Is serverless a trend that might be replaced by a future technology or programming paradigm, or is it a durable approach to development?

I've noticed more and more teams are moving towards serverless, but at the same time you see a lot of skepticism out there still. You'll definitely see folks trying to make a case that it's a terrible idea and you are going to inevitably hit the wall where you have to move away from it as your product and your system grows. So I guess it's up for debate still in the industry.

I'm biased. Certainly my perspective is that where those limits start to come into play is at a point where you would expect to have the resources to move in whichever direction you need to move, if you need to do something different than what's supported out of the box with Next.js and Vercel. So it buys two, three, four or five years at the start of a project and just completely eliminates all of this friction -- all of this cost of teams to stand up this type of infrastructure and maintain it.

I guess there's also some debate in the industry around whether you can design a system similarly, run it on your own servers and get your costs down dramatically, as you start to pick up a little bit of scale. There will be some questions there, as more and more teams start actually growing on this type of a platform. How soon are they hitting the point where the bills are starting to be like, “Whoa, this isn't going to work”? To be honest, I don't have too much experience actually getting projects to that point. At least I can say that there's plenty of room for a good solid year or two as you're starting new projects to not be worried about costs getting carried away.