Realistically, on a scale beyond very few graphs, it’s pretty likely some sort of software is going to be used to process data into markup. At which point, if the desired result is static markup, it doesn’t really matter (to the end user) whether that software is written in JS or some other language. That’s a toolchain concern. It probably doesn’t matter much if at all (to the end user) whether the resulting markup is HTML or SVG. There may be some imperceptible performance difference, or slightly different accessibility requirements. But neither requires users to run JS.
If I were evaluating this choice, it would almost certainly come down to what’s most familiar, or what most readily integrates with other parts of the project’s stack and/or data sources. And I could easily imagine even that evaluation being a wash.