Racket is great for some kinds of Web development right now. (Maybe more later, with the development of additional toolkits, and/or maybe a WASM backend after the Chez work.)
I've personally worked on a few industry Web "full-stack" projects in Racket -- including an important large server-heavy system that accomplished something on AWS that no one had yet done before, and a very complex data model HTML5 Offline app (which I did as a mixture of generated and Web services) -- that would not have been possible any other way with the developer resources available. I also used Racket for smaller Web tasks, as a researcher, to whip up a browser-based labeling UI for a large corpus (the corpus was also Web-scraped and massaged by Racket), and to save a big conference demo when the big Java server someone had made could not be deployed on the two-laptop LAN that would be available.
But if a bunch of people hear "Racket is great for Web development" without being more specific, they will run to go look at it, then they will be confused, then they will run at you with pitchforks.
Racket is potentially great for a startup doing a non-cookie-cutter MVP for which they'll have to be clever and invent some things in any case. But Racket is not going to be some variation on some currently-popular framework you already know (maybe it can be backend for one of the frontend frameworks), and you're going to have to write some code to do exactly what you want, not just glue together off-the-shelf components.
I wrote a full feature search engine and web crawler in Racket and it had everything I wanted off the shelf; I was really wowed by how batteries included it was for web development. This was in 2013. It's definitely easier than Node, in my opinion.
I've personally worked on a few industry Web "full-stack" projects in Racket -- including an important large server-heavy system that accomplished something on AWS that no one had yet done before, and a very complex data model HTML5 Offline app (which I did as a mixture of generated and Web services) -- that would not have been possible any other way with the developer resources available. I also used Racket for smaller Web tasks, as a researcher, to whip up a browser-based labeling UI for a large corpus (the corpus was also Web-scraped and massaged by Racket), and to save a big conference demo when the big Java server someone had made could not be deployed on the two-laptop LAN that would be available.
But if a bunch of people hear "Racket is great for Web development" without being more specific, they will run to go look at it, then they will be confused, then they will run at you with pitchforks.
Racket is potentially great for a startup doing a non-cookie-cutter MVP for which they'll have to be clever and invent some things in any case. But Racket is not going to be some variation on some currently-popular framework you already know (maybe it can be backend for one of the frontend frameworks), and you're going to have to write some code to do exactly what you want, not just glue together off-the-shelf components.