Ep 61: High School Dropout to Billion-Dollar Startup; How Guillermo Rauch Built Vercel

0:00For each hundred milliseconds that you slow down an e-commerce experience, you have a 1 % drop off in conversion. And we've all experienced this when you're like on Instagram and you go to a product, you need to buy it and everything has to click. For every 10th of a second, you just slowly lose people. And it's not just how fast the text loads, it's how fast it's interactive. You have to feel like you're clicking and it's responding. Yeah, it's a credit card auto-completed automatically. So our obsession is giving people at all these great companies and institutions the best possible tools to create great experiences for the customers.

0:42Today we're talking to my friend Guillermo. Guillermo is the CEO of Vercel. A huge amount of the web is built using the frameworks that Vercel has created that Guillermo and his team have created. He has a great story. He's from Argentina. He started helping people all over the world, even as a 13-year-old, and just really extraordinary talent himself. Websites like Washington Post, Under Armour, Uber, Mr. Beast all use Guillermo. He's a superstar. Excited to chat with him today. Guillermo, thanks for joining us today. Thank you. Glad to be here. Let's start off with your background. You grew up in Argentina.

1:12Yeah. I grew up in the outskirts of Buenos Aires. So kind of an hour away from the city and I would buzz in every day to go to high school and, uh, less than an hour. Yeah. Wow. Yeah. It was, uh, it was an interesting area to grow up in. Not, not a lot of computers in homes. Like my dad was the, the guy that was always all about the future of software. He had a vision for software is going to be a huge thing. What did he do? He was an industrial engineer. My mom was a chemical chemistry engineer and they both came from a fairly poor families, but they were very proud that by studying engineering in particular, they had an out and that was their, their thing.

1:56So grew up in a fairly poor area. Were there parents from Argentina? So side of my family was from Italy, But, um, so my grandma's parents were from Calabria and they funny enough met in Argentina. So they both emigrated, uh, uh, in a ship. Uh, and then from my dad's side, German and a couple other things. Um, so a lot of folks in Buenos Aires have this diverse, uh, immigrant backgrounds as well. And what, what led you to America? What led you to come to the United States? So my fondest memory is I was helping my mom out with her resume. when I was a kid, because I was all about, she needs to have her own online resume.

2:37I was like making websites for different things. And I remember how proud she was of the American multinational company that she worked in, in Argentina. It was called Fisher Rose Mount, a subdivision of Emerson Process Management. And she loved being a sales engineer for that company. And the things that she would tell me are like, they just do things better. They have better rigor for their engineering. They have better specs. It's funny. A lot of things change over the decades and a lot of things stay the same. Like a lot of what we do at Vercel, we take a lot of pride in the quality of our documentation.

3:16And I think back in the day, she was like basically telling me the same thing. Like they're documenting things great. And like, she just had a lot of admiration for, for that. And she would always tell me, if you just learn to speak English, that'll be a huge advantage for you. I see it at my own company in Argentina, like the people that can speak English and do commerce and do customer support or help out a customer and like break that barrier. So that was the thing that if I look back, what are the values that I was attracted to early on in my life? I've always been attracted to great companies, great industry, great engineering.

3:58And that was kind of the thing I got exposed to early on. And how old were you when you came to the U.S.? So my first visit was, I was about 18 years old. So that was over 12 years ago. Yeah. But then later on, it took me a while. We can get into the topic of visas. It took me a while to make the case to the immigration authorities of, hey, I have some skills that could be valuable for the country because I think the thing that's different about my story is I became successful as an open source engineer. Yeah. So my background is front engineering. I made a asymmetrical bet early on that front engineering was going to be more important than back-end engineering and that the language that the browser could run, which still to this date is only JavaScript, now there's only web assembly but i was like okay there's an asymmetry there like you could learn any back-end language but the browser can only run javascript and that's the language that you have to use if you want to do some really cool things like animations like real-time synchronization which i became obsessed with like chat applications gmail chat google docs google wave all that kind of thing so very early at a very early age i was like okay i love all kinds of programming and computer stuff, but I mean, I specialize in that.

5:19And I started getting jobs and all kinds of crazy opportunities. As a teenager? I was 13 and I had a, I was about 13 and I had a customer in the Netherlands that was running a makeup company. He was an entrepreneur. He was running a bunch of things. He was really well connected. So he had a fashion business. He had a diet, dieting business. It was kind of like a Weight Watchers of the Netherlands. and I was helping him out, like build out all these pretty innovative websites. I remember actually I built one of the first really interactive, uh, browser first client side, uh, calorie counters. You could go in and say like, what foods have you eaten?

6:00What exercise have you done? And then calculate the calorie output and all that kind of stuff. And I was like 13, 14. Like I started like not paying attention to high school. That was like my thing. And I always tell the story of like, I was very studios. So I really believed in the power of academia because my parents were like, oh, we started engineering and that's why we made it. And on the other hand, I was like, okay, I love studying, but I have all these opportunities to work with people from countries like the Netherlands. then I went at one point I had a client in Boston then I got this one client in in San Diego who was contracting for a company in Switzerland and that was the thing that really got me out of Argentina I was 17 and the job sounded amazing they invited me to go to Lausanne Switzerland which was like that was my first time out of the country it's like okay you're gonna get deployed to Lausanne which is not even easy to get to and the CEO of the company will pick you up at the train station and you're going to do this like really awesome, interesting engineering project for them.

7:02That's awesome. So it came out when you just started working. Yeah. So I started paying less attention to high school over time, more attention to all my freelance projects. You didn't need to go to college or anything. You just know. And that was the thing I made my visa case so hard for the U S because we always value college versus not, but actually you can be more advanced than the kids in college. This is, this is a problem with, we say we want to copy the Canadian model, but we shouldn't over-index on these markers. Peter Thiel famously thinks a lot of people should do what you do, actually.

7:29Yeah. And I always am cautious of, I don't want to be extreme because my case is somewhat extreme. I had sleepless nights in high school because I was awake for Netherlands hours a lot and things like that. But on the other hand, I just developed so many skills early on. I was making money for my family. I was actually at some point when we still lived in the outskirts of the city of Buenos Aires and weren't doing so well, I've loaned dollars to my parents. Um, Argentinian currency has always been bad. So I, we were benefiting from that ratio early on. Um, but yeah, like I, I self self-teaching was, was definitely, uh, valuable for me.

8:10Do you, do you think your parents would have been more successful if they were born in the United States, given their frameworks? Oh yeah, absolutely. Absolutely. Do they ever consider moving or their, their, their family's all there? I think so. Yeah. They, they still, the one thing is they always talk about is the language barrier, but I always tell them like, Hey, in California, like you, you can go so far and you can learn as well. So that's awesome. That's amazing background. Obviously they gave you a lot of things led you to succeed. Tell the audience a little bit about open source a little bit, because people don't always know what open source is, why people do it.

8:39You, you said you were involved in a lot of open source projects. What was that? Yeah. Open source is the best thing that's ever happened to me. So now it's super fashionable for enterprise open source first companies. turned into a business strategy now, whereas it used to just be this like movement, right? But for me, literally it came down to, it was the most open platform for learning in the world. I remember I always tell this story. I was trying to compile a C program on Windows and just setting up the tools and accessing information was impossible. Oh, it's so hard. I hated this stuff. There was this compiler called DJGPP of one like brave soul that was trying to create an open source free compiler.

9:20Everything was gated behind a paywall. The information was hard to access. It wasn't even digital at times. Like you'd have to buy books. And when I switched to Linux is when like, I think my developer tools career was like on an unstoppable journey at that point, because I found that installing the operating system already came with the basic tools necessary for you to create things, not just consume, install, pay for things. It came with the tooling to create. And I remember the day that I compiled my first Hello World program, I had already done a lot of things with programming, but just having that raw access to the machine and knowing that all my tools were free and open source was absolutely main belonging.

10:06And so you started contributing to these projects and helping them get better. Yeah. The other thing about open source that is amazing, it's not just, oh, you're getting the tools for free. There's this global community of people that are like teaching each other the best practices. You can Google any error message. As if those weren't substantial enough benefits, the other thing that it has is the ability to fork. The ability to take code that other people have written and incorporate, fork or integrate. You can take some component that maybe the author didn't think could have become a component.

10:38So open source has this fractal nature that is unstoppable because you can take anything and layer in more additions and improvements or use it as an implementation detail. The first thing that I used in that fashion was I used to run the online forum for high school. And I took a piece of open source code, it's called PHPBB, and I just modified it. And because I was obsessed with real-time, I added that real-time chat capability because I thought forums were too slow at the time. And that idea of like, your starting point is not a blank slate. You can piggyback off of many other things. And Naval has this quote that I really like that is, open source means that every problem only has to be solved once.

11:22And sure, there will be alternatives and things like that, but a big part of the world today runs on Linux. The efforts that go into creating that kernel, they only had to be done wants. So I think open source has this tremendous economic efficiency and power for the other thing I love is, which is a cliche, but I love sapience and the concept of cooperation at scale and cooperation in a way that is very energetically efficient. So when I look at open source, I find that it's a system for folks that are in Argentina and the US, in India, in China, in Brazil to cooperate. They're all collaborating.

12:06Yeah. And it's incredible. It's a very positive part of the ecosystem. So tell us a little bit about Vercel. What does Vercel do? Who are your customers? For sure. So we create a web infrastructure. That means we give you the tools and platforms for you to develop, preview, and scale any initiative on the web. So we have customers like underarmor.com, which they sell sneakers, or we have customers like Washington Post. What we've solved, which is a hard engineering problem, is how can you create applications that can scale to like billions and billions of traffic, but also be extremely dynamic in nature.

12:41So you can personalize, customize, and deliver them really, really, really fast. So the wisdom that gets shared in the industry a lot is for each hundred milliseconds that you slow down an e-commerce experience, you have a 1 % drop off in conversion. And we've all experienced this when you're like on Instagram and you go to a product, you need to buy it and everything has to click. For every tenth of a second, you just slowly lose people. And it's not just how fast the text loads, it's how fast it's interactive. You have to feel like you're clicking and it's responding. Is the credit card auto-completed automatically?

13:17So our obsession is giving people at all these great companies and institutions the best possible tools to create great experiences for the customers. And how big is the company now? Like what part of the internet is it running? So we have an open source framework that is the foundation of all of this. It's called Next.js and it runs already a very significant part of the largest websites on the internet. Like Walmart.com entirely runs on Next.js. Mobile TikTok runs entirely on Next.js on the web. So anytime you're using the web, there's a very good chance that Next.js and Vercel are behind the scenes.

13:54And when Next.js is there, how does Vercel make money on that? Yeah. So our main principle is our tools should be learned by every developer in the world that is interested in publishing to the open web platform. And to my origin story, I hated that I couldn't, like there was an obstacle for me to learn how to build. So Next.js is open source. Now what companies really struggle with is the infrastructure to deploy to the cloud, right? Like we've all had the experience when you go to the website and it's slow or it goes down during Black Friday. So Vercel comes in and gives you this global platform to deploy your Next.js projects and other kinds of projects as well.

14:34And it's custom-tator to the problem of front-end. So what does front-end mean? When you go to any website, you're experiencing the front-end first. The back-end could be all the servers and infrastructure and APIs and things. But what matters the most these days is how good of a brand experience you have, how fast it loads. So we empower companies to make a difference on that part of the stack. So Guillermo, Mr. Beast, I think he's like the most popular YouTuber on the internet. He's using Vercel. Tell us about him. What's going on? Yeah. So he's an incredibly smart guy that has just reached a hundred million subscribers.

15:17And one thing that our platform Vercel is known for is great scalability and there are lots and lots of traffic. So Mr. Beast was going to do a drop of a t-shirt, I believe it was, when he hit 100 million subscribers. And he did it on Vercel. And he created this incredible experience that combined games inside the browser and real-time elements. And the drop, of course, generated millions and millions in revenue. And the platform just scaled flawlessly. What kind of games? It's really funny. It's like there's a Lambo game where you're Mr. Beast and you have to hop over Lamborghinis. And another one where you slap him.

15:57and the one that whoever accumulates the most slaps wins. I think it's a reflection of what e-commerce is becoming. It's all about games and drops and exclusivity and crowds of people. Jumping over Lamborghinis and slapping Mr. Beast. This stuff is definitely above my level, but it's amazing how well that works. Yeah, check it out, shopmisterbeast.com. It's on Vercel. And so Vercel's origin story, it sounds like it's tied to you being frustrated with the fact that you couldn't learn how to build. And so without the gap in the marketplace, you want everyone to learn how to build and then do it more easily?

16:29Yeah. The most concrete and memorable story that I have is... So I sold my previous company to a company called Automatic that runs WordPress.com. I was very inspired. I've always loved that idea of an open source project that can change people's lives in running a business as well. And I remember I wanted to create the landing page for my startup. I wanted to create it and have my logo and an email input for people to submit and get updates about my startup. And the other thing that I want to do is I want to use the absolute best practices available at the time. So at the time, Meta or Facebook had open sourced their technology for creating UI at scale.

17:14Like we're talking about Facebook literally open source, the engine that powered the like button, the newsfeed, messenger. What was that called again? React. And I was like, holy crap. It's open source. It's a liberal license. I mean, I use that. And then on the infrastructure side, Google, another usual suspect of great technology, had open source Kubernetes. Yep. Which was the technology that empowered the orchestration of infrastructure. So I was like sitting down there. I'm a fairly experienced engineer. I sold the company. I'm in Silicon Valley. I have a lot of time on my hands. I'm going to deploy my web application.

17:51And it literally took me weeks to, on one hand, configure React, this engine for UI. And I realized I was basically having to build my own framework from scratch. And then once I went through all that trouble, and we're talking weeks for a single page, I had to stand up infrastructure. And what I was trying to do is use the best possible available technology. I knew that it could have gone to Squarespace and maybe in two clicks I would have had a website. But I wanted to build something durable, sophisticated, dynamic. But the best technology is just really hard to use. And that's where the inspiration for Vercel came because I was like, okay, every company in the world is going to want to build on top of these technologies, but zero of them are going to want to spend the time and effort on anything that's not their product.

18:38I was trying to get a page online and that was way too difficult. And so you went also with the open source route, but just redesigned it from scratch. Yes. It made it easier. Are there any unexpected network effects you found with the open source thing or any surprises to you? Oh my God. So because Next.js uses this API called React. So the innovation there is that you can model the entire world as components. When you talk to a programmer, they tell you, well, I model my world as functions, I model it as strings, methods. I remember Steve Jobs at one point was excited about object-oriented programming, objects.

19:17Yeah, objects, yeah. And it turns out that none of those abstractions really helped build products. So when you work at Airbnb and you're experimenting, you're creating a new experience, you have to bring in reusable components from almost like a bookshelf of ingredients. And you say, okay, I'm going to bring in the carousel that shows the photos for the property. I'm going to bring in the buy button. And what's important here is that when your company invests in these things, like companies like Amazon have perfected the buy button. They've iterated, tested, they've instrumented it with events and observability and metrics.

19:54They know what shade of yellow converts the best. So you want to reuse all those things. So that's what this component-oriented programming methodology enabled. And what's happened is that because our technology is open source, now there's an entire ecosystem of reusable components, which means back to that idea of even in Lanús in Argentina, I didn't want to start from scratch. No company in the world will ever have to start from scratch when they have to build a production-grade application. So you can use components for data visualization in real time. You can use components for chat, for layouts, for menus, and all kinds of things.

20:31So it's way easier to build a web interface now than it was five or 10 years ago. Yes. And one that scales like Google's and Meta's technology does, which is a big problem for a lot of American enterprises, right? Like Walmart did not have the luxury of inventing the cloud like Amazon AWS did. So how do you bridge that gap between all the companies that have to rush to become digital first, especially after the pandemic? So how do you think about it? And there must be a tough line between being a community first company and fully monetizing the opportunity, right? And you obviously don't want to do things that grow the community long-term, but you want us to make money.

21:07Like, how do you make those trade-offs or what trade-offs are there? For sure. I'm very, very passionate about this idea of education. So one principle that we have is we have to have developers learn our methods and techniques and basically have no barrier to entry there. So we give away a lot of resources for free because we know that the engineer that's working on a side project to evaluate the technology will bring it to a large enterprise the next Monday. And this has actually happened in ways that are quite fascinating to me. So I like to tell a story of therealreal.com, really big enterprise public company on the e-commerce side.

21:47the CTO was testing out our platform on a weekend with his personal email address on our platform. That's cool. If I had added a lot more friction to that process and like, oh, there's no free account, you can't deploy more than end times. He might not have tried it. He might not have tried it. And so we strike that balance between like there's an open source component, which is almost like the client. And then there's a cloud platform that is like the server. And really where Vercel makes its money is by providing high quality services. We battle the attacks. We secure your data and your code.

22:20We scale it. So anyone who wants to can design and build in the language, but then you help them in infrastructure. And so let's talk about the big picture. One of your things you've said in your goal is you help people go from global idea to global app in seconds. Sounds a little bit fast to me. But what is... We actually do it. Seconds, really? Yeah. How is this going to change the web in the next decade? It's amazing because I was recently back in Argentina for a trip. and like many other hackers, I love this website called Hacker News. And Hacker News grew out of some of the old hackers in Silicon Valley and it's hosted in US West.

22:58So it's hosted in California or about there. And I was in so much pain experiencing that website and so many other American websites that I would visit on a daily basis. It's just slow for you from there. Because it turns out the speed of light is not that fast. that's a fundamental problem in the world like if you're in Argentina if you're in Brazil if you're in Europe and the workload is not right next to you you're going to experience this load and there's just no way to overcome it if you go the speed of light from Argentina to California and back is that like a fourth of a second or what is that it's seconds and the problem with this is that yeah for example to set up a secure TLS it's not just the light it's the light and everything it has to talk to when it's there that's fair My guess is that in terms of raw milliseconds, it's probably, I'm going to get nuked if I get it wrong, but I would wager it's like 100 and 200 milliseconds, maybe 150 milliseconds.

23:54That's my intuition as well. It's on that order, basically. A fifth of a second. And so there's so many variables. So for example, you have to go through traffic exchanges. So the wiring in Argentina is not in a perfect straight line to Virginia. Of course, yeah. Virginia is where most of the cloud workloads in the world are. That's true. You're not just like shooting to a satellite and then shooting right there. You're going through some other stuff. So there's that layer of interaction. Then all the internet protocols, and some are fundamentally for security reasons, have to go back and forth.

24:21So long story short, it was taking seconds, maybe five, to do a cold read of a news article. Which in the old days would have been fine, but these days it's super annoying. Yeah. And if you're selling something, or if you're... Well, every millisecond is 1 % less people are going to buy it. So five seconds, that's 50 % less purchase. There's also just like annoyance. Like people, and this goes back to why we take so much pride in democratizing access to technologies that Meta or Google have invented. The problem is that if you're selling cameras and you have a great camera company, you are now expected to perform to the same level of Amazon.com or TikTok.com because that's what people are now used to.

25:03People are used to the instant gratification, instant response. There's a web that just performs right away in real time. Yes. Everything has to get pushed to you. That was the other big thing. So one of my proudest open source creations was this framework called Socket.io. I became obsessed with the idea of democratizing the data being bidirectional, which is really obvious. Like a phone call is bidirectional. I talk, it goes to you, you talk back, it goes to me. But on the internet, everything, and still to a great degree these days requires a refresh button. Like not every website has developed the technology that WhatsApp has created such that data can be pushed.

25:40Just push it to you. Yeah. So that's the bar. It's like, I call this negative latency using AI to predict what the customer is going to want and push it to them ahead of time. I love that. I love that. That has to work. You basically figure it out ahead of time. So that's how you saved not only the 200 milliseconds. Like, let's say that with Vercel technology, we will put your workload in Argentina, in Brazil, in Europe. So it's going to be like, you know, 20 milliseconds away from the customer. I love this. So you're actually doing the negative laser thing with Vercel itself for everyone. This is great.

26:10I've thought about this a lot and I always wonder why people haven't done it, but you're making it so it's easy now. Yeah. And this is the biggest distinction between Vercel and virtually everything else is that we're global first. So every other platform has historically asked you to say, where do you want the computer to live? Because they lived in the world of hardware, of data centers. But your computers live everywhere. Our computers are absolutely everywhere. Antarctica? Soon. Soon. Mars? When necessary. Mars when necessary. You're on your six continents then. Yeah. Well, we're upwards of 20 regions now.