I think maybe you're confused. Java drives a significant percentage of Android apps. It absolutely can do modern UI. I can almost guarantee you've interacted with a Java program this year that you never considered.

Pascal is more niche, but it can do modern, too.

Java was doing web clients before the web could and still can. I don't know much about Delphi's web stuff, but I know they've targeted it for years now.

WASM and transpiling blur the lines, too. LVGL can provide beautiful interfaces on the web as well as platforms Electron could never target, and works with any language compatible with the C ABI.

I'm not saying these strategies are without their own warts, but there are other ways to deliver good experiences across platforms with a ~single codebase in a smaller payload. But mostly nobody bothers because they just reach for Electron. It's this era's "nobody ever got fired for picking Intel".

We need more people working with and on alternatives, not just for efficiency but also for the health of the software ecosystem. Google's browser hegemony is feasting. Complexity has become their moat, preventing a fork from being viable without significant resources. Mozilla is off in a corner consuming itself in desperation.

A US-based company holds a monopoly over the free web and a hell of a lot of our non-web software. So maybe let's look for ways to avoid feeding the beast, yes? And we can get more efficient software in the process.