Arvind Gupta on Innovation, Investing and Planetary Health

Mayfield Partner Arvind Gupta recently appeared on “Move Fast and Fix the Planet,” a new podcast about climate and sustainability entrepreneurship from STVP – the Stanford Engineering Entrepreneurship Center. Listen to the episode where he discusses his mission to invest in science-based companies that could change history with Professor Mike Lepech, or read the full transcript below:

Arvind Gupta: We don’t innovate until there’s pressure to innovate. Problems don’t get solved, because they exist. Problems get solved, because there’s a reason to solve it. And so now there is pressure to solve problems that we haven’t really bothered to try to solve. And we’re seeing new technologies that are able to solve it with great unit economics that is not philanthropies, that they’re actually businesses. And so, what gives me hope is I do believe capitalism is going to be able to solve climate change and reverse climate change for humanity.

Michael Lepech: Hi, and welcome to the Move Fast and Fix the Planet. I’m your host, Michael Lepech, Professor of Civil and Environmental Engineering at Stanford. I’m also an associate faculty director of STVP, the Stanford Engineering Entrepreneurship Center, where we empower aspiring entrepreneurs to become global citizens who create and scale responsible innovations. Of course, one of the biggest global challenges we face is climate change and the sustainability of our planet. In each episode of this podcast, we’ll talk to a different expert about entrepreneurship in climate and sustainability, and what’s different about it, if anything, from entrepreneurship in other spaces.

On today’s show, I’m thrilled to welcome Arvind Gupta. Arvind is a partner at Mayfield, focusing on investments in human and planetary health, where his mission is to invest in science-based companies that could change history. He’s also the founder of IndieBio, a development program for biotech startups trying to solve major world problems. Prior to founding IndieBio, he was design director at IDEO in Shanghai. Arvind is also the author of Decoding the World: A Roadmap for the Questioner, a firsthand account of the science that’s shaping our future. He was honored with the F50 Global Award for Impact in HealthTech Innovation and is a frequent speaker at TechCrunch Disrupt, Slush, TEDx and Future Food Tech. Arvind received his BS in genetic engineering from the University of California, Santa Barbara, and holds eight patents. Welcome, Arvind.

Arvind Gupta: Thank you so much for having me. I really appreciate it.

Michael Lepech: Yeah, this is a lot of fun to have you here today. I want to dive right in, and you’ve been a pioneer in investing in science-based companies that look to tackle some of the world’s most pressing challenges. You’ve looked at reinventing the food system, to combating climate change, to a host of other really tough problems. How did you first get involved in human and planetary health? And maybe if even before that, you can discuss a little bit about what does planetary health mean to you?

Arvind Gupta: Yeah, thank you. Thanks for that question. It was a long and winding road to get to today, and I’ll share some of that with you guys, because it is interesting and there are some lessons that I’ve drawn from it. But to address your second question first, planetary health just refers to the idea that our planet has health, just like human health. People, we think of our own bodies and we get sick, we take medicine, we heal ourselves. Our planet also gets sick, and we have been making it sick. And there are ways we need to give it medicine to make it better and make it better for everyone, because when the planet gets sick, people die and a lot of people die. It’s projected over the next 50 years, more people will die from climate change related activities than all diseases combined. So, that’s a significant impact.

Also, as the planet gets sicker, what’s going to happen is our resources become much, much more scarce. And with scarcity, prices get driven up. Inflation goes up, as everyone talking about today, and that causes prices to rise and it increases inequality. So, planetary health is linked to two of the biggest problems that I see in the next 50 years, which is climate related deaths and rising inequality. Those are big problems to tackle, but the good news is human ingenuity and capitalism, combined with opportunity, can solve these problems in a way that we’ve never seen before.

Michael Lepech: I really like the way that you started with grounding this in the environmental health of our planetary ecosystems, but then interestingly, and I didn’t necessarily see it going this way, you use the economic piece to then connect it over to equity and inequality. And that brings together that triple bottom line that we often think about when we talk about sustainability. And it also relates it very closely to the topic for this podcast, which of course is a capitalism investing venture type topic. And so, I really appreciate the way that you were able to weave all of that together. So how did you get into this?

Arvind Gupta: Yeah, well, it’s an interesting question. I didn’t grow up knowing what venture capital was. I didn’t grow up knowing what entrepreneurship was. My dad’s a professor at UCLA, my mom’s an accountant. They’re immigrants from India. I’m first generation, grew up in Los Angeles, and it was in Van Nuys, which isn’t the world’s best neighborhood. And so, this idea of going out on your own and building a world changing company was completely foreign to me until just recently, really in the last decade or two, last 10 years. And so I grew up thinking, “Okay, I’m going to be a scientist.” And so I went to school at UC, Santa Barbara. I studied genetic engineering, got interested in economics through thinking about how DNA self-assembles and economics is basically the self-assembly of society. And so, thought there was overlap there.

Michael Lepech: Wow, that’s a very interesting jump there from saying that, “We were studying the genetics of DNA and the assembly,” and then you connected that to, “Wow, I think this is a lot like economics.” I just want to recognize that. That’s an interesting jump that I don’t know everyone would make.

Arvind Gupta: Yeah. No, thank you. So the way DNA coils itself, this is in 1992, ’93, ’94, it was a lot less known about genetics back then. And so in thinking about; how does miles and miles of DNA filament pack itself down into inside a tiny little cell, in a nucleus of a cell, and then unpack and get read to drive the instructions and then refold back up to… It’s fascinating to think about that. And so for me, just thinking about what are all the things that create this self-assembly, it just naturally led me… I don’t know. My curiosity led me there. And then I remember my first economics class, I was hooked. It was macroeconomics, introduction to macroeconomics. And the first lines of the class, the professor said, “Human desire is unlimited. And because of that, we have the study of economics, because economics is a study of scarcity.”

And I think that stuck with me in a way that I still to this day think about. And it drives the way I think about investing in planetary health, because I do take that to be a fundamental truth, a truth as much as gravity. And so, if we think about that and then look at the economic system we have, capitalism, capitalism will always deliver what the consumer wants always. Always.

Michael Lepech: I guess the equilibrium of economic theory is in itself a self-assembling construct and that equilibrium that we achieve in markets, whether they’re micro or macro in scale, I can absolutely, now that you’re discussing it, see how you would make that jump. So you’re studying genetic engineering, and then what next?

Arvind Gupta: So what next? Yeah, we could continue on tangents for several hours. So basically, genetic engineering I saw was extremely powerful, even in the mid-’90s. When you think of it at a high level, the ability to reprogram life, it’s just obvious that it’s going to be powerful beyond belief. The issue was back then, it was really a set of protocols. It was not a technology yet. It was restriction enzymes and random walks. And the things that we were doing were extremely basic by today’s standards and what would take me a week or a month of prep, now you do overnight and by mail, by mail order. And so I thought to myself, “Okay, well, I don’t know really, I don’t want to be a doctor, and I certainly don’t want to be a bench scientist. I know this is very powerful, but what else can I do with my mind?”

And so I go back to economics and I say, “Well, okay, if there’s a lot of different variables, all that are changing at the same time, maybe I can try to predict how those variables play out and make a bet on it. Where could I do that? The stock market.” So I moved to San Francisco and stood on the steps of the Pacific Stock Exchange. And everyone that came out, I asked, Hey, could I get you some coffee or something? Just give me a job.” And one guy was like, “Yeah, I’ll give you a job and you can get my coffee and check my trades.” So I got on the floor. There I started to learn what was going on. I asked a lot of questions, and a small trading firm backed me to be a market maker in the Microsoft pit. That’s where I learned making money without creating value leads you to a life of consumption as a way of measuring yourself.

Michael Lepech: Interesting. I’d also like to point out someone took a chance on you.

Arvind Gupta: Oh, many people over and over again in life have taken chances on me.

Michael Lepech: Yeah. And so now, I think ultimately we’ll get to you as a venture investor where your job is in some extent taking a chance on people, but I think that’s an interesting point in life when you think about the people who took a chance on you.

Arvind Gupta: Yeah, I can name… You’ll hear in my story crucial moments where I would certainly not be where I am today if it wasn’t for a number of people that believed in me. And so, because my life goes in a way that it wasn’t a straight line, and so therefore, when you go in a different direction, someone has to take a chance on you, because you’re not obvious. So anyway, I got on the floor, started trading, realized that this wasn’t the life that I envisioned for myself, and I knew I could do it, I knew I could make money from trading and options, market making. So I said, “Okay, well, let’s run the experiment. If I do nothing, whatever it is I do, when I do nothing, if I can make a living doing that, then I’m on vacation for the rest of my life.”

So, I decided to run the experiment, close my positions down, walked out the floor, and that’s when I learned basically to find the art side of my life. I started reading a lot of the classics. I basically lived in Moe’s Books in Berkeley. And I started to write poetry, things like that, make my own furniture out of driftwood. And a friend of mine, who I was teaching how to bass jump, I was a bass jumper at the time, and a big wall climber. He was like, “Oh, you should be a designer.” And I was like, “What’s that?” And he’s like, “It’s solving problems with science and art and blend it together.” And I’d never heard of design ever in my life. I was like, “You could make a living?” And he was like, “Yeah.” And so I enrolled in a master’s program at SF State. That was another person that took a chance on me. I had no background in art and Stefano… I was, “Hey, how could I be a designer? All these kids know how to draw and all this stuff.” And he’s like, “It’s not about drawing. It’s about life experiences and using that to solve problems. It’s about the ideas, that matters far more.” And I said, “Oh, okay.”

So that gave me the courage to even try. And then the head of the program took a chance on me that here’s a guy with no portfolio, no nothing, “Yeah, sure, come in.” So, two years in, I had a student show and a global lead of industrial design from IDEO, a guy named Paul Bradley, saw my work and he said, “Hey, you should come and work with IDEO.” And I said, “Okay,” so he was another guy that took a chance on me.

Michael Lepech: He’s coming from the IDEO office in Palo Alto, I imagine?

Arvind Gupta: Correct, with David Kelley, Tim Brown, the people that you know at the d.school for Stanford. And so that’s where I started really understanding. So design was formative for me, because fundamentally I’m still a designer. And what that means is I look at the world through the lens of finding problems and problems are far more important than solutions to me, because problems are persistent. You never just solve a problem. It’s funny, right? Humanity has the same set of fundamental problems that it did a million years ago, or even 200,000 years ago as homo sapiens. And we just solve it with different technologies. And human connection is the same fundamental problem. Now we have TikTok and Tinder and all these apps, Instagram.

Michael Lepech: Zoom.

Arvind Gupta: Zoom, we have… But the connection hasn’t been solved. It’s not like, “Okay, we can move on from that.” I’m hoping climate change is different. I actually hope we can solve that and move on from it. And I think we can, we’ll get to that in a bit. But problems are really… I always said as a design director, the better the problem, the better the product. And so you might as well spend your time honing and refining your problem. And that includes all the constraints around your problem, because the more that you find the constraints, the more your product design is itself in the right way. And if a designer is spending too much time trying to design, they haven’t found the actual problems they’re trying to solve, and their solution is going to suck. Their design is just not going to be something that resonates with people, because it’s too much of the hand of the designer and not the hand of the problem.

So from there, I always thought; what else can design do? How do we find more constraints? I started building businesses through product, as we started expanding what design could do at IDEO with business design and things like that. And I was doing a lot of the technology products and designs to the Samsung Galaxy Curve. I did a project with LG where the design brief was; design a $500 million business. And we did that with McKinsey. So I learned all these tools, and that’s when I was in Shanghai, my wife had this idea for a fitness startup when we were walking around in Xi’an. She built it, launched it three months later, and it was being used very quickly all over the world, thousands of people. She was getting emails from people saying, “Hey, you changed my life, because you helped me get started on my fitness journey.” And I was like, “Man, I’ve designed all this stuff, I’ve won all these awards, but no one’s ever emailed me that I changed their life.” It was a remarkable experience for me.

And so we were coming back to the Bay Area, and my friends had left and become early stage employees at Square and other startups. And this is 2013. And I met this guy named Shaun O’Sullivan who was running basically a family office, but he had this idea to start these accelerators that are vertically focused. And he had one in Shenzhen called HAX, and it was a hardware accelerator, first of its kind. And I saw that, and I thought always back to genetic engineering, I never lost my love of science. I kept up with papers, so I knew the advancements that were happening in the field, and I was thinking about blending them in with design the whole time. And I wrote a paper actually in the Journal of Commercial Biotechnology in 2011 that you could look up that asked the question; what happens if you mix design thinking with scientific method, and you can actually speed up R&D?

So, I saw this thing that Shaun was doing as an idea, as a test bed for what design and investing in early stage biotechnology could do. I thought, “Okay, well, what would a vertical accelerator for biotechnology, synthetic biology and genetic engineering look like? And what can we do with it?” Well, when you reprogram life, you make it more efficient. What’s the most pressing problem of our time that I could spend the next 50 years of my life working on? Climate change, disease. These are the things that are persistent don’t go away and expertise in that can continue over time. So I basically said, “Shaun, look…” He was saying, “Hey…” I gave a talk at HAX, and he was like, “I think you should be a VC.” And as I thought about it, I thought, “Okay, yeah, you’re right. We could do something like this and reinvent what financing early stage biotech companies look like.” And all of these postdocs were coming free. And there’s a phrase called the postdocalypse, where postdocs basically don’t have jobs to go to, because professors aren’t retiring and there’s more postdocs than positions available.

So, I saw this stock of human talent, real talent, it takes a lot of talent to be a postdoc, and they’d had no options. But if you give them a chance to work really hard and work on a hard problem, I bet they could solve it. So I proposed that to Sean. Sean agreed. So I joined SOSV, as a partner, and built IndieBio here in San Francisco in the Bay Area. And that was the simple premise, but no one really believed it at the time, because one biotech had to be expensive, and I was proposing financing it for a 100K, which is laughable, literally laughable. It’s naive to me even today to think about that, and that postdocs could be entrepreneurs, laughable. No one believed that. And that biotech had to be only for drug making, which you could make things cheaper and more efficiently. So, why wouldn’t you have economic value there?

So anyway, those things turned out to be true, and IndieBio took off, and we funded companies like Memphis Meats, now, Upside Foods, the first cell based meat company, Clara Foods, now EVERY Company, egg whites without chickens, a host of agricultural companies and a couple dozen therapeutics companies. The combined market cap is north of $5-$6 billion today. And about three years ago, I realized after building a New York office, I really wanted to work with fewer founders deeper. And I hired Po Bronson who I wrote a book with, and I knew he was more than capable of growing IndieBio and keeping its true north. So I thought to myself, “Well, why don’t I learn how to help these companies grow all the way to maturity and join Mayfield as a general partner in order to do that?”

Michael Lepech: Because 100,000 doesn’t get you very far.

Arvind Gupta: No, but the ecosystem of follow on financing does.

Michael Lepech: That’s exactly right.

Arvind Gupta: That’s right. And so, well over a billion dollars have been raised by IndieBio companies. But I was just spread really thin, and I worked very closely with every single company that moved through IndieBio for the first 10 batches. Every single company I worked with on a personal basis for four and a half months, and then stayed with them afterwards. As too many founders, you just don’t have time, so now I’m focused much, more focused, and selective in the problems that I’m working on. So yeah, that takes me to today.

Michael Lepech: Yeah. Wow. Well, so that leads me to my next question, which with regard to having a front seat to seeing the evolution of startups in this space by being such a force in IndieBio and then now being able to do rounds of follow on at Mayfield, do you think that investing in the climate and sustainability startup space is different than other sectors, whether it’s tech or medtech, or any other vertical for that matter?

Arvind Gupta: Yeah, it’s very different. The fundamental physics of business are always going to be the same, but the companies in planetary health often have different business physics, because of their sectors. In other words, software typically has very low manufacturing costs.

Michael Lepech: High margin.

Arvind Gupta: Copy, paste. So that generally leads to high margin. Now, that’s not always true, because oftentimes these companies spend a lot of money on marketing, on distribution. With the recent AI companies, manufacturing costs have actually gone up, because you do have to pay to train models or do API token calls. So, some of that’s changed. But basically, planetary health companies oftentimes are physical products, and so therefore you have very different gross margin structures, fundamentally.

Michael Lepech: Capital intensity is very different.

Arvind Gupta: Much higher capital intensity. Oftentimes, scale is the one thing that gets you to positive unit economics, which means $200 million might need to go into the company prior to seeing any real profit margins. You might see revenues before, but you’re paying through the nose for them. So, those need to be very respected, and you need to design businesses that could survive that kind of growth. So, through the valley of death, which is you have to get through that capital intensity in order to get to positive unit margins. And that’s typically called the valley of death. So, it is different that way, but it’s not different in the entrepreneurial pluck that’s required, the animal spirits that are required to find customers, land them, deliver to them, and have them order more. All of that is required. That’s the same.

Michael Lepech: Yeah, the business of business is still business-

Arvind Gupta: Correct.

Michael Lepech: … as they say. But that brings up an interesting question, and I know this is an odd question to pose to a venture capitalist. Is venture capital the right funding mechanism for these types of companies?

Arvind Gupta: Yeah, it’s a great question. I’m glad you ask it. There is no one universal tool. And so, a lot of these companies have different business physics, which lend itself to different rates of return. There are vast amounts of capital seeking different rates of return. So venture seeks extremely high rates of returns, the highest on the planet, the highest on the planet that I know about, and therefore only certain business physics will match that.

Now, for physical products, that means a very, very high total addressable market. So, $50 billion total addressable market, or more, preferably a lot more than that. So there aren’t many categories that have that high a TAM, which you could then make up for a lower gross margin in revenue. So let’s just say your price to sales or price to revenue multiple in your sector is usually, I don’t know, three. I’ll make up something. Consumer packaged good is usually lower than that, 1.5 to 2.5. Let’s just say it’s three. You could still get to a multi-billion dollar valuation, $10 billion on 3 to $5 billion of revenue. Now, it takes a long time to get there. It takes longer… So you have to figure out how you’re going to fit within your time horizon, investment horizon, which is 10 years typically for venture funds. You have to figure all that out, but if you can know your destination, you can design your way there. And I’m a big fan of destination analysis; where where am I going? What does it look like? Paint that picture and now draw your map and you have your assumptions and go test your assumptions on the way to victory. So I think that’s the way to think about it.

Michael Lepech: One of the things that we like to call that in our book is future back planning, draw out your future and then back plan.

Arvind Gupta: Great phrase. But is that venture? There’s private equity, there’s other pools of capital, there’s debt financing that have different return requirements, that might be better suited for different types of businesses. So I think that’s something that great founders can do, or even different times. Seed stage can be venture, but later on you can go find alternative non-equity forms of financing.

Michael Lepech: Yeah. Well, I guess that connects me to my next question is that you talk about the returns for venture being very high, but of course that’s necessary to offset the extreme risk associated with most venture investments and the Power Law of returns. If you’re only going to get a small fraction of companies that do provide returns, then they have to offset everything else. And so, do you think there is a model in this space where maybe the risk, because the TAMs are so large, because these problems are so fundamental, that maybe the risk of failure is lower and therefore you don’t need to get those high returns on the winners that you do pick? Or is the risk of a failed startup, in your experience, equally as high, and so therefore that Power Law that is the fundamental, underlying principle of venture portfolio theory, does that still hold in this space, or can it be broken?

Arvind Gupta: It’s a good question. I think in the end it goes back to what are your goals? So, our goal at Mayfield is to be a top fund, consistent top fund, which it is. And so, that means having consistent top returns. Top returns are always driven by outliers. And so, you can get more consistent. Let’s just say we’re going to do CPG. You can do something where you’re doing 200 to $500 million exits by building a certain revenue that then gets flipped at Nestle. But one, you’re not going to solve climate change when you sell to Nestle, generally, because they have different goals. And two, you’re not going to generate the world-changing types of returns either. So I think that goes back to the goals. For me, I want to invest in companies that can make history, and that means reinventing or reshaping sectors. And Tesla is a great example of a sustainability company in my mind that is a better car, and cars are a $1.5 trillion TAM, and it’s one of the great successes of Silicon Valley. And it’s an example of the number of sectors that are going to be reshaped by entrepreneurs in planetary health and the types of returns that are going to happen in doing so.

Michael Lepech: I think that’s a great example. If you were going to advise a startup founder in this space of just one or two things that you would make sure are right in order to talk to, whether it’s somebody at IndieBio to get some of that initial funding, or to go on to pitch to a general partner at Mayfield and say, “Look, I’m in the climate and sustainability space, but this is why I’m a great investment,” what are those couple things that you would advise them to make sure are right?

Arvind Gupta: Yeah. Well, one, I think they have to be uncompromising. What I look for is uncompromising founders. It’s the one key trait that I see that’s been common to all the great founders that I’ve backed in my own personal experience. And the one common trait when I look out to the entrepreneurs that I admire, people like Elon Musk and Steve Jobs, Yvon Chouinard of Patagonia, all of them are uncompromising, 100% uncompromising, and that leads to controversies at times. Steve Jobs was not an easy person to work with. Elon Musk certainly not an easy person or personality. I don’t know Yvon Chouinard as well, but they were all uncompromising in their values and what they were trying to accomplish. So I think if you go in to pitch, show how you’re uncompromising in what you’re building and how you’re going to use the market forces to solve the problem that you’re setting out to solve.

Second, I think you need to have an amazing team. No one person can build a company. And I think there’s a lot of founders that think about the product they’re building versus the company they’re building. And founders that think about the company they’re building invariably go much further. Finally, the physics of what you’re doing. If there’s only one thing to think about in terms of business physics, it’s the product. How are you 10 times better than your closest substitute? Now, that also determines your very specific total addressable market. So Beyond Meat was 10 times better, arguably, than a Boca Burger, a veggie burger, the frozen… It was not 10 times better, it still is not 10 times better, in most people’s opinion, to a ground beef from a cow that you could buy at Whole Foods or whatever. And so meat eaters don’t switch to Beyond as fast as everyone hoped, because it’s just not better yet, it’s not better health-wise, it’s not better taste-wise. So the market size of Beyond is actually a lot smaller than the market size of meat. It’s the market size of veggie burgers. So that’s how you get to think about what your actual TAM is.

Michael Lepech: Yeah, I often get questions of when I say, “You must be 10x better than your next closest substitute.” Folks will often ask me, “Well, what does that mean? Does that mean 10x cheaper? Does that mean 10x faster?” And what I tell them is you have to understand your customer. What does your customer value and what their most important value proposition is, you have to be 10x on that metric. And if you’re asking the question, “How do I need to be 10x better?”, you don’t know your customer well enough.

Arvind Gupta: Yeah, that’s a good way to say it. I think for me, there’s two parts to it. One is a simple definition of 10x is once I use the product, I can never imagine going back to the old way. So once you use an iPhone, you can’t imagine going back to a Blackberry. Once I use a Tesla, I can’t imagine going back to a gas car. These are 10x products. Now, can you be 10x cheaper? Can you be… Absolutely. So the whole idea to me, so the second part is understanding the dimension in which you’re 10 times better than anyone else. So, does a 10 times cheaper car have a place? Yeah, it definitely has a place in the world. So, that’s what I mean by then you’ll understand what your true addressable market size is, because how many people want a 10x cheaper car? That’s your actual TAM, not $1.5 trillion. So, when you get very specific, it’s very helpful. And that’s what new market creation does, is it takes from the total TAM into where you’re going. So, I think that’s the most important way, that’s why it’s a simple idea, but when you take it very seriously, it becomes a very powerful tool.

Michael Lepech: Yeah. I like that a great deal. So one of the things you had said a little bit earlier really caught my attention, “When you combine design thinking with scientific method, you can accelerate R&D.” And I really like that, because to me what that says is that the first couple stages of design thinking always focus on customer empathy and the definition of a problem. Whereas in the scientific method, we start with asking a hypothesis. And we’ve seen a lot of interesting work come out around hypothesis-driven innovation at startups, and it’s always an interesting mix, but you have a background in design and creative thinking and shaping solutions. How do you see all of this coming together as it relates to planetary health?

Arvind Gupta: Yeah, so for me, the simple way to think about it is scientific method does not presuppose a destination. Design thinking starts with a destination and comes backwards to today. So when you blend the two, you actually get a very powerful tool to accelerate research and develop to solve a problem. So that’s the high level takeaway. And so when it comes to planetary health, every single sector that we have has carbon flowing through it. So when you think about… A simple way to stratify the world is all the carbon humanity creates flows through three general buckets; one is feeding the world, the second is building the world, and the third is powering the world. And that is $100 trillion of market value in terms of global GDP. It’s an incredible amount of opportunity.

So when you think about it, let’s just take energy, I’m just making stuff up really quickly now, let’s just take airline travel, how do we decarbonize airline travel? You get to really start thinking about, “Okay, we have a range of advancements in biology, physics, AI, all of these in pure science.” You’ve got these problems out there, and you’re basically taking from this toolbox of scientific advancement and applying those advancements to that end problem of, say, airline emissions. And then you work forwards using the business of physics to understand how you can create a sustainable company that lasts 20 years and reshapes the sector.

It’s that simple, and great entrepreneurs do this implicitly. So what I’m saying is, it shouldn’t be a shock to anyone, it’s pretty obvious when I say it out loud, but that’s how I believe every single sector we have is going to get reshaped. And I’m very positive and hopeful about the future, because I’m starting to see market forces start to ramp up and start to reward companies that are thinking this way. Oil and gas really is paying for solutions. Now, call it greenwashing or not, they’re paying for solutions. If you look at the fashion industry, they’re paying for solutions. MycoWorks is alternative leather made from mushrooms that was funded by IndieBio, and they’re doing great. So food and ag, enterprises are paying for alternative protein. So, I think that’s where I see how all of these scientific advancements are going to be funneled into products, delivered through capitalism to fulfill the unlimited desires of people. But in doing so, we create a sustainable capitalism.

Michael Lepech: Yeah. I would agree with you that the opportunity is not just large, but unparalleled in human history to some extent. But that opportunity will take time and capitalism and markets take time. And so how do you think about balancing the need for urgency to treat a planet, that is not the healthiest it’s ever been, with the patience and resilience that’s required for building really big solutions? Those two things, how do you square them?

Arvind Gupta: Well, historically, private sector has started the big ideas, and then the public sector has scaled them. So, it’s interesting, fission reactors started privately, and then the government helped figure out funding, how to scale them into actual workable reactors. I don’t think this going to be any different, because as the problem becomes bigger and bigger… Unfortunately, we as a society focus on problems that are very short term. We don’t tend to give stock to problems that are smaller and further out, because we just don’t think it warrants our attention.

Michael Lepech: Going back to your economics analogy, we have a very different way of discounting future value in our minds than things that are valid today.

Arvind Gupta: Yes, absolutely. And so I think what’s going to happen is as climate change gets worse and crops start getting lost, people’s economic livelihoods start to get lost, migration becomes… Well, yeah, the governments are going to step in. There’s going to be a lot more money flowing through. IRA has already fueled a lot of innovation around hydrogen, for instance. I’ve seen more hydrogen companies in the last year get formed than all the prior years combined. So, the government does have a large role to play, and it will have a large role to play. And even in our more divided political system, it’s going to be a topic that does get attention, and it does get attention worldwide. Europe is leading right now in climate funding and getting initiatives through governments in Europe. So, I think that’s where… It’s going to take all capital to solve this enormous problem, and it will take time, but I believe in humanity. I’m a big believer in people and the power of ingenuity and the power of what we can do when we put our minds to it. I have a front row seat to it.

Michael Lepech: Yeah. I guess the analogy I would make is then when we were all locked in our homes, your field of genetic engineering was able to solve a global problem, because there was a global pressing need in a timeframe that we have never witnessed before.

Arvind Gupta: Exactly. And the pace of acceleration of these technologies is just increasing with AI. It is an amazing time to be alive in that sense. The pace of innovation is going to continue exceeding every year, the year prior by a lot.

Michael Lepech: It is a fascinating time, and this has been a fascinating discussion. Before we finish, we like to do a little segment on our show called Four to Fix the Planet. And so it’s a series of questions that we ask every guest. You ready?

Arvind Gupta: Okay, let’s go.

Michael Lepech: All right. What’s on your bookshelf, playlist or feed right now?

Arvind Gupta: Let me see. I’m listening to a book called “What It Takes” by Steve Schwarzman, “Lessons in the Pursuit of Excellence.”

Michael Lepech: Wow. Sounds pretty good.

Arvind Gupta: Yeah.

Michael Lepech: What’s keeping you up at night?

Arvind Gupta: So far?

Michael Lepech: Yeah, so far. What’s keeping you up at night?

Arvind Gupta: I think just making sure that we can design planetary health businesses in a way that creates resilience in this high interest rate environment, and that we’re able to find the enterprise customers and business customers for these companies that really do pay the premiums that are required to create venture returns for everyone. And I think, I wouldn’t say it’s keeping me up at night, because I’m seeing wins in that direction.

Michael Lepech: You talked about it being a great time to be alive.

Arvind Gupta: Yeah.

Michael Lepech: What’s giving you hope?

Arvind Gupta: A lot of stuff. How much innovation and how… We don’t innovate until there’s pressure to innovate. Problems don’t get solved, because they exist. Problems get solved, because there’s a reason to solve it, and it’s Darwinistic in that way, requires a pressure. And so now there is pressure to solve problems that we haven’t really bothered to try to solve. And we’re seeing new technologies that are able to solve it with great unit economics that is not philanthropies, that they’re actually businesses. And so, what gives me hope is I do believe capitalism, the greatest system for innovation humanity has ever produced, is going to be able to solve climate change and reverse climate change for humanity.

Michael Lepech: That is hopeful. What’s your favorite sustainability hack? Something that people could do to add to their day-to-say lives that you like?

Arvind Gupta: Gosh, I don’t know if I have anything super clever here. Yeah, for me, I am an athlete, and so I require high amounts of protein. I eat very well and healthy. And so being vegetarian has been hard for me in the sense that it comes with a lot of carbs in general. The protein that I get from whole legumes and things like that doesn’t have the same balance. So, for me, you probably saw it earlier, I drink these protein shakes as a way of getting the protein that I need and it’s milk-based, but that’s okay for me. And so that’s what I’m doing as my hack, I’d say, is from my own personal life, that was the way I found getting off meat to be the easiest and still maintain my competitive and fitness goals.

Michael Lepech: Very cool. One other thing that I’m wondering, because you brought it up, what’s your favorite classic?

Arvind Gupta: Oh, wow.

Michael Lepech: You spent all that time at Berkeley reading classics.

Arvind Gupta: I did. I did. That’s an interesting question. Jorge Luis Borges, I don’t know if it would be considered a classic, but he was probably one of my favorite authors. Also, Hermann Hesse influenced me a lot along with Albert Camus. The Myth of Sisyphus by Camus was probably the most influential book for me at that time. And this idea that loving toil… My takeaway from the book was if you’re rolling a ball up the hill every day, and that’s supposed to be be hell, because you never go anywhere, well, obviously it’s the absurdity of life is a metaphor for that. And so the only escape from hell is to love that toil, to love that rock, to love every rugosity, every divot, every crack in it, and to learn it. And then you’ve escaped hell and you’ve given yourself a way, a mechanism of really taking something from the toil and making it your own experience and reframing it in a positive way. And that goes a bit to Viktor Frankl’s thinking as well, Man’s Search for Meaning, another great not a classics, but I think important work. So, anyway.

Michael Lepech: Yeah. And I like that as an ending, because that actually connects well to being an entrepreneur.

Arvind Gupta: Yeah, absolutely.

Michael Lepech: You have to love the problem, not the solution.

Arvind Gupta: Yeah. You got to love the toil. I think work is… When you find what you love to do, the work is truly a vacation. It doesn’t mean it’s easy. It doesn’t mean it’s not stressful. It’s extremely stressful, because you have goals and those goals are hard, and building things and putting people together and leading, none of that’s simple. And so, yeah, I think loving it is what makes it all worthwhile, because there’s a good chance it’s not going to work anyway.

Michael Lepech: For sure. Well, thank you so much, Arvind, for taking time to talk with us today.

Arvind Gupta: It was my pleasure. I really enjoyed it. Great questions, Mike.

Michael Lepech: Today’s guest has been Arvind Gupta of Mayfield. If you enjoyed this show, be sure to subscribe to Move Fast and Fix the Planet wherever you get your podcasts, and help others find it by rating, reviewing, and sharing it. Learn more about this podcast and related work at stvp.stanford.edu/sustainability. Move Fast and Fix The Planet is hosted by me, Mike Lepech, and produced by STVP, the Stanford Engineering Entrepreneurship Center. This episode is supported by Stanford Ecopreneurship Programs. Our producers are Holly McCall and Anthony Ruth. Editing is by Stanford Video. For more podcasts, interviews, and articles, please visit stvp.stanford.edu/ecorner.

Fund/Build/Scale: Digging Your Moat: Customer Discovery + PLG for AI Startups (Transcript)

Here is the full transcript of the conversation between Fund/Build/Scale podcast host Walter Thompson and Rodrigo Liang, co-founder and CEO of SambaNova Systems:

Welcome to Fund/Build/Scale, a podcast that explores the tactics and strategies founders and investors are using to build early-stage startups. I’m your host, Walter Thompson.

Today, I’m talking to Rodrigo Liang, co-founder and CEO of SambaNova Systems.  

Launched in 2017, SambaNova has raised more than a billion dollars to create a full-stack LLM platform.

Our conversation covered a lot of ground, including his thoughts on digging a competitive moat, product-led growth strategy for AI startups, and the importance of aligning the needs of customers with the technology you’re developing.  

Before we wrapped up, I asked him to share some advice for founders who are fundraising in 2024 — and for anyone who’s thinking about joining an AI startup. More after this.

Fund/Build/Scale is sponsored by Mayfield, the early-stage venture capital firm that takes a people-first approach to helping founders build iconic companies.

The podcast is also sponsored by Securiti, pioneer of the Data Command Center, a centralized platform that enables the safe use of data and GenAI.

Walter Thompson

Thanks very much for joining me today. I appreciate it. Just a general question, what does SambaNova Systems do, and how did you initially get involved with the company? How did you initially connect with your co-founders?

Rodrigo Liang

So SambaNova, we’re an enterprise AI platform company, we’re focusing on helping companies train and deploy large AI, generative AI models, in a secure and private environment. And so when we started the company, there are two, three of us — two Stanford professors and myself — and we, we’ve known each other for years. 

And so a lot of the common history, common threads really pulled us together, as we started thinking about this next wave of technology transitions that the world was starting to experience, which is artificial intelligence, machine learning. And now Gen AI.

Walter Thompson

How long did you know them before you decided to join up and become a founding team?

Rodrigo Liang

The three of us — Kunle Olukotun is a professor at Stanford, he’s been there for 35 years, I want to say thirty-plus years. And so he and I have known each other for over 20 years now, probably 24 years now. And so it’s a long, it’s a long relationship that we’ve had, and certainly with Chris Ré, who’s the other professor at Stanford, in fact, we’ve known each other for over seven years, as well. And just being able to build that common thread of understanding and kind of how you would build companies, how you would innovate, how you deliver new products, new ideas — those are really, really important things that, you know, founders and co-founders have to share in order for you to actually have a common vision, a common destination, as the market evolves and as your company evolves.

Walter Thompson

In your mind, what sets AI product-led growth apart from traditional SaaS companies?

Rodrigo Liang

One of the really interesting things with artificial intelligence and the companies that you see today, coming into the artificial intelligence market, is that for the first time, you’re seeing products having to span a very broad market from consumer all the way into the enterprise. And so the technology for artificial intelligence is applicable broadly. And so historically, you’ve seen companies that come into the market and they do one well, but not the other. Right? You have companies like Dropbox or Zoom that you can look at, they were very focused on being able to address the rapid adoption of one of those markets. 

But with AI, what you’re seeing is this challenge of having to create something that applies broadly and you really haven’t seen that since probably companies like Apple and Microsoft started creating technologies that really crossed the entire segment of the marketplace. So AI presents an interesting opportunity and challenge and, and you also see some really interesting market dynamics around margins and things like that when people kind of release products and try to figure out well, how do you create something that allows you to continue to innovate, allows you to continue to actually build value. But you can’t just do it without a commercial model that works. So I think you’re seeing a lot of that play out on the market today. And I think you’re gonna continue to see that evolve.

Walter Thompson

So, you brought up Zoom. Zoom took off during the pandemic, because it was an easy, cheap way to stay connected. It’s a pretty classic example of PLG in action. Enterprise buyers have layers of compliance and procurement for companies to navigate. So how has PLG impacted your go-to-market strategy?

Rodrigo Liang

Zoom is a great example. Prior to the pandemic, it was competing with a number of other solutions that were at least in enterprise perceived as high security, high availability, high many things right, and so for the enterprise need, and then the situation, the commercial market showed up and the situation showed up that allowed you to kind of come in and advocate for a value proposition where some of these other issues that you were dealing with was of less priority than being able to be widely available at a low enough cost. So you can deploy uniformly across the entire enterprise, right. 

And so, so I think we’re seeing this as well, that you’re finding that artificial intelligence and Gen AI as a wave, it is challenging most companies in terms of their traditional policies, the traditional policies that they will otherwise adhere to, things about data security, data privacy, being able to use models that maybe you know, you you didn’t create there, there are a lot of things that companies have held for a long time, which are being challenged as a process because the value of artificial intelligence is so great, right, that you’re solving such a big problem that you have now have to revisit those constraints. 

Now, SambaNova, what we’ve done is, since we’re enterprise-only, really focusing on the enterprise need, and we’ve been in this market for a long, long time, we try to mitigate those, so that companies don’t have to throw out 20, 30 years of investment in security and data privacy just to take advantage of Gen AI. So what we’ve done is created a platform that allows you to get all the benefits of the best Gen AI technologies out there, but allows you to then retain a lot of the protections that you might want for data privacy, data security, and making sure that your data does not leak inadvertently into the open domain.

Walter Thompson

Can you describe your ideal customer? And how long did it take you to triangulate your ideal customer profile?

Rodrigo Liang

Well, I think for us, you know, because we knew from the very beginning where the value proposition was going to be, it was all around mission-critical applications. And this is, by the way, and we can talk about this a little bit, but I think, as we start, having a clear identity of what problem we’re trying to solve is really important. And that’s something that my co-founders and I had very early on, which is we need to create a platform that allows enterprises to be able to deploy this in a way that you can actually feel comfortable with all the policies and processes that you already invested decades into, right. And so we came in thinking about these mission-critical applications, by the fact that you have latency sensitivities you have, you know, availability sensitivities, you have scale sensitivities, you have security, privacy, these are all standard things that it doesn’t matter whether you’re selling hardware or software, right doesn’t matter whether you’re selling SaaS or services, it doesn’t matter what product you’re selling into those environments, they all remain true, right? 

They have to be scalable, they have to be available, they have to be mission-critical, they have to be secure, they have to be data private. And so we came in with a very, very strong thesis around the fact that artificial intelligence will be the dominant technology for the next 10-15 years for enterprises. And you have to solve these basic tenets of what enterprises care about. And so when we started, we effectively just focused on customers for whom those were the most true, right? That some enterprises, many enterprises will say that those things are important, but there are some enterprises, imagine a highly regulated one, you know, highly-regulated companies or highly-regulated organizations, governments and things like that, where that’s most true. That’s where we started. We started with places where, you know, our customer need was the strongest to allow us to actually be able to articulate and demonstrate the value proposition that we’re bringing.

Walter Thompson

So break it down for me a bit, your customer discovery process, how long —  just roughly ballpark it? How long did that take before you were certain that you had enough information from potential customers to validate the thesis you’re talking about?

Rodrigo Liang

I think, you know, for us, and I’ll say that this is non-standard, because many things about artificial intelligence and this cycle of, of technology transition has been non-standard, right from, you know, companies like OpenAI getting 100 million users in two months to, you know, to like the new technology companies getting funded, like we raised a billion over a billion dollars in cash through the first three years of the company, there are a lot of non-standard things that many of your audience who may be have, have been in the startup ecosystem for a long time, or they’ve been in technology for a long time, they know that those aren’t things that you you perhaps witness every day, over, over a long period of time. 

But in that construct, I would say that we had two classes of customer: there’s some customers that had been evaluating artificial intelligence on their own, and when we showed up, it was just so clear that they needed something, they weren’t entirely clear exactly where were all the perfect places to apply. But they were going to invest very, very aggressively. And so in those cases, which we have many customers like that, I mean, it was just a matter of a few months? 

There’s kind of traditional again, this enterprise conversation, you know, just being able to demonstrate what you do, for them to understand the capabilities, bring the technical people together, so that’s frequently the way that these kind of very transformational buyers think about it: “I know I need it, I don’t have to figure specifically all the places that I’m going to use it and how to calculate the ROI and all these different individual cases, because it’s a platform I need, and I’m going to need to what’s more important for me is to get going because it’s a competitive advantage for me, right?” And that’s a big class of our customers. 

There’s another wave of customers that’s basically looking at this as, “okay, how does your technology differentiate from the other offerings out there? You’ve got OpenAI and ChatGPT doing some stuff, and Google does some stuff, all these different people do their various different things? How does your technology fit into my overall hybrid solution that I wanted to plug with various different things?”

And in those cases, there’s a little bit more evaluation: “Okay, where would you fit? And if you fit in places where I have a couple options, what is your differentiation? How do I evaluate the choices?” And it’s a bit more granular in the way that they’re deciding: “Do I use A or B for particular tasks, maybe if you’re not good for B, I can use it between C and D?” So there’s, there’s a little more than that’s a little bit more like a SaaS cycle, traditional, SaaS buying cycle, but what you get out of it are customers that have a very clear idea on the value proposition to your bringing. They have a very clear understanding of how your technology compares to everybody else. 

And you’re able to then lean on that to then create replicable use cases within those companies. You can say, “look, use case A looks like this, your successful use case B also looks like that, so you can be confident that we’ll be successful.” And that’s really important to be able to actually build value for customers quickly and build their own confidence that using your technology, they can get value.

Walter Thompson

When did you decide to bring on a full-time marketing person in that customer discovery journey? Was it kind of after the process you just described? Or before? Obviously money wasn’t an issue as far as bringing on marketing support, So when did that become something you pulled the trigger on?

Rodrigo Liang

Well, I think all founders will know that a big part of your job from day one is marketing. Yeah. Because all you have is some slides, right? And you have to be able to clearly articulate your value proposition to investors Day 1 and it bleeds right into articulating your value proposition to the customer. And so I think so I think, as your audience is listening to this, and probably resonates with many other founders that they have to play the role, and depending on the capabilities of the founding team. If you have it, then you don’t need to bring a dedicated person right away, because one of the people, somebody has to step into that role very, very early.

And if you don’t have somebody, which sometimes you have startups done by folks that aren’t as comfortable with the whole explaining and articulating the ideas and and the concepts in the value proposition in a way that the non-technical folks or the people that aren’t in your industry can understand — which is ultimately the job of marketing. How do you take this very domain-specific knowledge and translate it into something that is actually understandable by a much broader audience. If you don’t have that, then you have to bring it on fairly early because a big part of startups is to articulate, “why this new thing?” There’s this new thing, which is, by definition, the charge of a startup — you’re creating new things. “Why is this new thing so much better than whatever else is already out there. And whatever else I already know?” 

And so having somebody that can articulate that is really important, we just happen to have I mean, we have got two incredible co-founders, professors that are so well-suited in being able to explain the technology side in a way that most people can understand. And then as part of our founding team, we had lots of people that came in with some background and being able to help articulate some of the more complex ideas around machine learning and artificial intelligence to a fairly broad base of customers, some very, very technical and some a little less technical and be able to then still convey the value proposition that we brought.

Walter Thompson

I read an interview where you said that SambaNova had “transitioned to a much more internally-led, and probably now even more customer-led, and customer-centric culture.” What led you to make that shift? Or was that always baked into the company’s recipe?

Rodrigo Liang

I think it’s always been part of the culture of the company, and harder to do early on, when you don’t have clear customers that have a clear view of exactly where they want to go with technology. And so, so I would say, in the first few years, again, we build our own chips. So we don’t use any NVIDIA chips, we build our own chips, we build all the way up to the foundation models, these trillion-parameter models that you see with GPT-4 and that class. And so for us, when we’re building that new one, we’re in that first few years of the company, when you’re building the investment stack, there’s nothing like it on the market. One, there’s nothing like it in the market, nobody knows exactly how to kind of get all that in an integrated fashion. And then, two, the people that see the value, they aren’t entirely understanding how that would fit into the rest of the ecosystem that is evolving. 

You have to have a conviction around how you see your product fitting into the broader scheme of the market. As we started getting customers 1, 2, 3, by the time you get customer 20, 25, 30, now you’re starting to get a very clear point of view of exactly how the customer is going to use it. And by the way, most customers are going hybrid with artificial intelligence, nobody wants to be single-source in a single vendor, everybody wants the latitude and freedom to be able to choose who they want over a period of time they want, they don’t want lock-in. So we come in with the idea that we’ve got to win every use case with every customer and allow the customer to choose us. 

Ultimately, if you start with that belief, that every deal every year, every month, every year that you go out and talk to customers, you have to win the customer again, that drives the behavior in the company of you’ve got to serve the customer’s need, versus the example of if the customer is locked in, there’s no choice. The customer’s need is maybe a little bit less important to the developers, because whatever you build, they have to buy.

Walter Thompson

It’s interesting. I mean, are you ever worried about taking on like a one-off problem, so to speak, like a problem that doesn’t really have a broad scale or scope. You’re creating a new feature for one customer that might not really scale or pay off down the road?

Rodrigo Liang

Yeah, every day, every day. We have meetings all the time about this and this is one of the hardest things, especially in a market that’s still evolving. If you look at Gen AI in the market that’s still evolving, you’re trying to figure out well, “customer A has these three classes of problems, it’s trying to solve it all through one supplier.” Maybe the right way they need to solve this is, “supplier A does problem one, supplier B does problem two,” but they haven’t arrived at that conclusion yet. Of course, as a new company, you always want to get as much of that business as you can, but what’s important is to be able to deliver that sustainably. What you don’t want to do is, say you can do it when you cannot do it. 

Or you can do it, but then it’s such a big push and such a journey outside what you’re good at that it takes up all your resources delivering that. And sometimes what you end up having to do is just really break the problem down to, what is the customer ultimately trying to do? When the customer says, “Hey, can you create this model for me?” And if it’s a model that we don’t necessarily want to create or we don’t understand why they want to create, be understanding what are you trying to do? “Oh, I’m actually trying to do this task and I want Gen AI to actually produce this result.” “Oh, okay. How about this model that we do do?” 

And so there are ways that you can actually really understand the ultimate problem that customers have, and try to potentially align it better with what you’re already doing, because two things happen in that case: one, your ability to deliver in a higher quality is much better because all your people are working on a focused set of things, right? You’re just gonna hit the dates better, your quality is gonna be better. 

But the second thing is, you’re going to be able to then leverage that technology into the field and actually get more customers using and robustifying on your behalf and expanding the feature set on your behalf so that all your clients can benefit from that experience.

Then all your clients come in and say, “oh, it would be great if you also put in these five features.” And so you’re getting that collective feedback on a single product, versus getting all these, you know, distributed feedback on products that aren’t shared between clients. And so then everybody ends up having a subset of the feature set. 

Even if you are really, really good at mapping the needs to kind of what is the main products that you have, they’ll always be cases where a very, very important customer or very, very critical need of a particular customer who creates a requirement that’s a one-off and those are individual decisions, you always have to make. It doesn’t matter if you are your three-year-old startup, or a 30-year old company, when I was at one of the largest companies on this planet, we were still making those decisions of special customers asking for this one thing, should we do it? And so I think that that scenario never goes away.

We’ll be right back after this word from our sponsors.

Are you thinking about launching an AI startup?

Mayfield’s $250 million AI Start seed fund is actively searching for idea-stage entrepreneurs who are working at the cognitive plumbing layer. That’s models, middleware and tools, data, infrastructure, and semiconductors/systems.

Mayfield has a long track record. Since its founding, the firm has been an early investor in more than 550 companies, which has led to 120 IPOs and over 225 mergers and acquisitions.

If you have a fundable idea for an AI-first startup, email aistart@mayfield.com

Every business that incorporates AI has to account for data privacy, compliance and governance. 

That’s why Securiti developed its Data Command Center. Instead of setting up different tools that can eat up time, money, and add complexity, Securiti’s award-winning technology: automates compliance with global privacy regulations; includes a library with more than a thousand integrations across data systems, and identifies data risk to enable protection and control. To learn more, visit Securiti.ai.

Now, back to my conversation with Rodrigo Liang.

Walter Thompson

A B2B SaaS startup might share their product roadmap with an early customer to get buy-in. You’re one of the best-funded AI startups in the game. Is that something you can afford to do? How transparent are you with customers as far as, “here’s what we’re thinking of, or here’s what we’re working on.”

Rodrigo Liang

Again, our stack includes chips through servers, through a variety of things around what we call the AI operating system, right? If you think about how these models all have to interact with each other, and interact securely, and privately, all those things will kind of generally name it in the category of AI OS, just being able to have many, many models, operating concurrently on the platform. So if you think about all those features, that to us, our unique advantages, we tend to be able to give them directionally what we’re doing, but not the specific features, just because these are things that competitively are important for us to make sure that we’re able to drive differentiation and ultimately deliver as well. 

As you get up into the use cases, we’re actually quite transparent. Today, we build our trillion-parameter mall models off oftll the most popular open source and we tell them “okay, we’re using Llama 2, we use Mistral, these are all the types of models we use. And the next wave, we’re going to use these models, these models, these models, these are the configurations of the models, these are the sequence length, these are the batch sizes, the vocabulary sizes, we’re very transparent about what we do as far as kind of how we do these large, large models, and then how those models then are then translated into AI business tasks that you can take advantage of.

Because you’re working on business features and tasks that you don’t care about, then what am I doing? So we’re very, very direct about saying, “look, here’s your call center support, here’s the code Gen support. Here’s  your legal assistant, here’s summarization, here’s your named entity extraction.” It will give you all the very specific things that we’re supporting, so that they can come back and tell them “Yeah, I don’t know, I’m not using those things, but I care about that.” And so we’re able to then, using feedback, adjust what we support more of, because it is all about making sure that the customers when they get a platform, they get ultimately what they need to accelerate delivering value into the corporation.

Walter Thompson

Interesting. What challenges are you seeing with regard to AI adoption, and how would you advise a smaller startup that’s trying to get buy-in with C level executives? My sense is this is something you do over time by building relationships, but how do you get that ball rolling initially?

Rodrigo Liang

A lot of it is a function of the timing of the market as well. When we started the company back in 2017, there wasn’t a very great understanding in the commercial enterprise of what AI could do for us. It took ChatGPT showing up for a lot of people to really, for the first time in a very hands-on way, get a sense of, “oh, I see what Gen AI is all about.” And so I think it depends on the market. Now, as you kind of move forward from here, I think there’s a lot early education that’s taken place, and I think a lot of the enterprises are starting to realize, “okay, in the flow of work for for AI, there are things that I can get from these vendors and the suppliers that exist, you know, SambaNova or NVIDIA or OpenAI, there are places that are delivering these technologies into the enterprise. And there are things that I’m not exactly sure how I could solve. And I think that customers are starting to get that level of insight into what can I do for them?

I think that having that very, very active conversation with customers, a broad range of customers, they don’t even have to be customers today. But our mentality is one where everybody we talk to will one day be our customer. Right? It might be tomorrow, it might be five years from now, but one day, right? And so if the mentality is one where you’re trying to find out what are they thinking about and how could we potentially solve it? I think it actually starts opening a lot of doors for people to say, “okay, well do we have something to my favorite problem and a gap, something that might exist?” So that’s kind of a very just — is there a product-market fit with a need that you might have? 

I think the other aspect for new companies that are trying to enter the market and trends we talk about is, look: another thing that’s interesting about AI is the cost of actually validating your technology now is extremely high, extremely high, right, you look at the cost of running these GPU cycles on the cloud. I mean, a lot of companies are trying to offset that by raising hundreds of millions of dollars in venture funding, right, but not everybody can do that. And so you kind of have to think about effective ways that maybe you can partner with folks, partner with other companies to leverage kind of the relationships that they might have and leverage some of the technology and tools they might have to kind of prove out some of the things that otherwise you have to do all on your own.

We’ve been very open about the partnerships we have with companies like Accenture, with the US government, with other startups, that is a very, very important way that you can actually engage your technology with others and in some ways actually get fast feedback on how well that technology actually solves a particular type of problem. Those types of interactions that allow you to shorten the cycle, allow you to actually prove it faster, allow you to actually make sure that you connect to a problem that people are actually struggling to figure out how to solve, right, those are very, very important before we even get to that commercial conversation that we have to find a procurement person inside that company. Solving a problem is much more expensive these days than it was before, just because of the cost of the computation. But still, the process of solving the problem and process of making sure that you have value in your tech, finding a very efficient way to do that before you enter the procurement process is really, really important.

Walter Thompson

For product-led growth to work, you really need to build consensus and alignment within the entire organization. A lot of companies don’t have that much in-house expertise. Has that been a challenge for you? Or is or is just kind of, “game on, as far as we want generative AI? How can we get it?”

Rodrigo Liang

No, it’s absolutely true. I think intersecting the customer will, wherever they are in their journey, obviously, of companies that are incredibly savvy in AI, they’ve been doing artificial intelligence for a long, long time, they have thousands of people in there. And so you have companies like that. And there’s some companies that are just getting started in their journey, they maybe just hired the first couple of people to really look at it. 

And so understanding where customers are under journey, figure out kind of what you can offer for each of those types of customers at those stages. SambaNova, because we’re a full-stack company, we can go from chips all the way to models, the reason we took it all the way to the highest level of abstraction was we wanted enterprise customers to have an interface that looked like ChatGPT. If you look like ChatGPT for the open market, I can give you the private version of that. It takes a lot of investment to take your technology stack all the way up to that level of abstraction. But what it did for us was, it opened up a whole class of customers that otherwise if we’re down below where we’re actually teaching you how to program our chip, there was no chance that they would actually have enough expertise to be able to do it. 

Now certainly, we have a large number of customers with that level of staffing for artificial intelligence that are able to come in and say “I want to create my own model on your hardware, and I’m going to just do it ourselves.” Certainly we have loads, but being able to then manage out your customer base and be able to map the offering you have to their lifecycle into where they are in the journey itself, I think is really, really important to actually be successful in this market.

Walter Thompson

How does your PLG approach enhance the customer experience in a way that encourages people to become advocates or evangelists?

Rodrigo Liang

We really are focused on getting quick proofs of success.So we have these things that we show a customer called previews and what the previews are is basically, if you think about use cases for AI, a lot of people look at, “oh, I see this model on the web — GPT this, GPT-4, GPT-3.5, I see Mistral.” Those are not use cases for the customer. Those are amazing models that ultimately integrated into a workflow that will deliver end benefit to the customer. And so what we’ve done is, we’ve created these previews, and these are business outcome previews, “how do I take the technology that we have, but instead of showcasing how much better technology is compared to somebody else’s technology, let me show you how this technology feeds into a solution that is important to your business, right? How do I create value for you and deliver something you can’t get from anywhere else to give you a sense of how quickly you can take this technology stack and actually create benefits into your business?”

And that’s an incredibly important way to actually create acceleration, because otherwise, you leave it to your customer to figure out how the heck do I actually show value? And we want to actually make sure that the customers can show value very, very quickly. 

Walter Thompson

Last few minutes of our interview here, but if you could share some advice for people who are starting  out  or  who  are thinking about starting out with an AI startup. You mentioned, pretty clearly,  SambaBova has raised more than a billion dollars in its first three years. Do you have any  general  advice for early-stage AI founders  who  are  trying  to  raise  a  seed round right now in  2024?  

Rodrigo Liang

Two things: it’s much better to be raising now than it was last year, right? So your timing is much better, I would say, the understanding in the market and investor community around what AI can do, and what all the possible things that add value, I think is significantly higher than it was say, three, four years ago. And so I think be very, very clear about the value proposition that you bring, just having a tech that does X, Y, Z. Imagine these investors are seeing this probably, you know, 20, 30, 40 times a day and so there are a lot of ideas out there, and many of them aren’t going to be able to actually deliver value into the customer base. 

Think about something that is of significant value and think of something that is of significant value over a protracted duration of time, not as only valuable today, but you can see how it accretes value over time and allows investors to actually buy you time so that you can actually deliver more and more traction in the marketplace. Right. And so those are the two important things that I can think of.

Walter Thompson

I think a number of my listeners are likely people who have an engineering or an academic background, if they’re, they’re thinking about starting up. But becoming a CEO kind of takes an entrepreneurial mindset. So was this a challenge for you way back when you did your first startup, and do you have any advice for helping someone overcome this hurdle? It’s just a mental hurdle, but it’s real. 

Rodrigo Liang

Some people will say this is still a challenge today. Every day is a challenge, you know, the role of and it’s not just a CEO, but the role of leaders in a startup. The startup needs what they need from you changes year after year, and you have to sit down and figure out what it is that the company needs from you. 

That was never my case, because we just had such an amazing technical team that could deliver on the technology that I didn’t have to be that person. But for many of your audience, you might be, you might be the technical guru for the company. And so there’s a stage where what the company needs you you figure out is to solve that hard problem, right, solve that really hard problem. And many, many startups in the early stages are formed because you have an incredibly amazing technical solution driven by your own capabilities. And as we grow, then the company’s needs start changing, because the hard problems have been solved. Now, it’s all about scaling, robustifying, creating all sorts of other things. And yet, what the company needs next for you to take this incredibly sophisticated, amazing technology and explain it to somebody, to a layman in a way that they can understand value. 

Now, suddenly, you became a marketing person, right? And so that’s it: the needs of the company change, change what you have to do very, very quickly, and our ability to go from fundraising, to explaining the value proposition to customers, to helping customers actually incorporate into your change management, all of those things, those end up adding skills that we sometimes have from before, but oftentimes did not. And so I would say, for entrepreneurs, if you want to become an entrepreneur, lead a startup, be very, very adaptive, be very open to learning new skills, that you’re not going to do them perfectly. But the company needs you to do some of it.

Walter Thompson

How would you advise someone who’s looking for a job with an AI startup? What should their due diligence process look like and what are some questions that you would ask during the interview process?

If you’re sitting across the table, like you want a job with an AI startup, what are some of the questions you’d be asking them to kind of figure out like, you know, “is this a job worth taking? Is this a bet I want I want to make?”

Rodrigo Liang

Yeah, okay. Yeah. So same, same question: “why did you decide to tackle this problem?”Try to try to align that with what you’re interested in? Why, why did they do it? Right. And the [answer] is, “I wanted to just make some money,” that’s not a great answer, right? There’s lots of ways to actually create companies that add value. And can you can do that, and frankly, you know, startups. You go in for many, many other reasons than that. So find something where your passions, your interests, your wellness, technical, or often a few perspectives match what they want to do, right? I think that that is really important, because near the journey of a startup is a lot of ups and downs, and incredibly exciting, incredibly exciting, but you got to the same page as the mission of the company. 

I think beyond that, the very, very next question that I would ask is, “why is it long-term viable? Why is this technology, whatever you’re building, long-term viable.” It’s an incredible competitive environment. The environment that companies big and small are tackling, it is not just the small players. I sometimes use the phrase, it’s a stampede of the largest creatures in the world. And you know, if you’re sort of in the middle of it, you’re scurrying around, right? And so you have to understand it’s a competitive environment. So you have to believe that there’s a moat in the technology that allows you to be able to differentiate, and differentiate not today, but for a long, long time. 

And then finally, I mean, with startups, you’ve got to have money, right? It’s a very expensive endeavor, especially today with the costs of actual building these things. So “what is your cash position, and how do you intend to fund your venture all the way until you’re profitable?” These are somewhat pragmatic questions, but it’s a reality that it’s a competitive environment and having cash is going to be very important.

Walter Thompson

So if I’m interviewing at an AI startup today in 2024, how many months of runway do you think they should have for me to say “yes, I’m willing to get on board a year, year and a half, two years.”

Rodrigo Liang

The way I think about that is if you’ve got to match it to the milestones that you have, and different companies have different strategies, and it’s probably a subject for a whole other podcast around how fundraising works. But I always advise folks that you don’t have to raise funds right away, but it’s always better to have more than enough, right? That’s step one, right? It’s always better to have more than enough. But the second one is to make sure that you have a clear idea of what those milestones are that you think will drive value. Right? 

If the next big milestone for your company is three months away, well, you should have money for six months. If that next milestone is a year away, maybe you need significantly more than a year, you know, I mean, like, you want to be able to actually have cash until that big milestone because those milestones drive confidence in the investor base that your technology is actually producing value. But these things are imperfect, right? The market is changing, you might have some surprises, and so you want to make sure that the cash covers the next big step in your milestone — step function milestone — in order for you to get to that.

The worst thing would be — this is kind of the tragedy I see sometimes — the worst thing is when you run out of money just before that big milestone, right. So you never want to be in that situation. You always want to be in a situation where the cash covers you past that major milestone, and make sure that when you raise, you raise enough money to cover the next milestone after that,

Walter Thompson

Rodrigo, thank you very much for your time. This has been a great conversation. I appreciate it.

Rodrigo Liang

Thanks for having me.

Walter Thompson

I’ll be right back with some show notes after a word from my sponsors.

Fund/Build/Scale is sponsored by Mayfield. If you have a fundable idea for an AI-first startup at the cognitive plumbing layer, email aistart@mayfield.com.

The podcast is also sponsored by Securiti, pioneer of the Data Command Center, a centralized platform that enables the safe use of data and GenAI. To learn more, visit Securiti.ai

Thanks again to my guest, Rodrigo Liang of SambaNova, for appearing on the podcast.

Coming up next: an interview with Ozzy Johnson, Director of Solutions Engineering at NVIDIA.

Ozzy and I talked about where early-stage AI founders need the most help, how people from academic/research backgrounds can more easily shift to an entrepreneurial mindset and we also spent some time exploring strategies for nurturing a successful AI developer community.

He also shared some thoughts about balancing initial spending with the need to drive early growth and which traits successful AI founders have in common. (Spoiler: they aren’t all developers.)

If you’ve listened this far, I hope you got something out of the conversation. Subscribe to Fund/Build/Scale so you’ll automatically get future episodes — and consider leaving a review.

For now, you can find the Fund/Build/Scale newsletter on Substack.

The Fund/Build/Scale theme was written and performed by Michael Tritter and Carlos Chairez. Michael also edited the podcast and provided additional music.

Thanks very much for listening.

Partnering with Highway 9 Networks on Mobile Cloud for Enterprise AI

Green highway 9 and mayfield logosWith the news of their $25 million inception round led by us, I am thrilled to welcome serial entrepreneur Allwyn Sequeira and the Highway 9 Networks founding team of Debashis Basak, Serge Maskalik, and Sachin Thakkar to the Mayfield family. Together, they have built several billion-dollar businesses and created multiple markets while at VMware, including software-defined networking (SDN) and private/hybrid clouds. As the first Mobile Cloud for the AI Enterprise, Highway 9 is an example of the infrastructure layer of Mayfield’s Cognitive Plumbing investment strategy. 

AI is increasingly powering massive transformation initiatives in the enterprise—like robotic manufacturing lines, drone-based surveillance, distributed real-time business analysis and more. Use of AI on distributed devices requires always-on, high bandwidth and low latency wireless networks. As wireless networks become higher throughput, the physics of wireless data transmission makes these networks increasingly prone to interference. While a dropped cell phone call causes annoyance or inconvenience, the unreliability of a mobile connection has much higher consequences for AI workloads. 

A leading automaker and Highway 9 customer found that the autonomous robots on their manufacturing line would stall when they lost connectivity. These stalls would stop or slow down the assembly line impacting production and create safety risks. After extensive research and testing for multiple WiFi and cellular solutions, they found that Highway 9 and its Mobile Cloud was the only solution that could provide the highly reliable connectivity required for that autonomous manufacturing. At the same factory, their employees and machines needed an integrated indoor + outdoor network for effective logistics and inventory control, and once again the Highway 9 Mobile Cloud was the only solution able to provide a private mobile network that allowed devices to move between customers’ existing WiFi and cellular networks seamlessly without even changing the IP address. 

Highway 9 addresses enterprises’ core AI and automation (aka the “AI enterprise”) requirements through a unique cloud-based mobility platform. The Highway 9 Mobile Cloud revolutionizes the deployment and operation of private and hybrid mobile networks through a true cloud-native, integrated set of virtualized mobile services that work with a wide range of mobile network devices. Highway 9 also seamlessly integrates these devices with all the existing enterprise wireless and wired networks and security solutions. Highway 9’s product is sold to enterprise IT infrastructure teams and allows enterprise IT to have full control of these cellular networks in an operating model that works for enterprise IT.

Enterprise customers will also be able to use Highway 9 Mobile Cloud to extend the signals of the public cellular networks on their campuses without having to install expensive dedicated physical infrastructure for each carrier such as Distributed Antenna Systems (DAS) that does not integrate with enterprise security and networking. This will make bad indoor cellular coverage and indoor dropped calls a thing of the past. 

Mobile networks have been designed for higher security, lower power, longer range and high quality of service, but they have been hard to use as private networks. There is a once-in-a-generation convergence of developments that makes this easy now. In 2020, the FCC authorized the full use of Citizen Broadband Radio Service (CBRS) spectrum to be used for cellular networks without having to obtain spectrum licenses. Before this, any private cellular service required an enterprise to partner with a spectrum licensee which was typically a government entity or a telco. Mobile networks used to require physical SIM cards but eSIMs are now pervasive. In fact, by 2025, there will be 3.5 billion eSIM enabled mobile devices such as laptops, phones, tablets, watches and other mobile devices. Windows, Android and iOS have added native support for private cellular networks which makes it very easy for enterprise mobile and IOT device management solutions to easily configure private cellular networks and prioritize them for mission critical workloads. 

Being a People First investor, it is our privilege to partner with this seasoned team to address the huge mobility needs of the AI enterprise. We are looking forward to the next chapter of their journey. 

Introducing Fund/Build/Scale

Fund/Build/Scale — a new podcast series co-created by Walter Thompson and Mayfield — will offer listeners a holistic look into the world of AI startups.

In Season 1, host Walter Thompson (TechCrunch, San Francisco Magazine, Hoodline) and guests will delve into the strategies, challenges and tactics shaping the future of artificial intelligence and entrepreneurship.

Listen now on Spotify or Apple Podcasts.

 

 

 

 

Fund/Build/Scale: Navin Chaddha on Building Trust with Your Inception Investor (Transcript)

Here is the full transcript of the conversation between Fund/Build/Scale podcast host Walter Thompson and Managing Partner Navin Chaddha:

Walter Thompson  00:01

Welcome to Fund/Build/Scale. I’m Walter Thompson. 

I worked in tech before becoming a journalist. Now I’m talking to a diverse range of established and emerging founders and investors to learn more about the tactics, strategies and frameworks they’re using to create early stage startups. We’re discussing challenges like funding a co-founder, developing a value proposition, fundraising, going to market, and other essential topics for tech entrepreneurs. 

In the first 10 episodes, I’m focusing on AI, mainly generative AI startups. I’ll be right back after the short break. Thanks for listening. 

Fund/Build/Scale is sponsored by Mayfield, the early-stage venture capital firm that takes a people first approach to helping founders build iconic companies. The podcast is also sponsored by Securiti pioneer of the Data Command Center, a centralized platform that enables the safe use of data and Gen AI. 

I’m here today talking to Naveen Chaddha, managing partner of Mayfield Fund. Thanks very much for making the time today. I appreciate it.

Navin Chaddha  01:07

Absolutely. Walter, it’s a pleasure to be with you. 

Walter Thompson  01:12

And thank you also for sponsoring this podcast, it’s been a real opportunity for me, and I’ve learned a lot and hopefully, the audience is going to learn with me.

Navin Chaddha  01:17

Thank you, it’s a delight to be amplifying the work you want to do.

Walter Thompson  01:22

So my first question, why did you launch the AI Start seed fund? And what are your long-term goals? If you’re successful, how would the world look different in 10 years or so?.

Navin Chaddha  01:30

Absolutely. So I think first and foremost, as everybody realizes and is saying, AI is a tectonic shift in the technology industry. It’s going to change the way we as humans work, live and play. And then it will go beyond humans to go understand and do things for nature, if you will. So why are we bullish on AI? That’s the reason. And our primary belief is AI plus humans is going to be “human squared.” 

The purpose of our seed fund is in our history of investing over the past 50 years, we have primarily been an inception-stage investor. Ten years back, inception-stage investing used to be Series A, then it became mango seeds. I don’t know what it is, but our focus is very simple: how do we partner with entrepreneurs at the inception stage, sometimes it will be called pre-seed. Sometimes it will be called the seed, sometimes a mango seed, sometimes a watermelon, sometimes a Series A. We want to just partner with entrepreneurs and help them build great companies, from their inception to iconic. And in order to do that, we have created a dedicated vehicle where we can start with smaller checks, which could vary from two to three million, four five million because in the past, Series As have been 8,10, 12, 14. So we want to meet the entrepreneurs where they are, and hopefully with what Mayfield has done, other established venture firms do that too. So the pie becomes bigger. We want to just foster innovation.

Walter Thompson  03:18

Mayfield talks about having a people-first framework: what does that mean in terms of evaluating new opportunities, talking to founders, how does that manifest itself?

Navin Chaddha  03:27

Yeah, so I think it’s fairly simple. Again, we have a point of view. And it could be different, or the same as others, our point of view is focused on human-centered AI, which elevates humans rather than displaces them. And for building those kinds of companies. First and foremost, we believe mission and values are critical. You need to think of doing good for the world at large, besides making money. Yes, it’s a capitalistic world, we’ll all make money. And in order to do that, you need to have a very strong mission and values. You need to be GenAI-native, you need to think from day one about trust and safety. And also keep in mind how you can take care of data privacy and security, and then focus on figuring out, can you really elevate humans and make them better? And that’s what our view on what a people-first framework for AI means.

Walter Thompson  04:36

So how far along does someone need to be in their journey as an entrepreneur before they can pitch Mayfield AI Start seed fund? What do they need to show you as far as validation, or social proof?

Navin Chaddha  04:47

Yeah, I think it’s very simple. As I mentioned earlier, we’re looking for entrepreneurs at the inception stage. Ideas can be half-baked. We have helped build companies over the last 50 years, where they were at the napkin stage with paper-and-pencil ideas. And we want to help them, partner with them and figure out what their journey looks like. So my call to action for the entrepreneurs, who are going to be listening to your podcast is jump in, call us, find how to reach us. And we are here to help evaluate what you’re doing. And if the stars align: you like us, we like you. Let’s go.

Walter Thompson  05:32

In your mind, what does a fundable idea look like — the rough shape of it?

Navin Chaddha  05:37

So what I would say is, whether it’s AI, or non-AI, it starts with is the product you’re trying to build? A must have? Or a nice to, for a certain segment of the market? Because our belief is, painkillers sell, vitamins don’t. Everybody’s busy. You need to rise above the noise. Second, is this a niche opportunity where we just have one hundred customers in the world who need this? A thousand? Ten thousand? A hundred thousand  Millions? Let’s get on the same page on what it is. 

Asking for what your revenue potential is, what’s the size of the market? That’s too early at the inception stage, right? So is your product a painkiller? How many people in the world need it? And what’s [it] going to be [like] if everything works? What’s going to be your unfair advantage, or moat, which allows you to win? It could be IP, it could be people, it could be knowledge, it could be a business process. It could be a business model. It could be pricing, let’s figure it out. Every company, a lot of companies will have the same idea. But you need some defensible moat. And then, are you willing to put the company first and surround yourself with excellence? Then let’s go, let’s go. So that’s what I would say. It’s not like, “you need to do this, you need to be this.” it doesn’t matter. But those are some of the tenets we look for. 

Walter Thompson  07:11

That’s interesting, because at this point in the hype cycle, I feel like a lot of people wonder how deep a moat they need to have dug before they can talk to an investor. It sounds like you’re saying the idea is key, but the founders still need to have dug a somewhat defensible moat before they approach you. Is that fair to say?

Navin Chaddha  07:27

Or, let’s figure it out around the way. I want to just make sure the problem you’re solving is a painkiller for somebody, and that somebody is not ten customers, it’s potentially many, many. And then if everything goes well, you don’t need to have figured it out. What do you think is going to be the moat? I’m just looking for a bet you want us to make, clearly, since you’re a paper-and-pencil idea, you’re not going to know what it is. Let’s engage and figure out.

Walter Thompson  07:55

This leads to my next question: how large is a typical seed round these days?

Navin Chaddha  08:00

It’s all over the place. There are even before the seeds, they are like pre-seeds then there are seeds, which could start at two, three, four million. And there are mango seeds, which have gone up four, five, six, eight million, but I’m seeing round seed rounds from 8-12 million now. And then, if you are “hard team to beat” entrepreneurs, the number is 20. The number is 30. The number is 40 million, 50 million. So I wouldn’t say these are seed-stage companies in the traditional sense that we have been using for the last five to seven years, it’s really inception-stage companies, what capital they need to be able to get to the next milestone, some of them will be capital-light, they need little money to get to the next phase. Some of them will require a lot of capital. 

So I would say seed should be looked broadly, as it’s an inception-stage company, what’s the right amount of money? Right? Let me give you an example. If you’re taking a short flight from San Francisco, to LA, you need a plane which can hold a certain amount of fuel. You go from here to Denver fuel increases. You go to Texas, we need more. You go to Chicago, a little bit more, then you go to New York even more. Right? Because if that’s your destination, you need to make sure at least you have capital, which is the equivalent of the fuel. So there is no right answer.

Walter Thompson  09:30

Can you speak a little bit to specifically what kinds of opportunities you are looking for in 2024? And are there specific subfields or industries you’re interested in or not interested in?

Navin Chaddha  09:39

Absolutely. So I think having been in business myself for 20 years as a VC, and another 10 years as a startup entrepreneur before that, I think there are two approaches. One approach, that’s the pertinent question you’re having, is to have a prepared mind: where is the puck going? Where are the opportunities? But unfortunately, the mega-mega companies, most of the time, you need to just have an open mind. Because nobody knows how greatness gets created, right. And then many examples in consumer and in business, you just don’t know. So I would say, you need to be nimble, have a prepared mind. But keep your vision not tunnel and have an open mind. 

As far as investing is concerned on the prepared mind-side, what I would say is, in particular, in enterprise and AI-enabled enterprise, we are looking at opportunities up and down the tech stack. So it starts from semiconductors, moves up to the infrastructure layer, which is as a service now, then goes to the data layer, then middleware and tools, and then applications, which are built on top of it. And the bottom four layers Mayfield is calling “cognitive plumbing.” 

And the application layer for AI will enable itself as applications, services, co-pilots, autopilots, agents — you just don’t even know what the possibility is going to be. So I think huge opportunities like there have been in the tech industry, similar to the PC and client server era, leading to the internet, leading to mobile, leading to cloud. This is just a continuum, the whole stack is being reimagined. And that track is going to enable net new things which weren’t even possible before. So that’s the way one has to look at it.

Walter Thompson  11:59

Any thoughts on how it might impact I guess what we think of typically as consumer internet or consumer products?

Navin Chaddha  12:04

Yeah, absolutely. So I think the first phase is this cognitive plumbing layer needs to get built. And OpenAI is a good example. What it’s showing is, you use AI at the back-end, you build cognition, you build reasoning, you build context, but the way you interact with your product and services, human language, it’s natural language. I think we should assume that interface is going to redefine every consumer, every business application, and humans will interact with machines and machines will interact with humans in human language. So no longer do you have to script, then there was a graphical user interface, then it was mobile, this is the new new thing. So it is a combination of back-end cognition. But the front end is machines that understand human language. So humans will converse with machines, as if it’s another human, and machines will converse back to us in the language we understand. And then as a result, what will happen is AI is really going to be my teammate, it could be my coach, it could be my genie. It doesn’t matter, right? Like, it’s gonna be a companion with you forever, either you support it, or it supports you, remains to be seen. 

I think it’s going to just disrupt the way we interact with computers with mobile phones. So it’s very, very exciting for entrepreneurs. But what it will be, this is where you need to have an open mind, open mind. And just imagine what could be done and don’t look back in the rearview mirror. That’s the wrong thing. If this platform is enabled and is available, what can you do? Don’t just think incremental in the consumer space, think bigger. Imagine the world, make it what you want it to be. 

Walter Thompson

Exciting. 

Navin Chaddha

Yeah, very exciting. 

Walter Thompson  14:10

I mean, this is very different from B2B SaaS. So what are some of the success metrics you’re looking for a year after that initial inception investment?

Navin Chaddha  14:19

So I think you’re absolutely right. Getting SaaS companies funded at the inception stage used to be the hardest, because people used to say, “hey, how many customers do you have? Have you built a product? What do you have? I don’t know this market, are the big companies going to do it?” So people used to bootstrap most of them. Make some traction, show some customers, and then get some money. 

This time is different, and the reason is the cognitive plumbing layer. Essentially, you need money. It’s capital-intensive, maybe the apps you can do with a lower amount of money. So what’s different this time is you need to lean in both as an entrepreneur and a VC be willing to take the risk and roll the dice. But what should one be looking at as the OKRs and metrics once you get this company funded is hey, “is your product that pain killer? Have you talked to 40-50 potential customers? Have you defined your minimal viable product, launched the product and start seeing is it a must-have or a nice-to?”

Rinse and repeat till you get some people who love you, and then raise more money after the seed to see if you can establish product-market-fit. And if you have that, the rest will be history. So learn, crawl, walk, run, and then we’ll fly. Seed stage? Let’s just crawl. To walk, you’ll need more money. To run, even more money. Then we’ll worry about flying.

Walter Thompson  16:07

We’ll be right back after this word from our sponsors. 

Are you thinking about launching an AI startup? Mayfield’s $250 million AI Start seed fund is actively searching for idea-stage entrepreneurs who are working at the cognitive plumbing layer. That’s models, middleware and tools, data, infrastructure, and semiconductors and systems. Mayfield has a long track record: since its founding, the firm has been an early investor in more than 550 companies, which has led to 120 IPOs and over 225 mergers and acquisitions. If you have a fundable idea for an AI-first startup, email aistart@mayfield.com. 

Every business that incorporates AI must account for data privacy, compliance and governance. That’s why Securiti developed its Data Command Center. Instead of setting up different tools that can eat up time, money and add complexity. Securiti’s award-winning technology automates compliance with global privacy regulations, includes a library with more than 1000 integrations across data systems, and identifies data risks to enable protection and control. To learn more, visit security.ai. 

And now, back to the conversation with Naveen Chaddha.

Walter Thompson

What are some of the biggest challenges in investing in AI startups? We’re seeing what keeps you up at night?

Navin Chaddha  17:27

I think what gets me excited first, is the massive opportunity. The biggest challenges today as a VC of investing in AI startups is that there is just way too much noise. Way too much hype. Raises are again, getting out of whack, valuations are getting out of whack. Most of the people just add “AI” as they used to add “dot-com.” Then they’re mobile, then they’re cloud. There’s no AI DNA. In most of the companies, they’re whitewashing the value prop as more of a vitamin, it’s not a painkiller. So you have to spend a lot and a lot of time to figure out what is real, and what is not real. 

What keeps me up at night is the AI startup market gets over-capitalized. And too many “me too” startups get funded, and essentially, the customers get confused and don’t buy and wait for the winner to emerge. And then the big companies get there. So I would like fewer companies, more focused, built for the long term, built in the right way, if you will.

Walter Thompson  18:54

I do want to talk about what you look for on a founder team. But going back to what we just talked about, how important is it to you as an investor to see a deep bench of AI expertise on the founding team? Are you happy to invest with academics and researchers? Do you want people who have business experience? What does your dream team look like for an AI startup founder team?

Navin Chaddha  19:12

So what I would say is, every team is not going to have everything. But if you’re going to play in the AI space, first and foremost, you need to have some technical chops, and some domain experience. Either you’re worked in a company, you’re worked as a PhD student, as a grad student, you need to have some experience to be able to figure out how to use this technology, how to innovate on it. 

Second, it will be great if you have some team members who understand the domain you’re going after. The good news is if you are playing the cognitive plumbing layer, it’s about the tech, it’s about the product. But if you’re working at the applications layer, then if you’re solving a problem for salespeople, you need to have somebody on the team who understands how salespeople work. If you’re selling picks and shovels, it’s easier. If you’re in the cognitive plumbing layer, it’s easier, you don’t need to understand the domain specific stuff of how sales works, how marketing works, how legal works, how finance works, how business development works. So there is no right solution. And most of these companies will be founded by first-time founders. So it’s the same story again.

Walter Thompson  20:35

Warm intros are hard to come by unless you’re already inside a Silicon Valley network. Can you share any advice for founders who are working outside major tech hubs when it comes to building investor relationships?

Navin Chaddha  20:46

So I would say, you can try everything, you can do cold emails, you can reach out on LinkedIn, the response might be low, because the spam filters, not that we don’t want to respond, have gotten good. So what I would recommend is really try to get a warm referral on somebody you see the VC is connected to. And you know, to get a referral into them. And maybe they’re speaking at a conference, go meet them. Because most of the VCs are approachable and they want to help founders. And if nothing works, cold-call their office, if that doesn’t work, to show up at the office. Just be persistent. Don’t, don’t back off, just keep going. But the easiest one is warm referrals. Look on LinkedIn. Who does this VC work with, which boards they are on, look at through your network [to see] who knows those companies, somehow get to that network and most of the time, you will realize there are only two or max, three degrees of separation from these people. Reach out that way, that will be the best warm referrals.

Walter Thompson  22:11

I’m sure no VC wants to be seen as just an ATM machine. Cash on demand. So what are some of the ways that early-stage founders can actually leverage their inception investors? And where do you expect them to make mistakes and ask for help?

Navin Chaddha  22:23

First of all, as I said, AI, you should view it as being your teammate, I think we see, you should assume that they will be your teammate, they will be your coach, they will be your mentor. So essentially, the first step has to be before you call for help, build a zone of trust with your venture capital investor, get to know them. Spend time with them. be secure in your skin, share bad news earlier than good news. And just establish the rules of engagement, and use them as a safety net. And then be shameless in asking for help and say, “I’m stuck here. I’m thinking about strategy. What do you think?” Engage them rather than evaluating them. “I’m hiring this person. This is the spec. Do you know people? Are you willing to interview them? Hey, I’m trying to get into this account. Can you make an intro? I need help with PR, I need to get placed in Wall Street Journal, I need to get to know Walter.”

So essentially, ask for those things. Right? And ask questions. No question is bad. Right? Learn along the way, make VC or partner and go and go. Don’t hold back! The mistake can be thinking of them not as a team. That doesn’t work! It doesn’t work.

Walter Thompson  23:52

Actually, I want to skip ahead to a question. So what are some of the top questions AI first founders should be asking potential investors in 2024.

Navin Chaddha  24:01

First and foremost, an entrepreneur should not be asking the VC, they should be figuring out, are these people really going to be backing me and my idea, or are they backing the market? Are they going to be there in tough times? Will they watch my back and support me through thick and thin? And rather than asking them, figure out where things didn’t work for them, and how did they act with those founders? And of course, you should ask them for on sheet references. You should ask, how do they add value? Do they have any experience in the AI space? Are they company builders, have they been entrepreneurs, but just spend time with them? Spend time with them like you spent recruiting or getting your co-founder. It’s very, very important. Read them, understand them, diligence them, backchannels on-sheet references, off-sheet references… don’t go wrong on this decision. It’s one of the most important decisions to get your first investor, as important as getting a co-founder, so do the work, whatever time it takes. Do the work.

Walter Thompson  25:25

What does it look like when someone’s rushing this process? What happens?

Navin Chaddha  25:30

I think they make a lot of mistakes. It’s not only just rushing, just going for the highest price sometime? It’s all paper and doesn’t lead to the best outcome. Just relax. Company-building is a marathon. It’s not a sprint, take your time. Opportunity will still be there. Right? It just takes time. This is a different game. This is a different game. You’re dealing with humans. As you said VCs, there are machines, they’re not ATM, you have to deal with them. They’ll show up at your company. They show up at the board meeting. 

What are you going to do when they want to be on your board? You report to the board. Spend the time, make sure they’re the right people. Because most of the time, companies can lose because there’s no market. A lot of times, there’s a market but there’s in-fighting. Well, I don’t have time and neither should you to be part of that. Take your time. Slow down. This is one of the most important decisions. Take your time. 

Walter Thompson  26:32

Going back to the pitch process for a second. And the idea of creating a painkiller not a vitamin. I’ve heard you say that before. But generative AI is still relatively new technology. So when you look at a pitch deck, how important is that TAM slide, or can I just leave it out entirely?

Navin Chaddha  26:47

Yes, I think at least for us, it doesn’t matter. Because whatever TAM you show us is going to be wrong. Majority of the time, the TAM will never get there. And for the companies which work, it will be 10x more than what you have thought. So at the inception stage, that’s where our AI Start fund is focused, I go back to the same thing. Don’t worry about revenue, don’t worry about market size. Focus on, “is this our top pain point for a certain segment of the market?” They have a headache man, they need Tylenol or Advil, you’re selling them vitamin C, they don’t even know why they need vitamin C. Second, right, like, think abstractly and engage with them. I have talked to 40 customers, 50 customers, I think they have this need and extrapolate it think —  how many customers in the world need it?. But tell me once you know that, how are you going to get there? Talk about fast iteration, the Lean Startup, because dinosaurs never survive. 

Tell me how you’re going to reach these customers? What do you think your business model will be? Let’s engage in those kinds of things. If you’re going to fail, fail fast, don’t worry about revenues at an inception stage, dollar amount of the TAM, dollar amount of the SAM, it’s irrelevant, irrelevant. Focus on basic principles of building a company. It’s a paper-and-pencil inception stage, you don’t even have a product. Right? It’s like you’re gonna build a tall, tall building, man, like, let’s figure out the architecture. Spend the time. Let’s build the blueprint. What will it look like? This is not good, that is not good. Let’s do it ahead of time, rather than putting the building and you keep adding more floors in the building tilts — just doesn’t work.

Walter Thompson  28:45

So there’s no building the plane in the air when it comes to AI, then.

Navin Chaddha  28:48

You can’t, right? Like, just take your time. Once you’re aligned, go, go, go, no looking back, go, go go. Because then you know, you’re prepared. You’re going to war, when you’re building a startup, come up with a game plan and a backup plan and go, don’t say “I’m gonna just land up in the field, and then I’ll figure it out.” It doesn’t work that way. The best teams in any sport, when they practice, they watch the videos of the other people they know their weaknesses. Think before you act, plan before you build,

Walter Thompson  29:21

But there’s still that Silicon Valley trope of building in public, you know, showing people what you’re working on and kind of doing it all very transparently. Does that still obtain for AI startups?

Navin Chaddha  29:30

Yeah, it does. But I think like, make sure you have something before you go talk about it. Because otherwise you’re gonna lose all credibility. If it doesn’t work. Take your time. Do it the right way. Do it the right way.

Walter Thompson  29:43

When it comes to first-time founders, how involved do you get when it comes to helping them define their value proposition and carve out that path to compete?

Navin Chaddha  29:50

I don’t think it’s about first-time founders or second-time or third-time founders. Our business is like that of a coach or a doctor, it depends upon what the need of the entrepreneur is. And where do they need help. So we don’t want to do backseat driving. They need to just tell us, “hey, I want to go from Menlo Park,” where you and I are to San Francisco, and engage in a brainstorm, “what do you think is the best route based on your experience?” That should be the discussion. If they don’t ask me that question, they have Google Maps, I know they can get there. So I think there’s no right answer, you sit down with the entrepreneur at the beginning, first, second, third time at every stage of the company and see, “hey, do you need help? Or you don’t need help.” If they don’t need help? Stay out of their way. Let them execute more power to them. I want more entrepreneurs like that. But if they ask for help, either jump in yourself, or have your broader team help, or you have a broad network, introduce them to the right people. They are the driver. Never as a VC, get into the driving seat, though — backseat driving doesn’t work, doesn’t work.

Walter Thompson  31:10

I mean, I think a lot of investors encourage founders to cultivate that mindset that they’re in this for the long haul. They’re building something, it’s going to take years. But it seems like a growing number of AI startups are finding the cost of developing their model so prohibitive, that they’ve started looking for a buyer trying to find an exit without access to compute and data. Is there any way to get a bootstrapped AI idea off the ground? And do you think any AI startups will be born in garages?,

Navin Chaddha  31:36

I think they’re already being born in garages, and I would say, don’t solve problems, which are already solved, solve new problems, raise the right amount of money. To get to the next milestone, maybe it’s four million, maybe it’s 10 million, maybe it’s 20 million, maybe it’s 30 million, find the right set of investors who can give you that money. But please, please, please, please, please don’t build a cloud. Don’t build an NVIDIA chip, don’t go compete with OpernAI, endorse them, use their stuff, and innovate around the edges or above them on the problems they are not solving. And the world needs help on or solve adjacent problems at the layer they’re playing.

NVIDIA chip, we have a very good company: When AI gets to the edge, they provide cooling, cooling and devices where fans can go, it’s great for the whole world solve an orthogonal problem, you go to the infrastructure layer, clouds are being used, there is need for trust, model evaluation, safety, security, go solve that problem. You don’t know whose model will get used, you don’t know whose cloud will get used. So solve problems. And all the cloud providers will be happy. NVIDIA will be happy, they can sell more chips, more cloud adoption, more model adoption, everybody wins. 

Align yourself with the roads, the waves that are being created, become a parasite on the marketing they’re doing, ride their coattails. Once in a while you can attack them, but make sure you have enough money and a defensible moat. And not a “me too” proposition, because in AI, besides domain expertise, capital, capital capital is very, very important. It’s very, very important. Because of the questions you talked about model development, all trading, compute, just doesn’t stop. So be mindful. Right? Don’t reinvent the stack. Endorse things which have been done, leverage them, be adjacent to them, maybe disrupt them once in a while. But figure out how you play in the ecosystem. Don’t become an island, no need to become an island.

Walter Thompson  33:57

Returning to that comparison, between AI-first and B2B SaaS startups, most software as a service companies will start with small and medium-sized businesses and work their way up to the enterprise customers. That doesn’t seem like that’s the path for AI, you’ve really got to go straight to the enterprise. But that path is fraught with difficulty. So what challenges are you seeing with regard to AI adoption at the enterprise level? And how would you advise a smaller startup that’s trying to get buy-in with CxOs?

Navin Chaddha  34:26

So I would say the market is still early. You’re absolutely right on working with the impedance mismatch between a startup and a Fortune 500 company. God, man: it’s 10,000x, 100,000x, a million x. So choose your customer wisely. My advice would be: the size of the customer doesn’t matter. It’s more important, like they used to say, a tech-forward customer. A cloud-forward customer, a mobile-forward customer. Now you need an AI-forward customer size doesn’t matter. 

Are they ready? Is the pain high enough? Will they jump in and make sure if they’re jumping in, you’re not becoming their development shop, you’re not building a custom product. Build a product that’s applicable to a broader set of the market. So that’s the way I would look at it: are they putting you in the lab, and will never see the stuff, or they’re ready to write the check and ready to deploy. And if you can’t match the impedance between them and yourself, get a big company like Accenture to work with you get somebody else. But make sure, right, this is a broad set of problems, you’re not their development shop, and choose the right company, the right company. Don’t jump into the first customer, it will bankrupt you. So I’m not sure your first customer should be a big company, find an AI-forward company. It could be a fast-growth SaaS company, which needs to leverage AI, or a lot of companies are looking at building on AI. Security is a huge issue. They don’t know what model trust they can do. Figure out who they are. get traction, get proof points, and go with customers who already?

Walter Thompson  36:21

You kind of touched on them briefly, but kind of go a little deeper: what are some of the downsides of working with a big company, and how can founders mitigate them? I’ve heard from CxOs that they don’t have a lot of time to work with startups, they only have time to work with maybe two or three at a time, because they require so much resources and overhead. And then also the issue of liability: it can it’s a make or break for the smaller startup, but let’s talk about the break aspect.,

Navin Chaddha  36:45

I have seen the way a big company works versus a startup. A startup has five employees, 10 employees, zero salespeople, and the founder is the one selling. And the enterprise brings 20 people 30 people, 40 people to their office, the office can’t accommodate them. The office was only for 5-10 people. So be careful, right, they might be wasting your time. So the way I would go about it, is figure out if the team you’re working with is nimble. They’re not at the customer. They’re not layers or decision makers. They drown you with one meeting, a second meeting, third meeting, three months pass, six months pass, nine months pass, you will be dead. You need early adopters, right? When you go through a tornado market, right? You’re not looking for majority or late-majority, find customers, if your product was built, can use it and use it very quickly. And want to die if they don’t use your product. If you have a product, go find customers. When they hear about you, they’re going to get to the top of their building, and jump up and down, up and down from the roof. Shouting, “man my day got made, I heard about this company, this is the problem I was trying to solve for 10 years, and I finally found it!” And that’s what you have to figure out, not “you come in, I’ll do this for you.” Like maybe you can, maybe you can’t, it doesn’t work. Alignment of interest.

Walter Thompson  38:22

Mayfield has a lot of in-house expertise when it comes to AI. But most investors, they don’t. They’re generalists. So do you have any thoughts about how that could affect the companies they’re investing in, and the sector at large as these companies mature?

Navin Chaddha  38:35

I think this is where first and foremost, I put the problem on the entrepreneur listeners: make sure you’re partnering with somebody who’s in for the long haul, believes in AI, is not going to run away, and has some expertise. And I think there are enough of us and Mayfield is one of them. So where are the entrepreneurs going to go wrong? If they don’t choose wisely who their investor is, they’re going to make mistakes. And nobody would have seen the movie before or have pattern recognition to keep them within the guardrails, right? 

Second, a lot of money will be lost. The AI sector will get over-funded. As I mentioned earlier, lots of “me too” companies and what will it lead to? Customers will get confused, pause, established companies will catch up. And this is what is called “the trough of disillusionment,” disillusionment in the hype cycle. Let’s not do that please, solve real problems. Fewer companies, must-have, do it the right way. Otherwise, some will make it on the other end, but 90% will fail. I don’t want that. The industry doesn’t want that. Let’s play wisely. Let’s have muscle memory. 

We made money in internet, lost a lot of money. We made money as an industry in mobile, lost a lot of money. That’s the nature of venture, it’s okay, let’s make new mistakes. That’s a muscle memory. Because if you’re not careful, the next tulips have arrived, this time they are the AI flavor. We saw it in the PC era when the tulips became.com, then they became social, then they became mobile, then cloud. Let’s be careful, huge opportunity. Make new mistakes, not the same mistakes. Let’s not create a bubble. When it bursts, it hurts. You know what I’m talking about? Right? There was a COVID bubble recently. Sure. Relax, sell. So if it’s such an important wave, let’s do it the right way, the right way, this time.

Walter Thompson  40:41

You touched on this a little bit earlier in the talk, but go a little deeper this time? How does Mayfield view the role of AI with regard to driving societal change? And how does that inform your investment decisions?

Navin Chaddha  40:52

So I think this is where you need balance as an investor. Mayfield pioneered the concept of conscious venture capital, where we believe beyond doing financially well, for the society, our limited partners, who are endowments, foundations, pension funds, people in need, who serve needs of other institutions who serve needs of people. Our job is not to just do financially well, but do good for the world. And that’s what we call conscious capital. So we look at our role as investing responsibly in the AI space. We’re looking for companies, which eventually Yes, there will be loss of jobs, those things in the short run. 

But in the long run, when humans adopt AI, do they perform at human-squared? The GDP improves, the income per person improves. Think about the end goal. And are they building the company in the right way thinking about trust, thinking about safety, thinking about privacy and security? If they don’t, they’ll be shut down. Forget about compute costs and data costs, nobody’s going to use them. People wanted self driving cars to fail. They are just looking: lobbyists will come, unions will come, this will come. So just keep that in mind. Keep the big picture in mind. huge potential. But let’s put guardrails. Let’s make sure it’s done the right way.

Walter Thompson  42:33

My last question, when I started this podcast, I realized I had to educate myself about generative AI, just so I could ask the right questions. And it’s been a real journey for myself. But can you share some of the resources you’ve used to learn more about AI? What are you reading or listening to?

Navin Chaddha  42:49

My belief is very simple: entrepreneurs need to surround themselves with experts and excellence. That’s what I do. Hire people in your team, or AI-native. Go talk to entrepreneurs, go talk to university professors, PhD students, postdocs. That’s how I learned: go talk to established companies. Jensen [Huang] at NVIDIA, Satya [Nadella] at Azure, or Microsoft, right? Go talk to TK [Thomas Kurian] at Google. Talk to the model providers, understand what they’re doing, read technical papers. We have 3,000 customers in our network at Mayfield — understand their needs, do surveys, you have to be hungry. 

But that’s the way I learned, which is essentially, talk to people, understand their needs, then go verify is this technology safe or not? And then sit back and connect the dots, and say, hmm, “I was an entrepreneur in both the PC and the internet era. Then I was an investor when social, mobile, cloud happened. Now it’s AI, can I connect the dots based on real data?” So I’m trying to learn the basic building blocks that are happening at universities, what engineers are thinking, what’s happening in labs, what are the big providers doing the customer pain points, and then figuring out how Mayfield should attack that market, and what kind of entrepreneurs we should be partnering with. So that’s my approach. 

And then we want to work with people like you who will make not only us, but the rest of the world aware of what the possibilities are. I think the thirst for learning at Mayfield and also me has grown 10x. And this is an amazing opportunity. Even for a person like me who has been in the business for 20 years. I couldn’t have wished for better because you need to start from scratch and realize: same rules, new technology. 

Use the past, learn from it. Have no shackles, apply some of those principles. But learn, adapt, and keep going.

Walter Thompson  45:13

Fantastic. I really appreciate the conversation today. Thanks for your time. Absolutely.

Navin Chaddha  45:17

It’s an honor to be here with you and to be partnering with you. Onwards and upwards. Let’s make the pie bigger.

Walter Thompson  45:23

Thanks, Navin.

Navin Chaddha  45:24

It’s a pleasure.

Walter Thompson  45:26

I’ll be right back with some show notes after a word from my sponsors. 

Fund/Build/Scale is sponsored by Mayfield. If you have a fundable idea for an AI-first startup at the cognitive layer, email aistart@mayfield.com.

The podcast is also sponsored by Securiti, pioneer of the Data Command Center, a centralized platform that enables the safe use of data and GenAI. To learn more, visit Securiti.ai

I’d like to thank my guest Navin Chaddha, managing partner of Mayfield. Coming up in Episode Two, I spoke with Rodrigo Liang, CEO and co-founder of SambaNova, about digging a moat, customer discovery,and product led growth. He also shared some advice for anybody who’s thinking about joining an AI startup, and had some thoughts about fundraising in 2024. 

Please subscribe to Fund/Build/Scale on your favorite platform, and if you liked this episode, I hope you’ll give me a great rating. For now, you can find the Fund/Build/Scale newsletter on Substack.

The podcast theme was written and performed by Michael Tritter and Carlos Chairez. Michael also edited the podcast and provided additional music, and I’m deeply grateful. Thanks again for listening.

World Economic Forum: How AI Can Usher In an Economy That Puts People First

  • Artificial intelligence (AI) is going to transform business and the economy, but what are its potential benefits and downsides?
  • Everyone in business has a role to play in ensuring AI is truly a force for good.
  • We need to approach innovation with the right motivations and accountability mechanisms to ensure AI’s promise becomes reality.

It’s hard to overstate the impact that artificial intelligence (AI) is going to have on every aspect of business and the economy. In the tech industry, we have long understood its potential. Much of the rest of the world woke up to its transformational power more recently, particularly following last year’s launch of ChatGPT.

Naturally, that awesome power has raised a slew of questions. What are AI’s potential benefits and downsides? Will the former outweigh the latter? What will AI do to productivity and jobs? Will it exacerbate inequality? Where is it going to take us? Will we be able to control AI, or will it control us? Each of these questions merits a healthy debate.

As a veteran of the tech industry and as an investor whose firm’s motto is “people first,” I bring an unabashedly optimistic perspective to the conversation. But optimism doesn’t mean complacency. Everyone in business – from the startups creating AI technologies, to the companies adopting machine learning algorithms to supercharge their products or operations – has a role to play in ensuring AI is truly a force for good.

Here’s why I’m optimistic, and what I think companies big and small should be doing today to ensure a positive outcome.

AI will augment humans, not replace them

Those of us who have been around the tech industry for years have seen it again and again – powerful emerging technologies raise fears about jobs. And certainly, new forms of automation eliminate some jobs. But they also create efficiencies and streamline repetitive tasks, allowing humans to move up the value chain.

Ultimately, more jobs are created. Personal computers, for example, eliminated some jobs for typesetters but helped to create far more employment through desktop publishing. I am convinced that AI will have that kind of effect, but on a much larger scale. It will create massive productivity gains that will allow businesses to invest more, innovate more and generate new jobs along the way.

But that’s just one part of the story. I also believe AI will give us new superpowers that will make our work more satisfying and our lives richer, leading us into an era I think of as “human squared.”

How? First, our way of interacting with technology will change. Going forward, our primary way to communicate with computers will be through rich and layered conversations. Perhaps more importantly, for the first time, technology will be able to perform cognitive tasks that augment our own capabilities.

Rather than merely speed up and automate repetitive tasks, AI will generate net new things much like humans do. The result is that we’ll be able to multiply our own capabilities with a human-like assistant. Whether you call it an intelligence agent (AI becomes IA), copilot, teammate, coach, genie or something else, it will make us immensely more capable, regardless of the task we are performing. I believe that we are witnessing a whole new layer of cognitive plumbing that is being built to drive this wave.

cognitive plumbing graphic

Founders should build trustworthy companies from day one

The entrepreneurs creating tomorrow’s AI systems and applications all face choices in how they develop and harness the technology. That’s why, as an investor in startups, I encourage all founders to build trustworthy companies from day one.

What do I mean by that? One of the first things I do when I meet founders is to probe their values. I want to know whether they are driven by human-centric mission, and whether their vision for how to deploy technology is aligned with ours.

These are probing and broad ranging conversations that inevitably cover a few key topics:

  • Trust and safety can never be an afterthought. In the past year, the potential pitfalls of AI – whether it is hallucinations, lack of transparency, inequity, bias, deep fakes, copyright infringement or other issues – have been well publicized. If AI is going to be a force for good, founders must not only be aware of them but also determined to address them. They must evaluate the trustworthiness of the models they develop and use, and ensure that they are compliant with a nascent but rapidly growing regulatory regime.
  • Data privacy is a human right. AI is fuelled by data and companies should treat that data responsibly. That means not only complying with regulatory regimes, but also embracing ethical practices around its use. That ranges from transparency about what they will and won’t do with it, to the handling, classification, and security of sensitive data, the careful inventory, lineage, retention and consent to use of everything that goes into AI models. The time to put guardrails around data practices is now, not after breaches or privacy violations are exposed and it’s too late to prevent harm.

Startups should not only dedicate themselves to AI safety, accountability, and averting harms, but also state their commitments publicly, as, for example, Anthropic has done, which should serve as a model for others to emulate.

Responsible AI is a practice, not a checklist

Just like the companies creating AI technologies, those that are deploying them have choices to make and, I would argue, a duty to do so responsibly. I’m encouraged by what I’m seeing and hearing.

Whether it’s through our annual survey of CIOs (chief information officers) or in conversations across the industry and at events like the World Economic Forum’s Annual Meeting in Davos, I have noticed a sense of collective urgency among many business leaders to do the right thing.

But that’s easier said than done. Implementing responsible AI practices across an organization is a challenge that requires resources, commitment and leadership. Like anything in business, it must begin with an adequate budget and a mechanism for accountability.

The person or group that oversees it needs to have the visibility and stature within the organization to be able to convene stakeholders – across tech, legal, compliance, audit, and other functions – and influence decisions. Importantly, the CEO and board should know how those groups are working together and the roles each has to play.

Some companies are waiting for regulatory regimes to force their hand. I think that’s a mistake. Deploying AI responsibly, and doing it in a way that doesn’t slow the pace of innovation, is not like flipping a switch.

Those who aren’t laying out thoughtful plans today are already behind. For those who don’t know where to start, a growing number of certification programmes from organizations like the Responsible AI Institute can help lead the way. The time to do so is now.

AI is already improving our economy and wellbeing in myriad ways, including higher worker productivity, more accurate health diagnoses, new forms of drug discovery, better decision-making, to name a few. But that’s just the beginning.

I am certain that businesses and entrepreneurs will be able to harness AI technology in new ways that will unlock untapped human potential and benefits on an unprecedented scale. We just need to approach innovation with the right motivations and accountability mechanisms to ensure that promise becomes reality.

This column was originally published on the World Economic Forum website.

Alchemist Announces Backing from the Mayfield Fund

Mayfield and Alchemist Accelerator logos

MENLO PARK and SAN FRANCISCO, Calif. – January 30, 2024 

Alchemist is announcing today an investment from the Mayfield Fund.

The money will continue to support Alchemist’s mission of backing and launching the most promising startups disrupting enterprise innovation.

“We are thrilled to continue our long-standing partnership with the Mayfield Fund, which was one of the original backers of Alchemist when we first started,” says Ravi Belani, Managing Director of the Alchemist accelerator. “Mayfield has a deep understanding of how to support technical founding teams building iconic companies over the long haul. We are honored to partner with Mayfield again to usher in the next generation of founding teams with both deep technical skills and shrewd market vision. These are the teams that change the world.”

Beyond just investing in Alchemist, Mayfield will also be supporting Alchemist founders with ecosystem events, advice, and thought leadership.

“Mayfield and Alchemist have always been aligned in understanding the power of visionary, tech-driven teams to change the world,” says Navin Chaddha, Managing Partner of the Mayfield Fund. “These are unprecedented times for being an entrepreneur, and we are proud to back Alchemist to be a launching pad for ambitious teams looking to change the enterprise.”

For more information about Alchemist and to apply for the next class, visit alchemistaccelerator.com.

About Mayfield

Mayfield is an early-stage venture capital firm with a people-first philosophy and $3 billion under management. The Firm has a proven track record of partnering with founders starting at the inception stage to help build iconic companies that leverage innovations in IT and biology. The Firm has invested in more than 550 companies, resulting in 120 IPOs and more than 225 acquisitions. Some recent notable investments include HashiCorp, Lyft, Poshmark/Naver, Mammoth Biosciences, Marketo/Adobe, NUVIA/Qualcomm, Outreach, Rancher/SUSE, SolarCity/Tesla, and Volterra/F5. For more information, go to mayfield.com or follow Mayfield Fund on LinkedIn.

About Alchemist

Alchemist is a venture-backed accelerator focused on the development of seed-stage ventures that monetize from enterprises (not consumers). Alchemist backers include many of the top corporate and VC funds―including Mayfield Fund, Khosla Ventures, Foundation Capital, BASF, NEC, Cisco, Siemens, GE, and Salesforce, among others. The accelerator’s primary screening criteria is on teams, with primacy placed on having distinctive technical co-founders. Notable alums include LaunchDarkly, Kyte, MoEngage, Foresight, Privacera, Matternet, and mPharma. For more information, please visit alchemistaccelerator.com.

 

Partnering with Sema4.ai to Accelerate the Agent Economy

Mayfield and Sema4 logos on blue backgroundAs the AI era moves from early adopters to the enterprise, I see the plumbing layer rising again. This vision has guided our latest investment in the cognitive plumbing of generative AI.

With a rational Series A round of $30.5 million from us and Peter Fenton of Benchmark, an investor I respect immensely, Sema4.ai debuted today to go after the $88 billion opportunity to accelerate the AI agent economy. Sema4 is led by a dream team of experienced founders — Rob Bearden, Ram Venkatesh, Suds Menon, Paul Codding — behind enterprise powerhouses Cloudera, Docker, Hortonworks, and JBoss. They have joined forces with Antti Karjalainen and the team of open source automation pioneer Robocorp to deliver on their mission to build an intelligent agent platform that uses semantics and reasoning to transform how knowledge workers collaborate with AI.

Sema4.ai Leadership Team
Sema4.ai Leadership Team (L-R): Rob Bearden, Paul Codding, Antti Karjalainen, Suds Menon, and Ram Venkatesh

Our conviction in making this investment began 25+ years ago, as I have been a believer in the power of plumbing to help consumers and businesses realize the benefits of technology waves. During the Web era, I joined a long line of tech industry school dropouts to co-found my first company in 1996 with my PhD advisor at Stanford University. VXtreme made it possible to stream video over the Internet, and after its acquisition by Microsoft, endures as Windows Media. While at Microsoft, I invested in Akamai, which delivered infrastructure to scale the Web.

In 2014, as the cloud era got into full swing, Mayfield partnered with Armon Dadgar and Mitchell Hashimoto, the open source superstar founders of HashiCorp, whose mission was to elevate the devops professional with a simple and comprehensive multi-cloud infrastructure platform. HashiCorp’s public offering in December 2021 valued the company at $15 billion, and it continues to power major cloud-based businesses today. Over the last decade, we have partnered with many bold enterprise founders including those at CloudSimple, CloudGenix, Elastica, Gigya, Portworx, NUVIA, Rancher, StorSimple, and Volterra who successfully realized their mission to cloudify the world.

As AI shifts from early adoption to widespread enterprise use, the importance of the “plumbing layer” is resurgent. The transformative potential lies in four layers of the tech stack: models/middleware/tools, data, infrastructure, and semiconductors/systems. The breakthrough of GenAI is in elevating humans by providing a natural language interface and doing cognitive tasks — an era I think of as AI+Humans = Human Squared. In this context, cognitive plumbing unlocks the easy creation of agents, applications, and services that can be built on the foundational technology.

Looking forward to partnering with many more cognitive plumbers of the AI age — we are People-First investors with over $1.1B in capital for new early-stage investments who are excited to guide your inception-to-iconic journey.

 

Top 10 Takeaways From Davos

 

Banner on top 10 takeaways from DavosI just returned from an energizing weeklong visit to the World Economic Forum meeting in Davos, Switzerland. I have participated several times, starting in 2002, when I was named a Young Global Leader. This was my first post-Covid era visit, and I am happy to report that the magic that brings world and business leaders to this cold mountain town is alive and well.

NC1 Davos

Highlights of my 2024 visit include:

  • Aligning to the broad cause of Rebuilding Trust around geopolitical issues, the climate and AI;
  • Listening to inspiring leaders such as President Volodymyr Zelenskyy of Ukraine, whose appeal for peace received a standing ovation and whose ask to “strengthen my economy, I will strengthen your security” is still ringing in my ears;
  • Hearing Satya Nadella of Microsoft (whom I had the honor of working with after I sold my first company to Microsoft in the mid-’90s) who spoke eloquently during the fireside chat with Klaus Schwab on how LLMs have been trained to understand human language and now they have to be trained to understand language of nature;
  • Getting insights from sessions that will hopefully result in behavior change in our daily lives – “Young Brains and Screens” and “Navigating Longer Lifespans;”
  • Elevating the conversation around AI as a Driving Force for the Economy and Society – lots of food for thought in the sessions on GenAI as the steam engine of the fourth industrial revolution, the boon or bane of creativity, thinking through augmentation, ethics and regulation and industry applications; my take is that AI is a paradigm shift that will be bigger than the Web but investors looking for tulips are overvaluing startups, which, in my two decades of experience as an investor, never ends well;
  • Thinking of quantum computing’s black swan potential to go mainstream for applications in drug discovery, materials science and supply chain optimization; my take is that it will take a unique founder/investor combination to build these kinds of deeptech companies;
  • Hearing leaders debate on how to develop a long-term strategy for climate, nature and energy by pursuing a systemic approach to achieve a carbon-neutral and nature-positive world by 2050 while providing affordable, secure and inclusive access to energy, food and water; my take is that it will take product-first founders with breakthrough business models (similar to our company SolarCity during the cleantech wave of the 2010s) to grow into enduring companies;
  • Sitting with a group of 25 global VC leaders around a table discussing the state of our industry, including LP mindset, distributions, fund size, exit markets and evolving fund models; my take is that venture capitalists need to be guided by cash-on-cash returns, by providing solid distributions to LPs;
  • Saluting the winners of the Crystal Awards (an architect & educator, an Academy Award-winning actress and a humanitarian), listening to an AI-driven immersive music performance, participating in sundowners but leaving before they turned into midnight parties;
  • Last (literally before I boarded the plane back) and not least, sharing my optimistic view of how global entrepreneurship will continue to thrive in a discussion on a panel titled “No Rain, No Flowers: Funding Start-ups.”

 

NC2 Davos

NC3 Davos

“No Rain, No Flowers: Funding Start-ups”

Watch the panel recording or read the full transcript.

 

“No Rain, No Flowers: Funding Start-ups” Lessons From Davos

"No Rain, No Flowers: Funding Start-ups" panel at WEF 2024

At the World Economic Forum’s 2024 Annual Meeting, Managing Partner Navin Chaddha spoke on the “No Rain, No Flowers: Funding Start-ups” panel. Watch the panel recording or read the full transcript below.

Sara Kehaulani Goo: Hi, good morning and thank you so much for joining us this morning for our conversation. We’re calling it “No Rain, No Showers: Funding Start-ups” here at the World Economic Forum. I want to welcome our guests who are here in person, and I also would love to welcome our guests who are joining us online. If you want to join the conversation, please use the hashtag WEF24. I’ll be your moderator today. My name is Sara Kehaulani Goo. I’m the editor in chief of Axios, a Washington, DC-based news organization, and thrilled to join you here to have this important conversation. And the topic here is we’re going to over the next 40 minutes discuss a global picture of funding for startups. And here we have some great panelists who bring different perspectives, not only from their different geographies but also their different aspects of the business.

And why does it matter? That’s what we say at Axios. The why it matters is that global funding and entrepreneurship is an important engine, of course, of economic growth, of jobs, a driver of the economy. And yet as many of the panelists will probably talk about soon, access to capital remains very tight. And just last year it was down 42%, global venture capital in 2023 year over year. So we’re here to talk about that, but also to look forward to 2024. As many of you know, there’s a lot of excitement also about AI and funding and investment in that space in particular. So without further ado, I’d love to get started and introduce you to our esteemed panel. First I’ll start with Navin Chaddha. He’s the managing director of Mayfield Fund based in the United States. Rishi Khosla is co-founder and CEO of OakNorth of UK. Ahmed Karsli is co-founder, chair of Papara from Turkey. And Jacqueline Poh, she’s managing director of the Economic Development Board of Singapore. So welcome to all of you. I’m so glad that you’re here.

Navin Chaddha: Pleasure.

Sara Kehaulani Goo: Okay, so we’ll start with you Mr. Navin. I read just a few months ago that you are a contrarian, venture capitalist and that you’re writing more checks than ever while others are pulling back. And you did raise a lot of money in 2023. So I’d love to hear a little bit more about your philosophy and this contrarian approach to investing.

Navin Chaddha: Absolutely. So first, for context for people, the firm I manage, Mayfield, we have been in business for 50 plus years. So we have seen many upcycles, many down cycles and have been very lucky to be part of a lot of iconic companies, 120 IPOs, 225 acquisitions. And what we have learned is when markets get troubled and when they are tough, that’s a great time to start an early stage company. And the aim is when you’re investing at the peak of markets, assets are inflated. And when markets dip, you can get in early, you can get in cheap and help build companies. And so our approach is when 2020, 2021 was happening, markets were peaking, money was pouring into venture capital. We slowed down based on our patent recognition over the last 50 years and we said this is the time to sell rather than to buy.

So we created a lot of distributions because paper gains IRR, TVPI is only so helpful. LPs want money back. Our job is to take small boxes of money, make them bigger and send them back. So I think our feeling is the moment the market’s got sanity and they’re tough, just lean in. So we have been growing year-on-year, are investing by 50% and it’s okay. And I think there’s a lot of talk about, and our focus is the United States, that funding levels have gone from 350 billion in 2021 to 240 to 170. But just as a reminder for people 2018 and 2019 was 140 billion and 2012 was 40 billion in the United States in venture capital.

So we should stop complaining, right? 170 billion is still 4X of what it was a decade back. And you can take any amount of inflation, put it cumulative, you still don’t get to more than 4X. So we are still up two to two and a half X. There was a lot of indigestion in private markets. It’s getting cleared up and it’s survival of the fittest. So we are extremely excited to be funding early stage companies, that’s what we do, and help them become industry leaders of tomorrow.

Sara Kehaulani Goo: And oh yes, go ahead.

Rishi Khosla: Can I just pick that up? So I’ll give a slightly adjacent viewpoint. So OakNorth’s a business, which is a digital commercial bank that focuses on really funding scale ups. And to take your point, Navin, if I look at 23 for us, you look at sort of the fall of SVB and then lastly, sort of First Republic and Signature. All of those banks really played into the entrepreneurial class, right, in terms of supporting the entrepreneurial class. And for us, again, it was one of those moments where we stepped up.

Navin Chaddha: Yep.

Rishi Khosla: Right. So last year, 23, we would’ve lent an additional $2 billion in terms of gross lending. We are a UK-based business, but we decided to enter the US because of,-

Navin Chaddha: Makes sense.

Rishi Khosla: Somewhat of the void which was left. And if I look back at our history, we started the business since September 2015. When the Brexit referendum vote happened in 2016, we were still a very small business, but we tripled the size of the business in six months, right, because everyone else retreated. Covid, right, at the beginning of Covid, everyone else retreated, we stepped forward. So I think there’s a lot to be said in that when there’s general exuberance, whether it’s in the equity or debt markets, it’s sort of the time to be slightly more cautious because everyone’s getting a bit heady. But at times like this is exactly when, if you lean in, our view is you end up building a better business and you support great businesses because the entrepreneurs which are actually doing something in this type of environment tend to be the stronger entrepreneurs who actually have the wherewithal to actually grind through.

Sara Kehaulani Goo: And can you go a little deeper on that? I’d love to get your perspective about where do you feel we are now? There’s a lot of buzz and excitement about investing in AI companies. Everybody’s putting AI attached to their business, whether that’s real or not because it does have a lot of buzz. Do you feel like we’ve hit bottom when you think about the market, down market, are we still there? Are you investing more this year?

Navin Chaddha: Yeah.

Sara Kehaulani Goo: Basically.

Navin Chaddha: We are going to be investing more this year and I think markets go through ups and downs. Humanity is optimistic. There’s always a next new bubble which gets created. The next new tulips get found. And I think this wave of AI is actually as big as the PC era, and it’s a fundamental transformation of how humans are going to interact with machines and how AI is going to be able to do cognitive tasks to amplify humans. And whenever that happens, right, like there’ll be a lot of excitement, a lot of buzz, and as a result, fewer companies, more capital chasing again, the valuations will go up. So what we have learned is we cannot, as a venture capital firm, look, as an early stage firm, we were investing in AI for five to 10 years back. And then with GenAI, it happened last September with ChatGPT.

But as a VC, what we have learned is you have to look for the next wave and catch it and be ready when it’s going to peak. But when it passes you, that’s no point if you have so far to go. So you have to have a long-term perspective. We already have like 25 plus investments over the last five to seven years and these will grow and we’ll be careful on which companies we invest in in the AI hype, but the stage we invest in primarily at the inception stage, very difficult to invest. It’s people with paper and pencil ideas and prices are still good. And our job is to find the diamond in the rough. And that’s what we love doing. And it’s a craft, right? There’s no numbers, there’s no market, there’s nothing. Sometimes there’s no teams and then you help build them and the follow on money just gets poured.

So it’s the same story. It’s the same story in which you move from one wave to the next, but this is real. Real companies are going to get created. And you guys do a great job of covering some of these things. You look at the market caps of the magnificent seven, like the way NVIDIA has grown, the way Microsoft is growing, Google, Amazon, Facebook, it’s just amazing. And today the value has been accreted with the hype towards these big companies, but there’s a lot of open problems.

Sara Kehaulani Goo: Yeah.

Navin Chaddha: So I’m very, very bullish, but very mindful of how VCs invest because everything is not going to be worth 10, 20, 30 billion. And also a word of caution to entrepreneurs, whether you take debt, whether you take venture money, it’s all borrowed money. Let’s build real companies and not just focus on just growth, growth, growth. Look at, hey, when can I be self-sustaining? So I think hopefully all of us have learned from what we have seen over the last 20, 30 years and apply those principles. Otherwise, our job is not as a venture firm. We’re not a media company to accumulate logos and buy beachfront properties. Our job is to create financial returns.

So it’s a healthy balance, right? At the end, investors, institutional investors, look at venture capital firms and say, okay, public market bonds do this, public markets do this, buyout does this, real estate does this, private equity does this. I’m giving money to the most illiquid asset class. And you’re supposed to make, as a top tier firm, 500 to 700 pips more per year than some of those other assets which are less risky. So at the end, it looks great. You’re creating startups, building brands, you’re a money manager, right, like at the end to be able to raise money. When you take money from people, you have to multiply it and give it back. And that’s where, right, people who have been in business for a long, long time just realize it’s not about this hype, it’s about making money.

Sara Kehaulani Goo: Ms. Poh, I’d love to bring you into this conversation because I read that your agency has been focusing on AI as a focus for supporting the growth of startups. So what is your perspective? We just heard the private sector perspective, the venture capital perspective.

Jacqueline Poh: I think there are many perspectives on AI, but one of the worst reasons to set up an AI company, or to invest in an AI company, or set up a fund, or aim to raise a fund is to tag AI behind everything in the hope that that’s the only reason why someone will invest in you. So I would be a bit cautious about the hype. Singapore’s put up a national AI strategy, a second version. We’ve put up our first version a couple of years ago here in Davos in fact. It combines governance as well as a big focus on use cases as well as what we’re doing for compute and the development of our own LLM, which is called SEA-LION, the Southeast Asian large language model. Southeast Asia is a hugely vibrant area and location for venture capital.

Singapore alone has 400 VC firms and funds, including some of the PEs as well and some of the biggest ones in the world. We have 1,100 AI startups. And even then I say, show me the business model. I think that we are at the stage in terms of generative AI, where it is unclear where the commercialization opportunities will lie. Just because NVIDIA is making money from GPUs does not mean that an applications company is going to make money on regs. So it’s very important for us, for any investor in any company to figure out first of all where the value is actually residing for AI. But I think that it’s hugely promising. I think the potential is there and I think that it’s given life to an otherwise possibly fairly more abundant sort of VC market in a time of rising interest rates.

Sara Kehaulani Goo: Right.

Jacqueline Poh: But the fundamentals will always remain and the smart money will follow the smart money. My own sense about the overall market is that similar to what my colleagues here have mentioned, it may be time to come in. This is a time where even though interest rates are likely to remain maybe a little higher than anyone would want for a bit longer, they are showing signs of moderating and valuations have definitely come down first in the United States and then in Europe, and then they decided to come down in Asia. And some things are looking a lot more reasonable, particularly growth stage companies being a lot more reasonable. So various ways of funding including acquisitions by existing companies. It is not a bad time if you are an existing company and you want to help that startup exit through an acquisition as well.

Sara Kehaulani Goo: And let me just mention something a little bit provocative here since you talked about AI, and this is really a question really about the distribution of investment globally. Yesterday at an Axios house event, we had Alex Karp, the CEO of Palantir, join us for a conversation. He said, this was the headline, that he criticized. I guess it was, I don’t know if it was a criticism or just an observation from him, but he said that “The European startups scene is anemic on AI and that Americans have a huge advantage in terms of their investment and the innovation they’re seeing there.” I’m not just saying that because I’m American. I would love to hear your perspective to respond to that. And is he right? Do you feel like there’s enough investment either from the private sector, barriers to entry lowered by governments or regulation to help that more even distribution?

Rishi Khosla: In terms of specifically Europe? So I think clearly if you look at where the large innovation has come from, right, so you look at OpenAI, you look at Meta’s model, you look at sort of Google’s model, Google DeepMind. DeepMind is actually based in London. Right. So fundamentally Bard is created in London. So whether the innovation, whether there’s the innovation talent, right, in Europe, I believe absolutely there is. Whether those are European companies which are actually taking those and sort of commercializing, clearly not. Right. They’re American companies. And I think that’s just a wider topic. Right. That’s just like, it’s not AI specific.

Sara Kehaulani Goo: Right.

Rishi Khosla: Right. It is a question of, again, the size of the ambition. It’s a question of the availability and depth of capital. It’s availability. I mean, I was at a previous session this morning and the Moderna CEO sort of shared the fact that they had spent $5 billion before they made their first dollar of revenue. Right. And that was in response to Oxford University talking about the fact that they’ve got a part of funding for startups, right, which lets be clear, is probably a small nine figure number. Forget about a 10 digit number. So the size of capital which is available for that for, I mean you look at how much OpenAI has raised, right, clearly from Microsoft and from others. So all of those things I think drive that fact. Now where are the users and where’s the application and therefore the outcomes from AI going to be felt? I think I’d say we’re very high on the hype cycle at the moment, right, but we’re very low on the actual outcomes delivered.

Sara Kehaulani Goo: Right.

Rishi Khosla: Right. And like with most of these things, I mean, I’m not old enough to remember the PC era cycle, but I’m old enough to remember the internet era and I remember sitting there in ‘97, ‘98. Right. And the whole world was going to change in the next 12 or 24 months and it didn’t, but it did in the next five or 10 years. Right. And I think with GenAI we’ll be exactly there. Right. And it may go quicker, it’ll probably go quicker. And I think that European companies will adapt and be users of and innovate with the technology, but they may not be the entities which are commercializing the technology but creating it.

Sara Kehaulani Goo: Right. But Mr. Khosla, you’ve also, I’ve heard your critique of there not being enough investment in certain types of companies, particularly in the UK. So I guess from where you sit, what more needs to be done?

Rishi Khosla: Well, I’m a massive, how can I put it? We’re massively passionate and vocal about the fact that the UK, and I’d extend this to most of Europe, we don’t have the right environment to actually enable entrepreneurs to scale businesses. Right. We have a great environment to help them start businesses, but the funding markets, the availability, or talent, right, in terms of just people who’ve had experience scaling. Right. You see it in the valley. Right. You want someone who can go and spin up a new product line within an existing business, right, and take it from sort of zero to one and then get someone else to take it from one up. It’s like you’ll have a list of people that have done it before, right? In Europe it’s like you’re scratching your head because it’s like the motion hasn’t been driven, right? Because there haven’t been that many businesses which have sort of gone from startup to true scale.

And so you’ve got that aspect. You’ve got the size of the ambition point, right, where if many European entrepreneurs, you get to the point where you create a business for, which is worth tens of millions, hundreds of millions. Right. And they’re willing to sort of say, you know what? That’s good.

Sara Kehaulani Goo: Yeah.

Rishi Khosla: I’m sort of going to step off this and let someone else take it forward. Whereas clearly at least the Valley mindset, et cetera, is that’s getting started. Right. That would not be a failure, but it wouldn’t be like a home run. Right. So I think that there’s the cultural aspect, there’s the availability of talent and then also the availability of funding and public markets. I mean clearly our public markets across Europe just are not, there’s no depth for growth companies and therefore they’re no growth investors.

Ahmed F. Karslı: I feel quite lucky by the way, when I hear about funding cycles, funding problems. Because as a bootstrap company, you actually don’t care. You have a chance to sort of focus on what you build, right?

Rishi Khosla: Right.

Ahmed F. Karslı: Which is one of the biggest advantages. And when it comes to tough times, as you mentioned, for a company operating in an emerging market like Turkey, tough times are just another Friday.

Rishi Khosla: That’s also true.

Ahmed F. Karslı: Just think about what we have been through in the last one year. Last year, only in the last 12 months in Turkey, we had elections; we had many political problems in surrounding neighboring countries. We had massive inflation, even if it’s getting better right now. And we had a massive earthquake where we lost more than 50,000 people. But still that helps a lot as well because you always have this advantage of building some resilience even if it is unintentionally. Just to give you a great example, we grew by four times in revenues only in the last 12 months while having those problems. So sometimes I feel like I’m quite surprised, but I here just said a year ago, close to a year ago, we made an acquisition in Spain and the investors of the company we acquired were actually complaining about inflation. And it was 6%. We were dealing with 65% inflation in Turkey. So sometimes as actually, like you guys said, those tough times might be a good teacher as well.

Navin Chaddha: Yeah, I would say you’re a black swan. That’s what venture capital and success is driven by and we need more and more of that. Right. So here is my take to be complementary to what was said. With a perspective of the United States, right, like first of all, technology innovation is happening everywhere. Right. It’s happening in Asia-Pac, it’s happening in the United States, and this is at research university level, it’s happening in Europe. So what’s different, even in the United States? If you look at 70, 80, this is in information technology, of the value gets created on the West Coast. And why is that the case? So here is what our analysis is: First, the mindset is a little bit different, I would say then ambition. Ambition is there. Is it okay to take risks? And it’s okay to fail.

Sara Kehaulani Goo: Yeah.

Navin Chaddha: So once you have that stuff right, like the sky’s the limit. Right. So you start with that. Second, and it applies to both, people who are starting companies and people who are funding them. Right. But this is what happens. If you are in a hotbed of established companies now you can poach employees from there and at the end people are building companies. It’s not the other way around. And then this business of taking companies from idea and turning them to iconic is all about mentorship. So this is where maybe VCs can add value, maybe they can’t, but there’s enough entrepreneurs and executives who are willing to give back as board members, angel investors.

So I think this is a network effect, this is a community. And what ends up happening is, it’s nothing to do with talent, it’s nothing to do with innovation. Basically scale just creates more scale and that’s what just ends up happening. And I would encourage in some of the other regions for there to be a change in mindset. It’s okay to take risk, it’s okay to fail because as Einstein said, “If you don’t fail, you’ll never succeed.” So to come up with a big success, it’s okay. Failure isn’t only an opportunity to do it right the next time.

Ahmed F. Karslı: I would like to add something on it, Navin. On Monday I had a chance here to listen to Sam Altman. You were there as well.

Navin Chaddha: Yep.

Ahmed F. Karslı: And before this panel I wanted to ask ChatGPT what advice can be given to bootstrap startups or bootstrap co-founders. And Sam Altman said on that day is that every advice is actually given and every information is there. So when I asked ChatGPT about what advice can I give to bootstrap startup founders, there were like 10 different pieces of advice there. And they all were actually more than, what can I say? But on the other side, I totally agree with you that I found out that actually the only thing missing there is the personal experiences. And I found out that currently when I want to give advice or take advice, AI plus personal experiences of entrepreneurs around me, of mentors around me is actually enough to get real advice because that is the only thing missing, which is not online.

Rishi Khosla: But even if you pick up those comments, there was a very clear view that actually his best sort of insight comes from discussions with people like his mentors.

Navin Chaddha: Correct.

Rishi Khosla: Right. And again, we’re massive believers in that, and I personally as well, and I would sort of say I’ve had people mentor me all the way through my career and different people and also for different sort of issues, questions, opportunities, et cetera. But I think that is such an important part of, again, the ecosystem.

Ahmed F. Karslı: We need to have more communities actually.

Jacqueline Poh: I have a slightly different opinion about what you’ve mentioned, which is to take risk and the existence of your company in an ecosystem that scales easily. There are many, many problems in the world, and there are many entrepreneurial ecosystems which are not built on a market that is very cohesive. So for example, Europe isn’t necessarily all one market in some senses. Latin America might not be, Southeast Asia might not be. America is a little bit more homogenous. And if you find yourself in Silicon Valley, you get a product and you get a sense that there’s a community around that, there’s capital, there’s talent, and there’s a sense of scaling muscle that works in a certain way. I think if you go to different parts of the world, and this is relevant in a world that is increasingly maybe a bit less globalized, a little bit more regionalized, that kind of scaling muscle has to change as well. So in addition to the skills that you’ve mentioned, I think that kind of adaptability is going to be something that is relevant.

Navin Chaddha: So what I would say is having invested in India and China since the mid 2000s, I completely agree with you and my advice to entrepreneurs is focus, and this is a little bit different, focus first on your home ground because there’s faster iteration cycles. Perfect the product, perfect the GTM, and it’s hard for an international company to come in and understand the local stuff. And at the end, the problems you’re solving there are very, very different. Right. I grew up in India in the late ‘80s, came to the US in early ‘90s, and there was like years before you could get a landline telephone. And necessity is the motherhood of all invention. The landlines never happened. It all went to basically cellular.

So that’s where the entrepreneurs have to innovate rather than saying, Hey, I’ll come from India to the US. First nail the problems. They’re there. The folks sitting in the US can’t even sell at those price points. They don’t even know the distribution channels. So you need something more than technology in today’s market. And if your product teams are closer to where the customers are, you win. Right. You win. And there are many, many examples, not only in tech, in CPGs, in financial services, in healthcare, the local companies in some of these markets have done phenomenally well because, so my feeling is it’s a little bit different than US, Israel, but some of these emerging economies are big. They’re big. So go solve the problems there and raise less capital. If the markets are 110th, it’s okay. You can still deliver a lot of value.

So I think that’s what I’ve learned, that get it right with something which is near to you where you have domain expertise and insight rather than somebody sitting in Silicon Valley. Right. That’s where I’ve seen the local companies we have been involved with, which are innovating and especially some of these low income markets at the bottom of the pyramid. Right. Those kinds of income levels just don’t even exist in the US. So I think that’s where my learning has been. Change the problem, focus on your unique advantage, whether it’s what you can charge, whether, how do you distribute it, like what do you do? Right. And the markets are very, very different in these countries.

Sara Kehaulani Goo: Mr. Karsli, I’m curious about your perspective. We didn’t get a chance to talk a little bit about your company. I wonder if you could share a little bit about your experience and whether the comments here and about culture, about regional versus local investment in communities make a difference. Was that the case for you?

Ahmed F. Karslı: I totally agree with everyone. Just to give you an example, the panel’s name is “No Rain, No Flowers.” But I think overwatering is also a problem, right? And I think one of the reasons why many startups didn’t focus on their own local markets, as you mentioned, was actually over raining. Because I have seen that many companies had a huge potential in their own markets, but they were actually pushed to grow as fast as possible. So they had to operate in multiple markets where they tend to lose their focus from their main market. And as a bootstrap company, another advantage is that you become really creative because you don’t have enough funding resources, which means that you have to do everything to be profitability, and you get really creative about when you’re building a product, designing the product, building a fee structure, et cetera. And that actually helps you a lot to do something in your own local market as well. That is a big advantage.

On the other side, culturally, we talked about AI. This year’s main topic is AI here. We all are talking about AI, but when it comes to AI, I really struggled in the first five years of Papara because it’s a regulated business. It was quite difficult to deal with regulators, especially when you’re a new and young startup founder.

So I noticed that we were trying to build something for some time, but the regulator might be a road blocker as well. And I’m really happy that we are talking about AI here for the last five days. It is great to see that many policymakers are happy to discuss AI. But on the other side, I’m a bit skeptical as well because I have seen many policymakers here in Davos. Actually, we’re talking firstly about the potential effect of AI on unemployment. And I can easily imagine right now that in five or 10 years, some political leaders will be campaigning against AI and they will be on the rally and campaigning like, we are not going to let robots steal your jobs. And probably there will be people there who are shouting like human first, human first.

So to prevent that, actually we need to have more connection with the regulators. And to do that, I recently, OECD did something great, OECD, maybe you have seen that, published a report on sandbox use cases of AI. We need more sandboxes like that to actually, like to have this connection between regulators, policymakers and startups. Because in the last five years we had this huge issue with policies about crypto, but crypto was only an interest for financial policymakers. AI will be interesting for any kind of policymaker including health, transport, whatever it is, which means that you have to have more connections. And it is also like the startup founders’ or startups’ responsibility to educate policymakers as well. So we need to think vice versa. I think maybe in the next five years, if you are working on AI, if it’s an AI startup, we should really start thinking about hiring a public police officer, maybe even before hiring an HR officer. So otherwise that roadblocks actually may prevent technology a lot.

Sara Kehaulani Goo: Wow.

Rishi Khosla: Can I pick up on a couple of those points? So I think on the bootstrapping point, the first business that my co-founder and I created was purely bootstrapped. We used $60,000 to build it into a 3000 person company. And the discipline that teaches you, like you say in terms of just business model, in terms of making sure you’re always doing the right thing in a way, is just dramatically different to, like you say, overwatering the business. Right. And the overwatering, I mean, we know very clearly from the last five, six, seven years that the, I mean, as some people have termed it, there was a fiscal easing rate, a monetary easing off the whole private capital market obviously caused by one very large fund and a number of other funds, which sort of went in proliferated after that. And that created a lot of, in a way funding of and spending of sort of money, which wasn’t necessarily anchored in a true business model. Right.

And clearly for that exuberance to being taken out of the market, our view is that’s good, it’s healthy. Right. I’ve been speaking to a couple of other sort of founders of businesses which are arguably larger than us, but similar stage in terms of maturity. And their view as well is that you sort of got the annoying ankle biting competition who didn’t have a model, but were just spending a lot of money to try to go and acquire customers, et cetera, that sort of paired away and therefore there’s more money to put in products, there’s more money to actually put in the right things which deliver a better experience for the customer. So that’s the first thing. And then on the other point, on the regulator, your point about having a policy officer, et cetera.

So again, like you we’re a fully regulated bank in the UK and from day one, again, we over invested in actually that regulatory relationship. Right. So we actually had someone to manage our relationship from that world almost from the get go. Right. And actually doing that and understanding where the regulation’s concerns are, the way that we think about it now is that 99% of regulation is aligned with creating the right long-term business. Right.

So if you work backwards, not from the rule book, but you work back from the spirit and the principle of which a regulator’s thinking through, how do you do the best for your customer? Is that a good thing for a business or not? I’d argue yes. Right. When it comes to financial services, how do you make sure you have the right liquidity, you have the right capital, are those good things for longevity of a financial business? You think so.

Sara Kehaulani Goo: Right.

Rishi Khosla: Right. So therefore, if you almost align yourselves with actually what’s the regulator’s mindset about what is the right long-term outcomes for communities, and you just think to yourself, how does that align with actually what would make a long-term success for your business? There is so much alignment there, and that’s the way we’ve always thought about it. And in a way, I say to our teams that sometimes when we interact with a regulator and they give us feedback, we should view it as good consulting advice. Right. And actually it makes our business stronger.

Sara Kehaulani Goo: Yeah.

Rishi Khosla: Right. So it’s just a different way again, to sort of think about these things.

Jacqueline Poh: I think that regulation is one of those interesting things. I’m very fascinated by your title, “No Rain, No Flowers,” because I thought about it a bit more and I’m like, here’s the assumption that all you need for flowers is rain. You don’t. You actually need sunshine, you need oxygen, you need carbon dioxide, you need a fertile soil, you need a whole range of things for companies to flourish and to be productive and to be profitable, make money, to be honest. And so it’s the whole environment, and regulation is one of these big pieces. Actually, you need a whole bunch of things and, because Singapore’s been trying to build this startup ecosystem for decades now. So each one of these little pieces, it comes into play. So not just the capital piece, which we’ve been speaking about, but the regulatory piece, the talent piece, the enabling environment piece, and the demand piece.

So these are the key pieces that you need, which is the soil. The regulation works really well when there is a less disruptive new technology. The minute there’s a new disruptive technology, regulators don’t really know what to do. They don’t know what to make of it. And when you have something like generative AI or precision medicine or quantum, I’ve been a regulator and you get stuck. You do get stuck.

Rishi Khosla: Because it’s not siloed.

Jacqueline Poh: It’s not siloed and all regulations are built for the last war, the last problem. So I do support the idea of sandboxes. And I think in any good enabling environment for entrepreneurship, you have to be very, very conscious about identifying the disruptive technology. In our case, it was alternative proteins. In our case, it was FinTech, it was Web3, now it’s AI. And create a little bit of a sandbox around that so that your entrepreneurs actually have room to play.

We are very cautious about over regulating. We’re particularly cautious about over regulating AI because that might kill use cases that have yet to be born and we don’t want them to be stillborn or we’re stillborn somewhere else where the regulation is more fruitful. But you need other things for this ecosystem. You need talent. You need to be able to bring in the best engineers, if that’s the case. You need to grow them indigenously, if that’s the case. You need a kind of funding environment that allows debt, that allows family offices, that allows angel investors to step in.

You need the kind of demand that is very fruitful for your company. So like in a B2C market, it’s a big consumer market. In a B2B market, it means that you do need all the companies present that can help you scale your company. And in this regard, the rain may not be venture capital, it may be venture procurement. And if you find that company that’s willing to buy from you something new and innovative on a venture procurement basis, that’s even more productive. So I’m just suggesting that the rain is not all you need.

Sara Kehaulani Goo: Yeah.

Navin Chaddha: Yeah. So I think, let me give you the history perspective on what happened in the United States because it’s getting repeated in other regions. So if you go back to 1960s, 1970s, I wasn’t also born at that time, but having Mayfield been around there, great companies like Intel, there was no venture capital industry. Arthur Rock essentially was in New York. He used to go to high net worth individuals, sell the business pitches and got the business going. Mayfield’s first fund, $2 million in 1969, and probably like 10 iconic companies came out of that. A lot of it was syndicated with other people and the best companies, Microsoft to a point, you mentioned Oracle, Apple. VCs didn’t give them money initially. It was too speculative like the AI wave is. They didn’t understand what it is. They got money from friends and family.

Even if you go to this date, Google, Yahoo existed, Excite existed. Every Silicon Valley firm passed on their seed and inception round, and then the first one or 2 million came from angels and they all became billionaires. When Google became the largest search engine, VCs came into it. They hadn’t figured out monetization. So I have seen this movie with all the iconic companies till the internet happened and after that, even Facebook didn’t raise money till it had gone to all the campuses when one of our peers invested in them. So my feeling is bootstrapping, getting customer validation, getting the customer to pay and establishing product market fit without any venture money. I would encourage every entrepreneur to go do that, right? And a lot of our companies, I would say 50 companies, which had IPO, didn’t get money at the first time. Twitter didn’t get money, right? Genentech, when the idea was there, like people look at it and say, you’ll discover biotech.

Sara Kehaulani Goo: Yeah.

Navin Chaddha: So I think whenever the idea is the crazy, it doesn’t get money, right, like basically whether it was Airbnb, all these companies, when you are the next, next thing. So I won’t get discouraged. Right. The bootstrapping point you have met and friends and family is a great way because everybody cannot, as a VC, predict where the future is. You just can’t.

Sara Kehaulani Goo: Right.

Navin Chaddha: Right. The other thing on the regulator stuff, I would say, it’s a healthy balance. So since the topics here have been genAI, my only request to policy makers and regulators is there’s a lot of fear around GenAi. Will it take jobs? What happens to trust, safety, privacy, security? Can we all get together as public and private and agree on the guardrails and norms and then let’s just go implement that, right? So if we can do that, please tell us all the fears and we’ll operate within it. Because many regulated industries, financial services, healthcare, we’re already doing it, right? Entrepreneurs are smart. You put constraints, figure out a solution. So I think time has come to lean forward and collaborate with private companies, financial investors, any kind, public policymakers and say, come on, let’s sit down. What are you afraid of?

Sara Kehaulani Goo: Yeah.

Navin Chaddha: Let’s find the solution.

Sara Kehaulani Goo: And maybe that could be our conversation for next year because I don’t think this issue is going to go away. Thank you so much. We have to leave it here. This has been an incredible conversation. I just want to thank all my panelists for the perspective you bring from the region you’re representing. It’s been fascinating. I’ve learned a ton. I hope you have too. And thank you to our panelists. Please join me in thanking them.