I haven't had a chance to listen to Nate's podcast yet, but mentions of DLSS ray reconstruction make me hopeful that it's not just there, but viable. As already mentioned, Ampere supports DLSS-RR, so as long as Nvidia adds support for it to NVN2, it should technically run, but there's a difference between running and running quickly enough to be a sensible choice on low-power hardware like the Switch 2. We know DLSS-RR is slower than regular DLSS, and that it can sometimes be quicker than running both traditional denoising algorithms and DLSS, but without actual performance figures it's hard to determine whether that's enough to make it usable on Switch 2.
If developers are talking about it to journalists, though, that makes me think it's more likely to be actually usable. If it's tucked away in the SDK with documentation stating that it's extremely slow and probably only of limited use, I wouldn't expect devs to talk about it, or if they did, to caveat it as "technically it supports DLSS-RR, but it's too slow to use". If devs are proffering up information that DLSS-RR is supported, without caveats, then my guess is it should be fast enough to be actually viable on Switch 2 at least in some cases. That's great news, as it makes a big difference to the image quality of games which make heavy use of ray tracing (or at least in Cyberpunk 2077, which is our only example thus far). Of course I don't expect Switch 2 games to make similarly heavy use of ray tracing as Cyberpunk does, but even for game with more modest usage, we should hopefully get much cleaner reflections/shadows/lighting/etc as a result of using DLSS-RR.