The Nintendo Switch 2's DLSS capabilities may not be as powerful as has been rumored due to the lack of a deep learning accelerator, which may limit its upscaling capabilities.
DLSS (or FSR/XeSS for that matter) isn't a magic bullet. I'd say the "Quality" setting (67 % per axis) is acceptable for some games if the output resolution is at least 2560x1440 (so it renders at like 1715x960), but even then it's very noticeable if you ask me. At even lower internal resolutions, it's getting a lot worse quickly.
It's a nice way to get games running that wouldn't otherwise run with an acceptable frame rate, but not much more.
It also got brought up again in the latest DF weekly. A recent rumor states thst the switch 2 soc won't have the dedicated tensor/ML cores, so DLSS performance would take a further hit.
This really ought to be Nvidia's chance to show off Gsync and DLSS. They control the entire machine - screen included. A zillion people are gonna buy this hardware. You figure they'd go hard on making 40 feel like 60 and 720 feel like 1080, but you'd have figured that for the first Switch too.