The Exchanges

Q a war. And so as a result, like, you just want to be like, no, like this is, we're now fighters. AI's are new weapons. And like, we're fighting over the same territory. We're just doing it with new tools. And I think that's just like a different mentality where like, in a fight, you're like, I'm fighting Bill. I'm fighting Cloudflare versus like, no, it's all new, you know?

A It, it really depends on who. For, for startups, we're, we're hardly ever, for AI startups, we're hardly ever fighting Cloudflare. So for example, uh, uh, just to use the example from the election, Polymarket, it was always, it was born on, in Vercel. Uh, for Washington Post, we're fighting Akamai. And for Minnesota Star Tribune, we fought on-prem. And they moved off of the legacy. Into the new homestead land of Vercel. So it depends on who you're talking to. I think for enterprise, you're absolutely right. Like when you get into like deep enterprise, everyone's fighting for your workload and like the big banks and, and, and, and the big insurance companies, they have 20 vendors courting them at any given point and it's more of a knife fight.

AI assessment note: “It, it really depends on who. For, for startups, we're, we're hardly ever”

Q Well, and I saw like your, your growth rate is also crazy, right? Like, is this all because of the agent swarm?

A Yeah, it's because it's that same intuition, right? That like humans, uh, Vercel's growth was capped by how many humans exist and how many times they deploy by submitting PRs. If now coding agents start taking that job of writing the code and deploying it and testing it and submitting PRs, you know, GitHub also published their data and it's insane. Now, all of a sudden, our infrastructure has like a hundred times the demand. Our signups are growing by 50% month over month. My high school friend, who's never written a line of code in their life, came to her office yesterday, and he's like, by the way, I, I've been following you, obviously, and what you've been doing for years, and in Argentina, there's always news and whatever, but I never understood what Vercel was, but now Claude introduced me to Vercel. And so there's this new, like agentic driven interaction, and I fully expect agents to be signing up for services directly, zero human intervention soon, making payments on behalf of people imminently, you know, getting credit cards and budgets and solving entire problems for, for people where the service provider becomes an, a vendor of the agent, not a vendor of the human.

AI assessment note: “Yeah, it's because it's that same intuition, right?”

Q And to put this in context, that tweet was what, less than 18 months ago? And remind people what the tweet was.

A Yeah, it was this idea that generating software, it was in competition now with Googling for software, but now generating software is not just in competition, it's going to win. It means that everyone can produce their own software for personal use or for enterprise use, because why would you go through the hard work of procuring a new vendor if you've already procured Vercel, you plug it into Codex or Cloud Code, and now you generated this software that you thought you were going to procure. Like, this is literally happening as we speak. What makes Vercel really good for this hosting solution for generative software? One of the things that we bet on was serverless. Serverless has this property of scale to zero. A lot of software will be so spontaneously created and used that it might just get intense use for five minutes, for two weeks, maybe a year. Maybe it's just for me. Maybe it's for a medium-sized group. Maybe it takes off on X. Like literally every day there is a new, a generative personal app that takes over the world on X. There's one called the world monitor, which is vibe coded Palantir. It literally became one of the top 100 host names on Vercel for a week because everyone in the world was talking about it. And so.

AI assessment note: “it was this idea that generating software, it was in competition now with Googling”

Q And I think there's some marginal advantage to it. But in my mind, that's like a five to 10% problem of like more great businesses, not like the 90% problem. That doesn't mean it's not worth doing. But like, it does mean like, to me, it fits very nicely into trajectory of like, yes, it's much easier to start all sorts of businesses for all sorts of technical reasons, right?

A Yeah, I think, uh, my point of view is, Generating software is actually the killer app of AI. I was interviewing Nat Freeman today for our internal Vercel podcast today, and he was talking about how much conviction Microsoft got into OpenAI because of the effectiveness of Copilot. Like when Satya saw Copilot, when, when they were doing the initial iterations, they had actually way more ambitious applications they wanted to create with AI. And the one that ended up being the killer app is Making people more productive at producing code. And what's happening now is that because the models are getting so much better, we're finding out that we need to make the models better for producing code. And so you use something like, uh, uh, uh, Claude and, and which is actually one of the main models that powers V zero. And it became like 10 times better at designing software all of a sudden. It became 10 times better at creating React applications. And so let's just look at the numbers, right? Like there's four million React developers. I can move that from four million to fifty million trivially, because now like you just speak English to that thing. Even if like it doesn't do a hundred percent of the job, it's just like this like infinitely available assistant.

AI assessment note: “my point of view is, Generating software is actually the killer app of AI.”

Q Like, why does this need to exist? Like, why and what are you, what are you fundamentally leveraging that all the other ways that people build websites today, um, are missing?

A Yeah, it really depends on the size of the customer. For e-commerce websites, for example, we just give them a lot of new revenue because we make their website so fast that you put it on Vercel and you get better SEO, you sell more stuff, people convert at a higher rate. So for example, Under Armour.com, Ransom Vercel, ASICS, like when you buy shoes, Vercel is behind the scenes, hopefully you have the best experiences on the internet and most visitors on the internet won't know. But they will like that website more because it feels better. So when, when think of it as like a global acceleration system, not only just the way to build it. Um, and the other reason that it's needed is that, uh, especially with, with the advent of AI and code generation, Versailles is almost becoming more like a design tool is that it's just so easy to create software these days.

AI assessment note: “think of it as like a global acceleration system, not only just the way”

Q this stuff are going to come down, you know, building software is nice. I do it all the time and like people will build a lot more of it, but if the marginal cost of building it goes towards zero, And like, that's the general alternative history. My big concern for your company will be, you're not going to get paid enough for it, right? Because like, it'll be commodity.

A Well, you still need the infrastructure to run it, right? So it's not just about like writing the code, but if you have, if you merge writing the code, if you merge a great experience for designing, if you, if you put in a lot of the taste that I think is really important into the process from the creator point of view, because AI mostly generates a slop. That's something I agree with you actually like, We do have to hold AI to an insanely high bar because there's way too much spending and way too much hype, and a lot of it is the slop. It's just like, a lot of it looks horrible. Uh, and one of the things that actually we, we've differentiated in is, uh, the vast majority of generations we produce, uh, embed our taste into how we would design software. They embed even our, like, sort of like, uh, design system component library, and from there on, we allow you to customize things. So you have to find that Sort of balance between it has to be really easy, but also it has to be high quality, but also customizable. And I do think that, um, making something 10 times or a hundred times more abundant will, will have very, very, very, uh, deep, deep, uh, implications to the industry.

AI assessment note: “Well, you still need the infrastructure to run it, right?”