Not to hijack this thread, but curious whether some map people out there might opine on growth areas for this space - technologically, artistically, product-wise.
A map made for a specific purpose will almost always be hand tuned. The first, second...nth draft might mostly or completely auto-generated but the final product (for a good product anyways) will be hand edited. It could be additional graphics, legend or chloropleth tweaking, emphasis or de-emphasis any of the thousands of things you can tweak on a digital document that is also mostly graphical and part art.
So tools for better intent manifestation, even more rapid live visualizations etc. Someone linked an Esri story map video in this thread and that's a good example of state of the art for map generation right now.
There's lots of room for innovations on the data side of things. Making spatial relations to non-spatial data. There's no unified spatial query language that allows arbitrary length depth, lambdas and subqueries. Yes there are spatially aware relational databases but it's a dream of mine to build a universal spatial query language that can do something like
Relate river to city
by nearest ordered by city.population
where river.distancefrom(max_elevation(city)) < 100
and max_elevation(city) < 0
Select river.name, (river.width.nearest_point(city)) as width, etc
Maplibre-rs is something to keep an eye on, but it's very much a proof of concept and early going. It's using wgpu (https://github.com/gfx-rs/wgpu), which is a rust-based "implementation" of the WebGPU api, so to speak, which of course would support running via wasm in the browser as well as having support for pretty much all the other platforms.