It does, but mostly in an indirect way.

See, companies that deal with a lot of traffic on static data have geographically distributed caches.

Let's say Steam has a major game release, and gets slammed with the DL traffic of 5 million gamers all around the world trying to get their hands at that new game all at once. However, Steam has an instruction manual that allows any ISP to set up their own cache servers. So an ISP that has a cache set up can convert a lot of that global traffic to local traffic, saving them money, and offering users a better experience.

(One small ISP I knew had it set up so that all traffic to their local Steam cache was fully exempt from client rate limiting, reportedly because the ISP's admins were avid gamers.)

Other services like major CDNs, YouTube or Netflix may have deals with ISPs to locate their caching hardware on ISP premises, or may buy their own caching servers in specific datacenters. Same idea applies - it's cheaper for both ISPs and web services when the users hit local caches than when they "cache miss" and generate global traffic.

VPN use is a "forced cache miss", so it's a loss-loss for both ISPs and web services.

Not really an L for web services, since the caches would just end up near the VPN locations (and sometimes inside the same DC).

Disclaimer: Former YT Engineer.