When 2025 rolls around, I can see the following:

1. Foremost, HTML Modules allow importing HTML files natively, that's key. New components are more stable, more accessible, and widely supported. Web Components are more stable than ever.

2. The most popular features of Sass/Scss are added into core CSS by this point

3. With broad support of features like :has, @scope, @property, subgrid, anchor positioning, MPA view transitions, and lots more

4. Modern JS is stable, most new versions are incremental in the 2020s.

It’s already a reality, I was doing it about ten years ago.

require.js was a big piece of the puzzle, allowing the use of modules in browser without a build step.

The only time a “build” happened was to concat/minify scripts for production release, but even this wasn’t strictly necessary.

There were other pieces involved, but we had a powerful stack (aforementioned modules, css preprocessing, reactive templating) that had a faster and easier workflow than anything happening today.

I mean, you can do this today right now. The "problem" is that many people rely on frameworks. Frameworks have dependencies, so you need something to manage them. Then you need a way to make sure that you have no namespace collisions.

You probably want to optimize your your build to not include code from libraries that you aren't using.

Boom, you've recreated modern build tools.

I think the reason is because employers are putting in various requirements for a job. 'Must be proficient with React' or 'Must have experience with $FrameworkOfTheDay' is a common occurrence. Really, all you need experience with is HTML+CSS+JS/jQuery and maybe PHP for back-end stuff. You can build anything with that stack.