You Don’t See the Path, You Take the Next Step | Sujal Patel (Part 4/4)

Intro - 00:00:06: Welcome to the Biotech Startups Podcast by Excedr. Join us as we speak with first-time founders, serial entrepreneurs, and experienced investors about the challenges and triumphs of running a biotech startup from pre-seed to IPO with your host, Jon Chee. In our last episode, Sujal shared the intense EMC acquisition negotiation, calculating $33.85 per share on a Saturday phone call and serving as president of EMC's Isilon storage division for two years. If you missed it, check out part three.

In part four, Sujal unpacks taking four years between companies, evaluating biotech and clean energy ideas, and receiving an email from Parag Mallick in 2016 that said, "I think I've come up with something important". He also breaks down why proteomics matter when 95% of FDA-approved drugs target proteins, but current methods produce irreproducible data. He discusses how Nautilus's platform analyzes billions of molecules simultaneously using iterative antibody binding and why building four technical pillars took half a billion dollars and nine years. He reflects on learning to be a biotech CEO by watching YouTube chemistry lectures at 2x speed, managing PhD scientists who think differently than tech engineers, and why the early access program with Tau proteoforms presents a key commercialization milestone.

Jon Chee - 00:01:51: What an origin story. And now you're like, "Okay. Parag's not crazy". Talk a little bit about the technology and, like, the state of the market and how this not-crazy idea disrupts all of this.

Sujal Patel - 00:02:03: Yeah. Let's walk through, like, why is this an important challenge to go tackle?. So over the last decades, humanity has conquered genomics. If I take a drop of your blood, I can tell you for a few hundred dollars what all your DNA is, your whole genome. The thing is that your genome doesn't really change from the day you're born to the day you die, and your genome is the same in every single one of the 37,000,000,000,000 cells for the most part that are in your body. So it gives you a very rough idea of what color your hair is gonna be, and you might have propensity for this genetic disease, but it tells you nothing about the real-time state in your body.

All of your cells are made of proteins. Proteins do all of the work in your body; they're the little machines that do all the work. Because of that, 95% of our FDA-approved drugs target proteins. Most of our molecular diagnostics, even with this huge revolution in genomics, most molecular diagnostics still target proteins today. And that's great, but we have a very poor understanding of proteins. So with the genome, I take an Illumina sequencer, I tell you what all your DNA is, it's reproducible, it's accurate, it's a commodity. In the protein world, these proteins are very complex. Every cell has a different makeup of proteins, and all of them are in these various modified forms. There's layer after layer after layer of complexity in proteins, and all those layers of complexity are critical to understanding biology.

What we do today, the gold standard to understand proteins in a complex sample like a drop of your blood, is that we run it through a very complicated, sophisticated set of sample preparation techniques that isolate the proteins. We physically break the molecules into little pieces, and then we shoot them through a mass spectrometer, which is essentially a fancy weighing machine that weighs all the pieces that it saw. And by the weights, we back into what we think the sequence might have been for that short peptide, short piece of protein, and then we infer from that what the proteins might have looked like. It's a fuzzy process; it's incomplete; the data is not really very reproducible. And if you don't have reproducible and reliable data, you're not gonna find the best biomarkers that are indicative of disease. You're not gonna find the best targets for a drug. You're not gonna be able to apply AI to solve humanity's medical problems.

And what Parag envisioned was that there has to be a way to do things completely differently. Parag is a rare creature: half computer scientist, half biochemist. Parag's academic degrees decades ago go back in both disciplines; he's a true interdisciplinary scientist. And what he realized was that one of the methods that you can use to identify a protein molecule is kinda like how we deal with things in computing. Your iPhone, for example, if you want to go and get a very accurate read of where your GPS coordinates are, it pulls information from lots of places—the cell tower over here, the satellite above us, the Wi-Fi access point from Starbucks—and it combines all those data points together to come up with a highly accurate identification of your current location.

What if you could do that for molecules?. With molecules today, the state of the art is: I have an EGFR molecule, I have an antibody for EGFR, if there's a binding event, I know I saw EGFR, and somehow I detect the binding event. That's great, but then I have to build one antibody for every protein, every protein variant in the world. If you include the variants, there's millions of those; that's never gonna happen. And how do you do that molecule by molecule?.

Well, Parag envisioned a scheme where we take a sample, we separate billions of molecules, put them basically on a giant chessboard where there's one molecule per spot, and we probe those molecules over and over again, learning different features and characteristics about it and stack up those features—hundreds of those features—to come up with a shockingly precise identification of what the molecule is. We keep adding more and more information as we want to understand how is the protein modified, what form is it in. And Parag envisioned a system where we do that for billions of molecules at once because proteins are complicated. A thousand cells, which is a standard kind of biopsy, is 10,000,000,000 protein molecules. That kind of scale is a thousand times more than the genome, and Parag envisioned a system that would be able to operate on that scale every single day, which is what we've built.

Jon Chee - 00:06:35: Holy moly. And yeah. I mean, like, you're right. You're talking about when the very beginning is, like, Nautilus is solving a big problem. It's solving a big problem and talk about it—an ambitious one as well. And you've raised this, you said, five and a half million. You know where you need to be. When that first 5 and a half million come in, what was the first—like, your day zero? What are we doing?.

Sujal Patel - 00:06:57: Yeah. It's interesting. So let's kind of zoom out from it. We raised 5 and a half million in the seed, we raised $27,000,000 in our A, we raised $76,000,000 in our B in the pandemic, and then we went public and raised 346—so about a half billion dollars. Every time we've raised money, we've done it in a position of strength and raised money for the next milestone. Every single milestone in this company's history has taken longer, much longer than we thought and we wanted. Hard, hard technology.

Jon Chee - 00:07:28: Because Isilon was—you said between your A and B, it was, like, two years, maybe less?.

Sujal Patel - 00:07:33: A and B was a little less; B and C, it's probably like two years. Because from '01 to '05, we raised four rounds of funding in Isilon—so that's like every 18 months, something like that.

Jon Chee - 00:07:44: Yeah. Pretty quick. And exactly what you said: it's hard tech. It's very hard tech.

Sujal Patel - 00:07:48: At Nautilus, every one of those rounds of funding was focused on solving one massive piece of the challenge. One piece is that nobody has ever built technology to take billions of molecules and separate them spatially on a chessboard and space them evenly out. Because if you're gonna analyze 10,000,000,000 molecules at a time, which is what our system is built for, that means you're immobilizing molecules at roughly one micron apart from each other, and you have to be able to detect antibodies binding. It's a very complicated process. That piece of technology, which is a chip, a flow cell, semiconductor fabrication processes, a set of chemistries—that was, in earnest, like six or seven years of solid development before we really figured out how to nail that. That's one huge body of work that we had to solve over the last—what is closing in on or I guess it's nine years now.

The second is that we have to create antibodies that bind to each molecule over and over again and give us little bits of information that then we combine computationally. That is a whole different class of antibodies, and we continue to this day building those antibodies, refining our methods, and building our library of those.

The next thing we have to do is we have to build a set of instrumentation in an assay that could take a molecule and show it one antibody after the other and gather information about the molecule. There are no assays that do that. Everything is like, "Here's a binding assay like an ELISA; it's bound; I'm done; I get my data point". Here, we have to introduce new antibodies over and over again, and we have to know what the binding events are per molecule and have all that information be very clean. There's a whole new set of assay and configuration that's required to do that—that was the third body of work.

And then the fourth is a massive body of work around algorithms, machine learning, and having the ability to take all of this noisy data and come up with very precise IDs of what each molecule is. Over a course of nine years, those are the major, major workstreams that we've had to build.

And if you kinda look at where we are today—we're nine years in, we're in a year now where we're moving towards a launch of that instrument. Today, we're doing a set of experimentation where we're able to dive very, very deeply into single molecules and map the landscape of modifications on those molecules and show a set of biological insights that you just can't get elsewhere. The next big application for us—and frankly, the biggest application—is this application where we can tell you what everything that we see is without having any prior knowledge of what molecule you've given us. That's the whole enchilada that's still coming. But today, enough of those pieces have come together that we are running customer samples through the system. We're providing biological insights, all of that starting in neurology today.

We've been working with a number of partners like the Buck Institute for Aging and the Allen Institute for Brain Sciences, studying Alzheimer's disease and the various modifications on the tau protein, which is the key implicated biomarker in the space, and some really exciting work that demonstrates at least three out of four of those pillars are functioning well and producing exquisitely reproducible results today.

Jon Chee - 00:11:07: Holy moly. And I was gonna say, when you and the team were tackling this, maybe I'm just observing there's a ton of Isilon learnings that could apply here. What were some new challenges that you thought, "Here's a lesson I learned in Isilon, but, oh, it does not work in life science and hard tech?".

Sujal Patel - 00:11:24: Let me kind of walk through the various pieces. There are less things that translate over. This is a startup, and you have to have startup DNA. Not a single piece of equipment is getting bought new; don't even come to me with that PO, and everything's gotta be done on a shoestring budget. Everyone's gotta work their butts off, and we move fast; we're agile—all the startup experience is there.

Two is that the go-to-market for what we're selling—it's a direct sales force; it's selling a big capital piece of equipment. The buyers are a little different, but the selling motion's the exact same as Isilon. It's just the titles of the people have changed; it was sales engineer, now they're field application scientist.

And then a lot of the pieces of what we're doing are familiar to me. Inside of our company is a very interdisciplinary team. There's biochemists, single-molecule biophysicists, electrical engineers, machine learning engineers, and AI engineers. There's people that are working on software, mechanical systems, regulatory, and supply chain. A lot of the same sort of stuff, but then a whole set of new experiences in the biological parts of what we do.

So for me, it was a huge learning process first to get caught up on the space. The minute that Parag and I inked our deal and said, "Let's get going," Parag went off into the lab to start working on basic experimentation and prototypes. I was working on two things: I was working on the algorithmic side—I'm still a software engineer; I could still program. There was no AI back then, so somebody had to write the full algorithm because Parag had just done a prototype. That was a nine-month process for me; I wrote the first one, and now others in the company manage the algorithmic side. Part of it was raising the money and keeping the trains running on time, part of it was writing this code, and part of it was learning what it means to be a biotech CEO.

I had to start from the basics. I went on YouTube, I found biology and chemistry classes, I'd put it on 2x speed to get caught up on those. And I wrote this list every day of what I called the "dumb questions of the day" list. Parag's professor; he's very patient. I would go to Parag with the list; sometimes it was twenty minutes, sometimes it was three hours, and he would walk me through all my dumb questions. In year one, I built this massive spike in our area. That got me to the point where I could start understanding research papers. I still have "dumb questions of the day" to this day, but I probably read a thousand papers the next year. I still read probably a paper a day.

And then my knowledge started widening, and then I started hiring PhD scientists. And boy, it's interesting: they're risk-averse; they wanna complete everything completely, then bring you the result. They like working independently; there's a lot of competing things that you need them to do when they're in a startup environment. And so the ability to lead PhD scientists effectively is a really hard thing to learn, and sort of—Parag wasn't super helpful there because all he had done is an academic lab, but we figured it out over time.

Jon Chee - 00:14:44: Absolutely. Know your personnel. It's kinda like learning a new field, new domain, new personnel, and people sleep on how much you can learn on YouTube too.

Sujal Patel - 00:14:52: For sure. And it's different as well. In the space that we were in at Isilon, hiring people wasn't the hardest thing. You'd find great people out there. Here, before Nautilus, I didn't know what an O-1 visa was where I could import anyone from anywhere in the world because they have some extremely specialized, rare experience. I've got people where Parag says, "I need to hire this person," and I'm like, "How many of them are there?". It's like—there's six in the United States. Like, you have to hire some of those people because there's very specialized pieces of what we do; it's very unique for me.

Jon Chee - 00:15:29: Wow. That's amazing. And you talked about how there's these four kind of pillars during the past nine years, and now you're about to start commercializing. Talk a little bit about who are the folks that you're selling to—customers, your archetypical customer, partner, collaborator. Who are the folks that will find the most value from Nautilus's platform?.

Sujal Patel - 00:15:48: Yeah. Let's look at it from two different lenses. Our customer is either an organization that is an academic institution or a nonprofit research organization or someone who's doing basic science research or translational research that ultimately leads to improvements in human health. That's one category, and it's a very large market.

And then there's the other market, which is the more commercial market—either a pharmaceutical company building drugs or a diagnostic company that is building a diagnostic. It doesn't have to be "do you have a disease or not"; it can be monitoring therapeutic response, helping you choose therapies, and personalized medicine-type of environments. Those are the places that we're going to sell, and that represents billions and billions of dollars of market opportunity.

Inside of those customers, there's a bunch of different types of buyers. There's a type of buyer who is already using the mass spectrometer with a complex workflow. They already know how to analyze proteins, and they probably love their mass spectrometer; they give it a hug in the morning. It's hard to get them to think about something a little different. There's another type of customer who has been doing genomics and they wanna get more involved in proteomics. They're more likely to change and want to try something new, but their knowledge in proteomics isn't super high. There's also the biologists—the people who are looking for the answers; they don't care how the answers come. They have the most need for the data and understanding of how novel information can provide them with a better outcome, better drug, better diagnostic.

Ultimately, those are the buyers for us. Today, we're largely focused in neuro over the course of the next few years. When you look at where we'll go in the long run, it's really to try to get the early adopters to start getting the product into the hands of their end users, get that first instrument in there, start to grow the number of use cases, and get them to call their friends—all the traditional things you would do as a startup.

Jon Chee - 00:18:30: It reminds me of when you're talking about Isilon and having to convince someone who's like, "No, I have all my storage ready," and it's kind of similar to someone here who's hugging their mass spec.

Sujal Patel - 00:18:41: But that's always the case. When you have a piece of technology that you have built your career around, that you've used for ten, twenty years, you develop some emotional attachment to it. A lot of the mass spec researchers have built their own custom techniques on top of it to improve the results. And it's complicated—if we come in and say it's 10 times better, but they have decades worth of information and data that's not on our platform, how do I compare?. It's a little complicated transitioning to something new even if it's as disruptive as Nautilus's product is built to be.

Jon Chee - 00:19:18: For sure. And I guess the question is: how do you communicate that? How do you get them across the line?.

Sujal Patel - 00:19:25: Getting them interested at Nautilus is something that doesn't take very long at all. At this point, we've talked to probably more than three or 400 potential customers. The pattern is always kind of interesting: you explain what we're out to do and show them how it works. This is such a unique assay and methodology that as we're talking, they try to say, "Oh, it's like an ELISA; it's like this Olink assay". And then at some point, the light bulb turns on, and they're like, "Oh, it's not like any of those things". Their facial expression changes and they're like, "Oh, now I get it. Can you build that?". And then suddenly, it's like, "I wanna stay up to date. Tell me about the next thing". They get incredibly excited about what we're doing.

Jon Chee - 00:20:25: Very cool.

Sujal Patel - 00:20:25: Takes less than an hour, but there's some inertia with all the other products that are out there, and they get over it.

Jon Chee - 00:20:31: You kinda just, like, hold their hand all the way through. It's like a lot of high hand-holding. I love that. And you're talking about how building this organization, you have to get a lot of specialized talent, and you have to—and then I remember earlier on too, you were talking about the importance of, like, promoting from within. How are you thinking about Nautilus from just a company-building perspective? What is it like working at Nautilus—a day in the life of someone at Nautilus?.

Sujal Patel - 00:20:56: Nautilus is a company where we attract people who want to do big things. What we're out to do is really hard; it is full of frustration. If you're in our R&D organization, something's not gonna work every single week, and you have to have the mindset to persevere through it. You've gotta be able to be agile and flexible. We have a collaborative environment, a really tight-knit team, and they're all really focused on building this product and getting to this huge dream.

Often when we're hiring, we are prioritizing raw potential and raw talent over the experiences that you've had. If you've got the talent and the energy, that's the type of person that we want. It doesn't matter if your resume doesn't look linear; we want someone who has a high ceiling, is passionate, and is willing to work hard and learn. When you hire those types of people, you end up relying on them to do really critical things early in their career.

You also end up in a situation where there is quite a bit of churn. In a startup, you have to have the stomach for a lot of churn because a lot of things change in the nine-year journey in terms of what you need in people for the scale they're at. Priorities change—for example, after we went public, we changed our entire strategy of building probes from aptamers to actual real antibodies. It was a massive shift, and some people have to get moved around. It's a very dynamic environment. Otherwise, it takes twenty-five years and $10,000,000,000 to build what we're building. Half a billion dollars sounds like a lot, but it's barely what we need to get this thing done.

Jon Chee - 00:23:09: Yep. And it—it almost sounds like it's like a living organism.

Sujal Patel - 00:23:12: For sure. Every day, it's changing.

Jon Chee - 00:23:15: Yeah. Absolutely. And the work that you guys are doing sounds incredibly exciting, and I'm rooting for you guys. It's also cool that you dove in and just learned up and are fully embracing it. And something that I'm noticing with your journey too is this constant learning machine element: you're always absorbing and learning and building these accretive experiences. And it shows in your hiring—do you have the hunger and high potential? You can't train hunger either. So I love that that is the litmus test too. So as you're looking forward, one, two years—I know commercialization's on the horizon. Talk a little bit about what's in store for Nautilus, one, two years out.

Sujal Patel - 00:24:13: So you have to think about Nautilus almost in two stages. Right now, we are in a highly constrained mode where every single dollar of spend is scrutinized and focused on getting the full Proteome product out the door. It is a very tight environment; if you go to our website, there are very few open jobs. We have things to prove before we go and raise more money to get to where we wanna get in the long run.

But we're preparing for a phase of what we expect will be hypergrowth. The product we're selling has an initial deal size of a million dollars—the instrument, site prep, install, software, service, and support. As you start to scale in a few years, your consumable spend every single year on that instrument could easily reach a million dollars as well. There's the ability to ramp revenue very quickly with a high-priced product where the comparables in the market are high-priced. I know where the 15 to 20,000 mass specs in proteomics are in the world; it's easy to reach those customers. So we expect to ramp quickly, which means we have to be ready to hire aggressively on the sales side and have a strategy in place.

We're laying that foundation right now, getting ready for commercialization. One of the key guys is Ken Suzuki, who I hired as Chief Marketing Officer. He's responsible for our product management and go-to-market strategy. He spent decades at Agilent running the mass spectrometry group there; he's a guy who has all of the experience and passion to build those strategies. All of that foundation is being laid today for that point where it flips and we enter into that hypergrowth revenue environment.

Jon Chee - 00:26:49: That's awesome. It kinda reminds me of Isilon when you got brought back in and you're starting to build the foundation again. But it sounds kind like an analogous phase right now for you guys.

Sujal Patel - 00:27:34: Yep. That's right. Isilon was a hard product to build, and the only company that followed in its footsteps was founded by three of my engineers at Isilon. But compared to Isilon, this is much harder and much harder to replicate. If you look at our guidance to get the full proteome product out roughly ten years after founding, with about a half billion dollars raised—even if I gave a large company the recipe, if they said, "Sujal, come be in charge of that project," I would never do it again. It is so difficult; I don't know how any large company would get it done in a decade. You have to build four distinct, extremely hard pillars and get them all to work together.

Jon Chee - 00:28:35: And the amount of work and heartache that goes into these things—they really are just like little miracles, and I commend you for embarking on something as hard as that. And I guess—when you were doing Isilon and obviously now Nautilus—when it comes to supply chain, were there lessons that you took from there, or is the supply chain on the Nautilus side a completely different beast?.

Sujal Patel - 00:29:13: They are very different. Isilon is commodity hardware—computer case, motherboard, memory. Other than one custom board, everything else is commodity, so the supply chain is more simple. At Nautilus, it's complicated. There's a very complicated imaging module; the stages move at very high speed; there's temperature and climate control inside the machine; it's sensitive to vibration. There's thousands of parts; everything's custom. It's a much more sophisticated product to build. And the flow cell and chip is extremely complicated—built with semiconductor processes and tons of surface chemistry. It's a sophisticated animal.

Jon Chee - 00:30:52: It really is. It just sounds like you turned everything up to 11 on intensity and difficulty. I'm super excited for you, and I can't wait for when the end users actually get to utilize the Nautilus platform because it sounds like it's gonna be a huge unlock. As we're wrapping up this conversation—thank you, Sujal, for your time and insights. In traditional closing fashion, we have two questions. One, would you like to give any shout-outs to anyone who supported you along the way?.

Sujal Patel - 00:31:45: Boy, shout-out. In these two big company journeys, the people I've relied on most are my board of directors. At Isilon, folks like Matt McElwain at Madrona Venture Group who bet on us at the beginning; folks like Greg McAdoo from Sequoia, who is a dear friend; Bill Ruckelshaus, who was my chairman—a public company CEO and head of the EPA. At Nautilus, some of the same names—Matt McElwain is on my board again; I've been on four boards with Matt since 2001. Matt Posard, our chairman, built the commercial organization at Illumina. Those are the folks I rely on. Being CEO is a lonely job. My wife doesn't wanna hear it, let me tell you.

Jon Chee - 00:33:05: If you could give any advice to your 21-year-old self, what would it be?.

Sujal Patel - 00:33:21: There are tons and tons of things in my career that required absolute perseverance. It doesn't matter how hard the thing in front of you is: go sit down and figure it out. Figure out what the possible solutions are and how you're gonna get it done, even if they're hard. Figure out the right solution and just take your next step. I tell people that a startup journey is not one where you see the end goal; you don't even see the path in front of you, and you just have to have faith that next step is there when you take it.

Jon Chee - 00:35:21: Thank you again. You've been so generous with your time.

Intro - 00:35:49: Thanks for listening to our four-part series featuring Sujal Patel—from New Jersey to RealNetworks, founding Isilon through the dot-com crash and $2,600,000,000 EMC acquisition, serving as president of EMC's Isilon storage division, and now building Nautilus as CEO after raising half a billion dollars. Sujal's story shows how entrepreneurial perseverance combined with technical depth can revolutionize entire fields.

Join us for our next series featuring Richard Yu, co-founder and CEO of Abalone Bio, a therapeutics company developing functionally active antibody drugs for challenging membrane protein targets like GPCRs. Richard's journey from molecular biophysics at Yale to pioneering GPCR antibody agonists demonstrates how synthetic biology combined with machine learning can unlock previously undruggable targets. The Biotech Startups Podcast is produced by Excedr. Excedr provides research labs with equipment leases on founder-friendly terms to support paths to exceptional outcomes. To learn more, visit our website, www.excedr.com.