Jeanne DeWitt Grosser on Vercel’s ambition to power the infrastructure behind the web and the agentic applications taking shape on top of it.

A founder can now describe an idea to an AI tool and, in hours, have the beginnings of a working application. But getting from “it works” to software that is fast, secure, reliable, and capable of serving millions of users is another challenge entirely.

That gap is part of what makes the infrastructure layer around AI so compelling to Geodesic. As software shifts from applications built primarily by people to systems increasingly built, operated, and improved by agents, developers need a new set of tools beneath the surface—tools designed for the agentic era rather than retrofitted from the cloud era.

Vercel is building for that transition. The company already sits behind many of the digital experiences people use every day, powering everything from consumer websites to parts of AI applications and agent infrastructure. And its ambitions go much further. As COO Jeanne DeWitt Grosser puts it, Vercel wants to become the underlying layer that helps power “every pixel” on the internet and “every token” that runs through an agent.

Jeanne brings an unusually valuable vantage point to that mission. Before joining Vercel, she helped scale Google and Stripe through periods when entirely new technology platforms were reshaping how businesses operated. Now she is applying those lessons to what she sees as the next major computing era.

In our latest Geodesic Spotlight, Jeanne explains why AI could dramatically expand who is able to become a founder, what “self-driving infrastructure” might look like in practice, and why the winners of the cloud era are not guaranteed to lead the agentic era. She also shares why Vercel is making a deeper investment in Japan—and how its partnership with Geodesic is already helping turn that strategy into enterprise relationships and new business.

Your career has taken you from Google to Stripe and now Vercel, helping scale companies that have transformed how people work and build businesses. Did you ever envision this path, and was there a defining moment that inspired you to pursue it?

JDW: Probably not when I was 22 and first working at Google. But I think pretty early on—my father had been an executive in the technology sector, and I got to see all sorts of high-performing executives at Google—I aspired to go be one of those people.

In my early 20s, whether or not I would have taken this exact path, I don’t think I could have predicted that then. But I was definitely pretty thoughtful about who I followed at multiple points in my career. That’s what led me to Stripe: following a leader, Claire Johnson, who I thought was incredibly impressive and who I wanted to learn from. She tended to build orgs that had the types of cultures that people thrived in.

Then at Stripe, I learned the value of picking companies that are playing in really large markets with highly innovative, market-leading products. So as I was looking for the next home, that led me to Vercel and wanting to partner with a real technology visionary in the industry in Guillermo. And so here I am.

For someone who isn’t a developer, Vercel might be a company they already interact with every day without realizing it. Can you walk us through a real-world example of how your technology powers something that people actually use or experience?

JDW: Absolutely. That’s easy enough. Most people are probably touching Vercel every single day because we power so many of the applications, and now agents, on the internet. But Vercel is under the hood, so you would never know it.

Great examples of this: if you go and buy something from Under Armour, Vercel is powering that site. If you go and use Muse or go to Meta AI, we are powering parts of their agentic infrastructure. If you go and use some brand-new startup product, there’s a very good chance that Vercel is under the hood of that.

So I like to think that, if we’re achieving our goals here at Vercel, in the fullness of time, every pixel on the internet and every token that runs through an agent will have been facilitated by Vercel under the hood.

Is there anything else you’d add to help explain Vercel in simple, non-technical terms?

JDW: Sure. In layman’s terms, you can think of Vercel as the AWS of AI. When anybody wants to go out and build an application or an agent, you need all sorts of stuff to power that. You’re typically building what a customer is ultimately going to see and interact with, but in order to make that experience exist, there’s a bunch of infrastructure that has to happen.

During the cloud era, you might have gone to AWS or GCP or Azure—the sort of hyperscalers that everybody knows well. During that era, which was more pixel-oriented, Vercel was sort of like the easy button for that. Instead of an engineer having to go to AWS, take a bunch of different types of infrastructure off the shelf, and do a lot of engineering work to put them together, you could go to Vercel and we would basically do all of that for you. You’d get a really performant, super-fast website as a result.

Now you’re in this agentic era, and the things that you need to build an agent are, in some cases, similar, but in many cases different from the ones you needed to build an app.

Today, if you went to a hyperscaler, a lot of the stuff that’s on the shelf was built for the cloud era. What we’re doing is building the things that you need for the agentic era because we’ve built things like v0, which are some of the largest agentic applications out there. We’ve gone through the pain of seeing that you can’t always easily take something off the shelf.

So what we’re doing is building all the sets of things that ultimately a company will need to build and run these agents at scale.

Vercel has had a presence in Japan for some time, and the company is now making a much bigger commitment to the market. What’s driving that decision, and why do you think Japan is especially well positioned for the agentic era?

JDW: Japan is incredibly well positioned to do well during this agentic era. There are a lot of benefits that Japan can get from AI, given some of the worker shortage that they’ve had. If each employee can be 100x more productive, that’s great for Japan.

It’s also a country that’s exceptional about processes and automation. As an example, one of the topics that’s really hot right now in the industry is software factories. Every single time I talk about a software factory, I can’t help but think about Toyota Production System and the original auto manufacturing factories that Japan built and was best in class at.

I really think this is going to be a boon for Japan, and we’re excited to be a part of it.

Next.js [Vercel’s widely used framework for building modern web applications] is popular in Japan. We have a strong developer community there, and lots of both startups and parts of enterprises are building on Vercel. 

Geodesic’s investment approach centers around connecting companies with opportunities across the U.S. and Japan. How has your partnership with Geodesic supported Vercel’s expansion in Japan, and what’s been particularly valuable about working together?

JDW: Geodesic has been incredibly helpful in Japan. I personally attended Geodesic Forum last year and will absolutely be back at it next year. They’re great at connecting some of the largest organizations and key leaders who can actually make decisions at those organizations with the companies in their portfolio.

As an example, at Geodesic Forum last year, I was able to meet with C-level executives at numerous banks, potential partners of ours, and other large conglomerates in the region.

Even though we haven’t gone big yet in Japan, as a result of Geodesic Forum, we actually already have closed additional enterprise business. I don’t think I can name names quite yet, but one of the larger Japan manufacturing companies is now using Vercel for some of their internal app and agent development. That was a direct result of conversations that we had at Geodesic Forum last year.

AI is changing not only how software gets built, but who has the ability to build it. How do you see this shift creating new opportunities for entrepreneurs and businesses, particularly those without large engineering teams?

JDW: It’s basically dramatically expanding the market of who can be a founder and who can build.

It used to be, if you were someone like me—I have a French degree and an MBA—my only route to becoming a founder would have been finding an awesome technical co-founder who could take my vision and bring that to life.

Now I can really do that soup to nuts. At some point, if I’m trying to build a large software application or a massive agent, I’m probably going to want some engineer somewhere in the mix. But certainly I can easily get to that Series A-plus point with all of what AI is bringing to bear.

Then there are all sorts of other opportunities in a lot of different sectors where someone can easily start a small business in spaces they hadn’t previously. You’re seeing a lot more entrepreneurs getting into digital services, et cetera. So I think basically we’re going to see an explosion in entrepreneurship over the coming years.

You’ve helped scale some of the world’s most influential technology companies, each of them during a period of significant change. What lessons from your experience at Stripe and Google are shaping how you approach this next chapter at Vercel?

JDW: Both Stripe and Google really thought at the scale of the internet. My first job was on Gmail, and you’d have a million new users join Gmail every month, and then every week, and then every day. That was unheard of at the time. 

So you were taught to think in orders of magnitude more scale than most companies thought to think. I think the same is true for Vercel.

Like I said, in the fullness of time, every pixel and every token should touch Vercel in some capacity. If you’re going to do that effectively, you need to be a systems thinker and be able to not only solve the problem in front of you, but think about how to design the solution for the issue you’re experiencing currently in a way that is going to scale meaningfully further.

The good news is, if you’re a systems thinker and somebody who’s process-oriented, that actually lends itself very well to agent building these days.

I find that the way I was taught to think at Google and at Stripe is now setting me up for success in how to think about building an AI-native go-to-market org.

Looking three to five years ahead, what do you believe will fundamentally change about how software is created and experienced? And what are you most excited to help Vercel build or enable as this next chapter unfolds?

JDW: With software, you’re just going to see more and more of it run soup to nuts by agents. As an example, today, agentic engineering has taken off with your Codex, Claude Code, et cetera. But there’s a whole part after the code that’s ripe for additional help from agents, and Vercel is focusing on a lot of this with Vercel Agent.

If you write code, but then you get it into production and it has performance issues, or it has bugs, or it’s working really well but all of a sudden you get attacked, we have all the agent capabilities to help you with that.

So I think you will get to much, much more of what we call self-driving infrastructure. You build the thing—just like you get in a Waymo—and, hey, we’ll take it from there and turn it into world-class software that can scale as far as your business’s capability to acquire customers. I think you’ll just see a lot more of that.

What are you most excited to help Vercel build or enable as this next chapter unfolds?

JDW: What I’ve been having a ton of fun with is building this AI-native go-to-market org.

A lot of what’s incredible about bringing AI to bear is that I actually think it’s disproportionately unleashing human potential.

There’s been all sorts of work that salespeople have had to do throughout their careers that is not what gets us out of bed in the morning, right? People in go-to-market orgs get out of bed in the morning to go talk to customers.

But in between each of those conversations, there are emails that have to be written. Salesforce has to be updated. There’s other prep work that you’ve got to do. So I’ve been really excited to bring AI to bear on all of that, so that you can create a sales org that is fundamentally out there interacting with customers and bringing way more insight to bear because we actually can use AI to put that at your fingertips.

Is there anything else you want to add that I forgot to ask, or anything you want to make sure people get right about Vercel?

JDW: I think we really have audacious ambitions. One of the things I think about, and that just gets me excited for what Vercel is trying to tackle, is that throughout computing there have been different leaders in different compute eras.

Once upon a time you had mainframes, and IBM was the clear leader in that era. Then you switched to client-server, and even though IBM had been the clear leader and was still relevant in client-server, Microsoft really won that era.

Then you went to cloud, and again Microsoft remained relevant, but AWS really led that era. So it’s not a foregone conclusion that the leaders of the last compute era will win the next one.

Now we’re heading into this agentic computing era, where I think Vercel is really trying to develop all of the primitives and agentic capabilities that you’ll want for that.

I like to believe that 10 or 20 years from now, when you look back on this time, you will have called it the agentic computing era. And while many of those players will continue to be extremely relevant, the name you’ll put in the box as the primary winner is Vercel.

Learn more about Vercel at vercel.com.

Note: This interview was lightly edited for readability.