'm very handy, I think, because of that. And I took every single science class that was available. And when I was done, I got permission. I bought myself a cancer biology textbook and I taught myself cancer biology.

And were you— so you were always— I saw you did the robotics club. Like, okay, you struck me as like a very high achiever, like with science. I have to assume that was what you were most interested in and called to you, or were you seeing that as, as like this path to success and, and, and, you know, changing your life, I guess.

Uh, it was my passion. Yeah. Um, the cancer biology textbook was because I lost my grandfather to cancer when I was in middle school. Um, and so I just wanted to understand and learn more. But I loved sort of learning about how the world works. And, um, so I had two passions. I love kids. I always love kids because I felt like so many people growing up had given me a chance, and like, my way of paying it forward was going to be to work with kids. Um, and I love science. I was really, really nerdy, and like, I love the logic and the beauty of how biology and science works.

Did you think you'd be a doctor or a scientist? What did you picture in your head?

Well, I actually taught, um, elementary school science before I went to medical school at Harker. At Harker. Yeah.

Um, and like, when you were growing up, what were you picturing you would I just, uh, I thought—

I probably thought I would be a doctor, but it was really most important to me that I work with kids.

Okay, okay. Um, I always wonder about Harker, the, uh— yeah, do you have thoughts on, on Silicon Valley schools?

Oh, you know why I did Harker? Um, I, uh, well, intersecting into the part of my life that more people know about, I've In 2003, 22 years ago, and he moved out to Silicon Valley in 2004, and I was still just finishing my freshman year of college. And we were like, is this— we continued dating long distance. Thank you, JetBlue. And we— and then we're like, okay, are we going to make this work? We're going to try to live in the same place. And what I really wanted to do was Teach for America, and, um, they couldn't guarantee that I could come to the Bay Area, so, uh, I ended up at Harker instead.

Okay, okay. I mean, it's a fantastic school. Well, at least it has a fantastic reputation. Yeah, um, yeah, um, okay, all right. Sorry, I wanted— these are all things I was trying to picture in my head when I was reading, um, some of your background. So that's really helpful. Um, when you When you start something like CZI, you know, we have seen in the past, I mean, going back even to, well, probably even further back, but like Howard Hughes, you know, has, has gave all this money to science and medicine, has had this tremendous legacy. More recently, we've seen some things that came after CZI, like Ark Institutes, where it's a group of wealthy people trying to go after science in their particular way. And I know you consulted with people before you started CZI, but like, how would you define how your model is different to other people trying to approach furthering science?

Yeah, for me, I'll start with me personally. Like, I was going to be doing this work regardless. Um, I, uh, was ready to practice and serve, um, And giving back was not a career shift for me. This was my career. And I think we're at a really cool moment where my interests and Mark's interests and my skills and his skills came to a really interesting intersection. We started from that. We were like, 10 years ago, we want to give back, but like, you know, we have resources. Money is obviously part of this, but what else can we bring to the table?. And for me, it was my experience as a practitioner of medicine from biology, and for Mark was about building technology. And we were like, okay, those are the 3 things that we have. And so it was never just resources, it was about sort of other skills, knowledge that we could bring to the table. And we looked at the problem of basic science, we were like, okay, what can we do? And I think we were the first to really in a serious way bring together biology and technology. And we, we said, okay, there's a few core principles we want to really focus on. One, we want to fund unmet needs. Two, we want to really build a way for people from different disciplines, um, starting, you know, like from our two different backgrounds, bring people from different disciplines together to solve hard problems. And three, we, we want to build technology. Like, there's got to be cool, interesting things to build, um, that can help accelerate science. All of this— when we say we want to cure, prevent, and manage all disease, our strategy is not disease by disease. Our strategy is how do we make all of science faster, accelerate the path to impact so that we actually can get to that goal. Um, okay, so then when we started, we were like, okay, well, we're going to build a biohub. Biohub is super cool, and our first one was in San Francisco, where it was a collaboration between UCSF, UC Berkeley, and Stanford. And we got these universities to partner together, bring their breast and bite— breast best and brightest, um, to work on a problem that's going to take 10 years to solve. We gave them a long runway, which is—

uh, sorry, just the particular problem.

Oh, in San Francisco, the particular problem was really understanding the cell at the single-cell level, both sort of in its innate state and under sort of perturbations, including infectious disease. And so there's the Biohub. And the reason why that collaboration is so cool, like UC Berkeley, fantastic biology, bioengineering department, home to Jennifer Doudna.

Pomona College grad.

Yeah.

There we go.

And then we have UCSF, one of the best medical systems in the world. Like, we were able to bring those groups together. Stanford engineering came together to work on that problem. Super cool. Um, we wanted to fund, um, unmet needs, so we did grant funding. And we funded things— we knew that the academic system is really good at funding the first, getting that first cool paper in Nature, Science, Cell. Like, that works and it's rewarded, but no one was really rewarding finishing complex problems.

I see. Okay.

Um, and so Human Cell Atlas falls into that category. Like, the methodologies were out there, but like no one was excited to fund or incentivize completing, uh, you know, an atlas, which is like millions and millions of cells. And like, how could we support a burgeoning community to complete these atlases so they can be useful artifacts for everyone to use. Um, we funded patients to help make biologists better. Um, and then we built tools. Um, we built a sort of industry-standard, uh, software, uh, engineering team, and we built tools like Cell by Gene, and we supported open-source tools that help scientists do their work faster and better. And then we learned that First of all, the biohubs were awesome. And now we have 3 more.

Chicago, New York.

Chicago, New York, and then imaging, which is sort of not a— it sort of supports the other endeavors. We learned that we could build cool, big datasets. Cell by Gene was both a useful tool used by thousands of scientists on a weekly basis. And it has a dataset that people use to build a lot of these models on top of now. Um, and, uh, we could build useful technology. So we're like feeling like, we're like, okay, there's a little bit of flywheel going here. And then we're like, LLMs come on the scene and we're like, holy cow.

Yeah.

That was the moment where people went from being like, we can't even say your mission with a straight face, to being like, well, probably 20 years, you think? Like, why do we, why, who's, why are we dilly-dallying? And we're like, okay. Okay, now, and then we realized that between our ability to do science in our biohubs, our ability to bring the community together to build assets, we could build datasets, we could build the compute that's necessary. We have one of the largest compute clusters for the life sciences. And we also had the scientists, the patients, the doctors to make sure that what are the models that we're building using large language models to elucidate interesting questions and answers in biology apply and have human impact. And so that's sort of where we are today. And in science, there's all sorts of— the fact that there are— we're not an entirely original idea. We sort of come from inspiration from others and that there's others following us. Like, that's awesome.

Yeah, yeah, yeah. I think— I mean, I really want to create like a science on science war or anything like that. And I think, I think you're probably at least vaguely friends with, with Patrick Collison and all these people, some of the people behind the Ark Institute. I mean, there's some similarities, right? Because they were trying to— they came after you guys, but now they're trying to create this link between the universities and apply AI to— to—

well, you know, Silvana was a BioHub investigator.

Okay. Okay. And you don't feel like you're not upset or anything?

No, no. Well, Silvana was a Biohub investigator, and I think a lot— she's like, she's able to take things that she feels like works well and build upon that. And like, we are friends, actually. Okay. Yeah.

Well, and then I, you know, I talked to the people behind Convergent Research a couple of weeks ago. They sort of have this— they saw these gaps where things that might— research that might not be clear enough to become a startup, it may or may not have a business model around it. And it's also maybe like too expensive for a university to keep funding. I mean, you outlined very clearly your attack. I was just wondering if you also saw these gaps where something like NIH funding gets you like so far, but, but there's always going to be these massive projects that just require more capital than probably makes sense, unless, unless that's just what you want to do.

Well, I think the point at which biology is at, like, there is a lot more sort of open discovery science that needs to happen. Yeah, but biology is actually kind of merging towards where physics is, and you start needing sort of larger scale teams, um, interdisciplinary teams. The equipment is getting very serious. We're not at the point of like LIGO, Uh, but I think we're getting closer and closer to being a team science, shared resources. And so my sense is our group, other groups that you're, you're talking about, are trying different ways to attack that problem. Like, what does biology look like beyond the individual wet lab to solve modern biology questions?

Okay. And you, you I think it's about $7 billion to date that have gone into CZI. And I, I ask this next question with all due respect. I mean, you guys have pledged, I think, 99% of your wealth to this. How do you— you know, sometimes I look at something like the Connectome, which is, is this quest to make a map of the brain on this very detailed level, every neuron, synapse, axon, and It's something I've been fascinated by for many years.

It—

I think there's open, fair debate as to like how useful it may or may not be, which is maybe part of the problem. But, you know, I go to these labs and I'm at Harvard and I'm with this Jeff Lichtman and he's got one brain slice.

You know, I worked in Jeff Lichtman's lab.

Oh my God. He's one of my favorite people.

As an undergrad.

Did you? Okay. So did you work? Was he doing— Yeah, he would have been doing.

I mean, I don't know. I was. 19 years old. My job was to dissect individual fluorescent axons.

Okay.

In mice.

In mice. Okay. Okay. Well, so, you know, he's— he's one of my favorite people. He's got— he's got this brain slicer. I think he has two, maybe. And I'm like, Jeff, what would it take to, like, you know, we've done a worm's brain and we just did a fly, and the quest is to get to a human. And I'm like, what would it take? You know, he's like, well, We'd have to have like 1,000 of these brain slicers and then just a ton of computing. And like, at some point it's a resource problem.

Yeah.

And then I'm like, well, it's the brain. Why don't— there's so many rich people in the world. Like, how do we not want to know? Like, what the— somebody must want to fund this. And so, yeah, I'm just— this is a very long-winded question of like, how do you decide how much you're going to give away when? And some of these things might be like $10 billion projects on their own. Yeah.

You know, we, for us, it's, we started with a big portfolio and I don't have a metric that tells me like, that's a winner, right? But what you do is you look at like, why us? And like resources, money itself is not enough to be like, we are differentiated to do this. We have to look at what else do we bring to the table?, and how does it make the sum of the work better? And so we started really broad, and we've been looking at like, okay, some of the stuff is like, you know, pushing a rock uphill that's like not going well. You've got to be real. And it's good work, and good work that we care about, um, but it's like we are not winning at this. And then some stuff you're like, oh my God, this flywheel— like the Human Cell Atlas, that flywheel just went. And we're—

that's a consortium of people that have been trying to build a— it's like a virtual, like, a complete body at the cellular level, right, of a map. I mean, sort of like the connectome, but yeah, but for all cells.

And yeah, that worked. And so we want— when we see things that work, we should do more of them. Can I take an aside on Human Cell Atlas?

100%.

So I'm going to knock it down to pretty basic for a second. Uh, so we've now understood the human genome for decades now. The human genome, you know, the way to think about it is that it's like— you have kids, um, you're familiar with Legos.

Yes. Yeah, in my foot.

And, uh, yeah, yeah, deeply familiar. All the shapes by touch. Uh, so the human genome basically tells you, like, for you, what are all the possible Lego pieces that are available to create you. And we have been able to map that for a long time, but we have never been able to say, how do these pieces come together to create you? And that is actually really important because when you get sick, it's because some parts of you, some Lego pieces are broken, missing in the wrong places. And what the Human Cell Atlas does, we have trillions of cells in our body, and it turns out each cell takes a— expresses a different subset of the Lego pieces. And that makes a certain pack of, of Legos will create a heart cell. A certain pack will create a kidney cell, a skin cell. And then you can start seeing, oh, this is what makes each one of these cells unique and different. And then the next step will be to look at those spatially. Where are they in the cell? How do they come together? And then you start getting, hopefully, video too in live cells. Then you start understanding sort of what the molecular machinery inside of you and how all of that works, then you can start understanding like, you know, sometimes like what is even that cell? What does it do? And where has it broken down? And how can we design a therapeutic, a drug, a protein to actually change that cell state? So suddenly we went from static to more and more dynamic through the Cell Atlas. Um, and you know, sometimes the question I get about the Human Cell Lesson is just like, is this a stamp collecting activity? Is it worthwhile? Like, what is the impact on this? And I think the thing that unlocks the power of, uh, of these bodies are the large language models, and we can start looking at differences. And turns out all your cells are kind of different. And you want to know what cells are doing related activities. For instance, CD47, very active in a number of cancers. And you can try to drug and target CD47, right? But the problem is, who else expresses CD47 and is necessary? And the cool thing you can— someone figured out by looking at these atlases is that It's true, there are normal cells that need CD47, and there's the hijacked cells, but the normal cells are expressing a variant that looks slightly different. So can you take advantage of that difference and design a cancer drug that is specific with lower side effects than, you know, a molecule that is looking at the CD47 in a more general way? So that's sort of how to think about The Human Cell Atlas.

And I was just looking because isn't— you have a software tool that does exactly that, right? I thought I found the name here somewhere.

It's Cell by Gene.

Cell by Gene. I was looking at— oh, there was a different thing that I was checking last night that I wanted to ask you about. Okay, I'll find it. We'll come back.

Well, are you talking about some of the models?

It was— oh, here it is. Transcriptformer. Is that— is that—

Transcriptformer is great. And, you know, our ultimate goal at CZI is to build a virtual cell. And I'm super passionate about that. We can talk about that in general in a sec. But the thing that Transcriptformer does is that it allows you to look at the transcripts, which are the mRNAs that we're looking at, and compare across species. And look, there's lots of model organisms. You were talking about worms, flies, mice. How many times have we solved diabetes in mice? So many.

Yeah.

They are diabetes-free. But we are not. So what's the gap? And so you can use a tool like Transcriptformer to start mapping model organisms, their cell activity, to understand, like, does that really map to the human? And actually vice versa. So only two species in the world get leprosy. Humans and armadillos.

All right. Yeah. Okay.

Okay. So how can we— what is it about these two that make them in common? And how do you use, say, armadillo as a model organism? So the transcript former allows you to map across species because it's obviously not always the best to study in humans. So how do you map across species and make sure that you're doing science that is meaningful? And applicable to humans later on.

And so it's some sort of tool that is like taking in these databases and then letting you— I mean, it seemed like it was playing off LLMs.

It's based on an LLM architecture. It is a large language model.

And in the sense that it's also letting you ask like natural language queries of this stuff, or do you have to know what you're doing when you—

I think it's— you can use it as a sort of a biologist. Okay. I, you know, Like, I, I would say I'm not prepared to ask it specific questions, but, you know, in the context of biology, you don't need to be a computer scientist to use the tool.

Okay.

It's meant for biologists to be used— to use.

How much, like, on the LLM stuff, because it's— it is so exciting. And, um, I think there was like a thing last week, this CHAI-2 came out, where they were, um, coming up with antibodies on these like one-shot run— I mean, stuff that like seems almost hard to believe. The, you know, I looked at what happened when Greg Brockman had his sabbatical from OpenAI. He went to Ark Institute and he helped fix some of their models, which was awesome. But the one thing that was like triggering in my head is like, you have this massive bidding war for talent among all these companies, as you would know, for, for AI talent. And, and, you know, they're getting gobbled up by, by like all the big foundational model companies. Like, are the science people going to be able to get the quality of AI talent to push all this forward? You know what I mean? Like, we're kind of using all our brainpower on this, this language coding quest over here.

Totally. Well, I, I see what you're saying. So, uh, I look at the market. First of all, it's not easy. I'm not going to pretend like I have the answer.

Yeah.

Um, and there are folks who want to build things like superintelligence, and we're not that we're not even sort of building these large language models from scratch, right? And so we are build— we are using fine-tuning, we're building off of other models, we're building upon those that have done the work ahead of us. So what we do have is mission. So people who want to work on biology, like, there's a much smaller subset of places to go. So if you want to work on biology, like, there are a smaller group that we're really discussing and competing with on talent. And then the other thing is like, there's, you know, compensation, which is obviously important, and we cannot compete with the tech companies on this. Um, but the other thing researchers really care about is access to GPUs. Like, like, you're not going to make the most of someone if you don't actually have the GPUs for them to work from. So, uh, we have that at CCI Our cluster is about 1,000 and we have goals to grow it larger. And we— those are— that's what we have. Like, come work on with us on biology and come work with us because we're going to have the compute to sort of support the research that you want to do.

So we should be forcing Jensen to like keep 50,000 GPUs for science every year or something like that.

Please.

Okay. You know, I noticed the areas that— and this was, this was taken from my, like, rushing— not rushing, but going through your website.

And I don't think you rushed. You know what my tennis team—

at the different spots you're investing in. I mean, I spend so much of my time on, on brain-computer interface stuff, on, um, epigenetic reprogramming. I don't know, these are just two of the things that have really grabbed me lately.

I—

and, and so my worldview is a little skewed just because I've been in those two worlds so much. I didn't see a lot of BCI stuff that you guys were looking at. I mean, it seemed, it seemed a lot around building these foundational models of the body. I think in Chicago the focus is on inflammation, or maybe that's just part of the quest. Yeah. So, yeah, I mean, this was a— it's a long— another long-winded question, but, but it felt to me like you guys were not following the some of these areas that the rest of the world— longevity, all these things that people seem to be chasing. And you were, you were kind of doing your own thing on these very, like, deep foundational technologies that unlock, um, unlock things.

Yeah. So that comes back to the strategy of we're not trying to solve one particular health problem. We are really trying to build tools that makes everyone better at their work. And so if you look at the things we're trying to do, it really is about tool building that's gonna be open to the entire scientific community to build. On the virtual cell model, we really believe that biology is gonna be more creative and faster if you build an in silico cell that biologists can flip the ratio on. Right now it's 90% wet lab, 10% computational, and it's slow and it's hard, and it makes people risk-averse. It's going to take so long to do that. I'm not going to test it unless I'm really convinced. If you can build an in silico model and flip the ratio, they're 90% computational, 10% wet lab.

So you want to make a cell that's— this is a cell model that is so rich.

Yes.

A virtual cell that I can run experiments on it. I can test molecules.

Exactly.

Okay.

I want to be able to de-risk that, do that in silico so that in the wet lab I can just confirm it. That's gonna make science faster. That's gonna make people willing to test riskier ideas. And that is going to be something that accelerates a lot of science. It's gonna change the paradigm, and it's actually gonna very directly impact human health. And then at our BioHubs, we do a lot of work around engineering tools in addition to sort of the large language models and modeling work that we're doing. In Chicago, what Shaina is doing, which is incredible, is that she wants to understand inflammation at the tissue level, at the organ level. And, you know, right now, you tell a doctor, you're like, I don't know, like something my grandma would say, you know, my joints hurt when I eat cucumbers. And you're like, what? And like, you know, I'm a doctor. I'm like, I don't know what to do with that. But, you know, everyone's body works differently, and there's probably truth in a lot of these things that cause inflammatory states, whether it's, you know, arthritis like my grandmother, neurodegeneration, cardiac health, aging. And you want to be able to understand what the inflammatory patterns are. But a lot of these changes are on a timescale that aren't captured by blood tests that you get however frequently you're taking them. You're just not capturing these more fine changes that impact your health and the way you feel. Shayna developed a sensor that can be eventually used very similar to a glucose monitor that folks are using to actually look at these spikes in inflammatory states and to be able to pick up super early signs of autoimmune flare, cancer recurrence, whatever it might be. And to be able to look at it with very fine temporal pattern. When you do this, this happens.

Does— yeah, I mean, I saw photos. It's quite— it looked quite small in the photos.

Teeny tiny.

Does it get placed— does it get placed like an insulin sensor in one spot or you're putting it— you're putting it near where you're—

So right now it's mainly only in animals. But the idea is that you could put it in your skin just like an insulin sensor. But you can also, in animal models, use it, put in the brain and look at inflammatory states in the brain. And the novel part of that technology is that they use sort of a physical vibration to be able to attract, record, and flick off the molecule. So you get that fine pattern. Like when we tested it in mice, we could actually see when you poked, the animal did not like it when you used a needle to inject it. You could see the immune response just from that. So you can start looking at much more clearer patterns and look at immune profiles of different states of health.

Okay.

So that's Chicago. In New York, this is like the fantastic voyage. In New York, what Andrea is trying to do is to be able to engineer immune cells to say, 'Go to Ashlee's pancreas and tell me what's up.' And you can give it instructions, and it can go to your pancreas, observe, and write into its own DNA, like, what it observed. Yes, no, of whatever signal you're looking for. And then it would lyse and release cell-free DNA into your bloodstream, and then I could draw your blood— actually, much like a lot of the cell-free DNA that's used around pregnancy now. And I could read out what that cell saw. And then the next step would be to engineer immune cells to be able to take an action. And so these are all— they're tools, and they're tools that you can imagine using in a lot of different contexts to accelerate science and biology.

And so in each Biohub, just has a slightly different focus. Yeah.

And then they come together. You know, there's sort of like an immune system theme because in San Francisco we do a lot of transcriptomics and imaging and infection, which is another way to perturb the immune system. And they're all generating interesting observations and datasets that feed into our AI modeling work.

Okay. I mean, it seemed overwhelming. You guys must— you must be bombarded with so much information. And then I mean, look, I worked at Bloomberg for a while and I used to sit in the room with some of the people in charge of investing Mike's money from a philanthropic perspective. And I know he was always so concerned with doing it efficiently and actually having the money do good. I mean, it's like hard, you know, it's like everybody's like, oh, I wish I had all the money. I would fund this and this. You know, it's like a logistically hard thing to do. How do you like— what is your normal day like as you're trying to triage both the science and then what you're going to back?

I spend a lot of time with the scientists to understand sort of like what they're excited about, what's feasible, and what is the plan to do there. And the most common question I ask is like, are we being ambitious enough? And where is the risk? How could we make this go faster?

Okay.

And, you know, I, I want to invest in a way that optimizes our chance of doing something truly incredible.

Yeah. So the big stuff.

Yeah, yeah, yeah.

Um, I'm just keeping an eye on time because there were actually a couple, like, things that have just happened. Oh yeah, I want to talk about— yeah, I mean, so I've covered rare disease for a really long time, um, quite often children, and it's just terrible on all fronts. I mean, it's often these genetic diseases that are devastating. Children, you know, might will have very often quite short lifespans. It's a part of medicine where pharmaceutical companies tend not to be incented to pursue something because the market's not big enough. But then when you add up all the people with rare diseases, it's actually quite a large population. They just—

hundreds of millions, right?

But they just all happen to be quite different. And so we talked about Jennifer Doudna before. And what was it? It was like 2 months ago, I think there was this baby. Was it— is it KJ?

Baby KJ.

Baby KJ. So had had one of these rare genetic diseases. We actually created a therapy after KJ was born and to treat it with these liver or kidney—

Liver.

Liver. And so, you know, so this is just a little context of we just had this momentous thing with this baby where regulators and the therapies all came together quickly to help this child. And now it sounds like you guys are going to kind of broaden this out. Yeah.

This is— this is one I'm super excited about, and it brings together all the different threads that we've been talking about. Um, Baby KJ had what's called a urea cycle disorder, which means that his— his liver is unable to process a byproduct of protein, and the byproduct is toxic. And so every time his liver is unable to keep up with the protein that he's ingested, he accumulates toxins. He would have Sorry, he does not, which is the cool part. He would have accumulated toxins that led him to have brain damage, and eventually this accumulates in a child to— can be pretty severe intellectual disability. And from there, when he was diagnosed, he— a team at CHOP, between CHOP, IGI, Jennifer Downer's group at Berkeley, and Danaher, actually had all the parts already ready to go to understand his specific mutation and to use CRISPR to go in and correct that mutation. And he was given— within 7 months of his life, he was given 3 doses of CRISPR, a CRISPR-powered therapeutic, to change that gene. And he is now able to normally process protein with no fear of going into these metabolic crisis. That's incredible, right?

I Yes, like, because gene therapies are expensive, rare, and we hadn't tried these new techniques.

KJ was the first— we've had gene therapies, and we have also used CRISPR. KJ was the first one where we looked at his specific mutation and designed a bespoke drug for him.

Okay.

And what we're excited to do is how do we continue going down this path and use the CRISPR-Cas9 platform to use the molecules that have been designed and to partnership with the FDA to build this into a platform where it becomes easier and faster to be able to design a bespoke gene therapy to a patient, most often a baby, and to correct a mutation that would have dramatically altered their life. And at— what we're doing in partnership with Jennifer and the IGI is to create a center at UCSF It's like this one, I feel so passionate and it sort of makes my heart, I don't know, it's both excitement and sort of, I don't know, passion, a passion, I guess that was what I call it. It's because we are literally going back to the teams that I trained with as a pediatrician that motivated me to invest in basic science. And with this program to, uh, gene edit children who are born without immune systems and without functioning livers to be able to process, um, uh, their food. And these kids otherwise would have had progressive illnesses and, um, really severe, unpleasant, long-term years in the hospital, um, and sort of never going back to a normal baseline.. And so we're going to be able to really closely tie together what we know from the basic science and deliver a therapeutic to impact people's lives.

Yeah. Like, and I don't want to give people false hope or even set timelines on some of this stuff because it's such difficult— but like, do we— so you— we feel like what we saw with baby KJ, it proved enough stuff out that we feel like we will be able to do this for multiple other conditions. On a— I mean, none of this is easy, but like, I'm trying to find the right word, you know what I mean?

Feasible?

Yeah, yeah.

Yeah. So yes, and I think you have to be super careful about the cases that you choose because there are unknowns and there are risks. And so you want to choose a case in which it would be the risks out sort of are outweighed by the potential upside that the patient has. And so we're going to be working with a clinical advisory group to work on that. You choose the cases carefully, and you work with the families, and you work with the physicians to figure out how to deploy this, and you monitor what's going on. But the interesting thing is, as you get to more and more cases, you understand sort of how long it lasts, what are the downstream effects, like, what are we seeing? Because you're worried about off-target effects. And this is where actually some of the cell modeling work comes into play as well. Because you wanna know, you know, in the case of baby KJ, we edited his liver. What are the other effects that this edit will have? And, you know, you test this in mice, you test this in non-human primates, but, you know, what you ultimately would be incredibly helpful is to be able to look at this edit in a virtual setting. And look at the downstream effects to be able to predict what sorts of consequences, intended or non-intended, you would have.

And is it gonna be all monogenetic, single gene conditions for the foreseeable future?

Yeah, we're planning at, in this model, in this investment, we're gonna, are targeting about 8 patients.

Okay.

And we'll learn a lot from these 8 patients, both from the regulatory perspective and the scientific perspective. But you really wanna know the exact mutation. So these are— we're going after really well-characterized Mendelian monogenetic diseases where we understand the pathway. But the really cool thing, and the other thing that's happening for us this week, is that we're releasing an AI model called Gremlin. And the thing that Gremlin does is Gremlin looks at non-monogenetic diseases. It is looking at the protein expression of a constellation of of your gene profile in looking at where a disease has a sort of a lever or a sort of key intersection that would help in sort of a lot of other diseases like ALS, Parkinson's, autoimmune disease. They come— a lot of factors come into play, and you want to understand where the linchpin is. So the one I find helpful to think about is like, say, autoimmune disease. In autoimmune disease, you have usually a genetic predisposition, and, you know, the dreaded variant of unknown significance, right? It's great when it's a slam dunk, you're like, I know, but then there's usually a variant of unknown significance. There's usually sort of an environmental factor, whether it's an infection or pregnancy, and then something happens where it triggers some people to flip into illness and some people don't. And what Gremlin could do is look at all these pathways, and instead of looking at a single pathway that you can correct, like we were talking about with, uh, the work that we'll do with Jennifer, Gremlin looks at the multiple pathways. And individually, it's unclear sort of at what point you should intervene, but together you can look at the intersection and say, ah, it's this protein that goes wrong when that happens, and help you— Gremlin is aimed to really help you understand these non-Mendelian multifactorial illnesses that so many of us deal with.

And so it's— and I have in my notes, but it's an application. I mean, it's a— what's the best way to describe it? A tool?

It's like an AI agentic model.

It's a model, right? Okay. And then you guys open source like a ton of this.

All of it. Yeah.

I'm keeping an eye on time. I'm sorry to do this. I have to ask you a couple other questions. Just a couple stories about you guys lately. And so you've been at this for 10 years. I think when you started, it was It was health, well, science and education. And there were a couple of buckets. And as time has gone on, it feels very— science is the priority, or at least that's my interpretation. And I think you wrote a newsletter like to this effect. You got it. So the Standard did a story and then a couple of people followed and I guess they were arguing in these pieces. They felt like politics is coming to play on some of CZI's mission and and that you guys had gotten rid of some of these other things and then emphasized science. I mean, I don't actually care because I'm pretty pro-science and money going towards science. But I, you know, I can see why some people would. I know you wrote a newsletter of this. I just wanted, you know, if there was something you wanted to address on this front and sort of how you guys are thinking about it and some of the criticism, I guess.

Yeah. You know, you mentioned this, but we, over the past 10 years, we've invested $7 billion in science, in the community, and all these different efforts. And a lot of good work has happened, and we're proud of that good work. And I, I also, like, not all of it has gone the way we expected, and some of it has been sort of— it just hasn't hit that flywheel. And meanwhile, in science, we have hit an incredible, interesting inflection point where putting more resources, we think, actually can produce outside results. And, uh, and it's something I care deeply about. This is not— I didn't stumble upon science yesterday. Yeah, this is something, you know, I've been reading that biology textbook, I've been in medicine, and I, I'm thrilled to be able to, with the work with Jennifer, to go back to thinking about patients. Like this is ultimately my life's work. And so when given the opportunity to focus our efforts and invest more in something that is going to improve the lives of a lot of people, like, I'm excited to do it.

So when you, when you see criticism like that, I mean, does it— I don't know, does it, does it bother you? Does it not? Is it just the reality of the situation? Yeah, I don't know how you process it.

Uh, I think in scenarios like that, you just— well, everyone's entitled to their opinion, especially when you do something so publicly. And I think the parts where I, I want criticism to make us better, um, and so when I get criticisms like, have they thought about this, what will this look like, yeah, that's great, and I'm going to take that and internalize it and make the work better. And then sometimes people are expressing their opinions and you just have to respect it for that. I mean, I get some—

it's that, you know, wealthy, powerful people doing big things are always going to be under the microscope. You're doing big stuff in the world, you know, like in the arena, I guess, as people say. But then there's an element to some of this. It's like, God, I mean, you're putting like billions of dollars into science. And it's like, I don't know. I don't know. Seems like a decent thing to do. Yeah, the— so you just came out with these two big announcements that'll probably be coming out just like a couple of days before this podcast airs. What else are like— I mean, it seems like there's tons of priorities, but is there like one big project you're really eyeing within the next year?

Yeah, I mean, we were lining up all of our ducks to really accelerate science. With the— at the intersection of AI and biology. Yeah, like, that's what we're going for. We want to be able to, like, build useful tools to make work like Jennifer's better. We want to build tools that help us understand complex diseases. We want scientists to be using our tools, whether it's, you know, Shaina's sensor or the engineered cells coming out of New York. And we actually have a bunch of stuff that we're building out of San Francisco that we didn't get to, but Like, we want all those tools to come together to make science go faster.

Is there, like, any worry that we— science— I am bombarded by AI meets science startups, and everyone's so excited. And a lot of the medical stuff, this technology definitely seems to work. I don't know. Is there any worry we put too many eggs in this basket? And because, you know, there's still the fear that, like, oh, these LLMs are amazing, but You know, we may be hitting a wall. We may— there might be limits to this and that we go sort of too far that direction. And then it turns out these things can't do what we thought they could do.

I would, you know, my approach here and Mark's approach here is like, let's go all in to know whether or not it was possible and we'll learn. And even if it wasn't possible, great tools and great things will come out of it. And if, you know, if we actually get to the moon, that's awesome.

Yeah.

But it's better than sort of giving it a half-hearted hedged effort because we'll never really know. Yeah. Yeah.

And I don't know if you get too into this stuff, but like the state of US— I mean, obviously there has been some cuts to NIH, like drastic cuts at some of these labs. Yeah. I don't know.

I mean, how do you feel about the state of American science at the moment and I think, uh, the foundation, the bedrock that we all get to build from in American science is incredible because of the investment that our government has made. It's so important, and the reason why we're able to take these sort of high-risk, high-reward projects with it is because the broader ecosystem exists. And I think we need to continue ensuring that, like, American science is fantastic.

Yeah.

Um, and I think sometimes we, we sort of lose the thread because the time between like a basic science discovery and impact in medicine, for instance, is decades and decades. People just lose the thread. That's partially why this case, uh, CHOP with baby KJ, is so inspirational. Because, you know, uh, when Jennifer got the Nobel Prize in 2015 and— oh, 2020. Um, she won the Breakthrough Prize in 2015. Uh, she won the, you know, she won the Nobel Prize in 2020, and today in 2025, a baby has had its life changed, his life changed.

Yeah.

And so I, I think we've, we've got to reignite people's imagination about how impactful and cool science is. And, um, and I'm not saying the system's perfect. There are imperfections in the system that I think we would all like to see improved. Like, right now we're funding scientists way too late in their career. You know this, but the average R01 goes to a 45-year-old. Yeah. And like, I would venture to say our younger people are more creative, bolder, and want to take risks.

Yeah.

Like, we— there's— the system isn't perfect, and so we need to sort of give it a boost. But it's, it's what incredible breakthroughs are built off of.

Okay. Well, I know, um, I think you've got to go and I know you're on a busy schedule. Thank you so much for coming. We have, we have a story coming on Baby KJ actually written by a doctor. I don't know when it's going to publish, but it's kind of the, the story behind the story on the regulatory journey that happened. And then I know I'm determined. I live here, so we'll do some UCSS stuff, but I want to go to Chicago and check out the inflammation work for sure.

And our, our imaging institute where we have our cryo- AM is actually just in Redwood City.

And I'm super into that 'cause it feels like we're on the verge of so many breakthroughs on imaging.

Imaging is going to be huge. Yeah. And they'll show, I mean, they'll nerd out with you. With Holger Müller from UC Berkeley, they just built the laser faceplate and lots of really tiny mirrors.

Yeah.

That has sort of blown the contrast and resolution through the roof even for the latest cryos.

Okay, well I'm into optics. All right, my very last question is a very quick one on these famous 4th of July videos. I mean, is that like just a dude's thing or are you behind the scenes or, or, uh, are you—

I'm there. There's shenanigans available for everyone. Uh, we even just filmed the Avengers video with our kids. Um, so it's, it's a whole family affair.

Okay. All right. I thought it might be. Well, I was just checking. Um, well, thank you so much for coming in. It was such a pleasure. And, um, thank you.

Thank you for having me.

Core Memory is hosted by me, Ashlee Vance. The show is produced by David Nicholson and me. Our theme music is by James Mercer and John Sortland, and the show is edited by the John Sortland. Thank you all for listening. As mentioned, the show is sponsored by Yuan Ventures. Thanks.