I think if DLSS is a part of this next hardware, then it has to be a new node and architecture. My understanding is you cannot simply “bolt on” tensor cores to a Tegra X1. It has to be integrated into the design, and software has to be specifically coded to use it.
But let’s real talk here for a minute. This is Nintendo. Everything we know about the company says they are far more likely to, at most, yet again node shrink their existing X1 and simply up the clock speed to make up that power draw difference, rather than paying for a new design. That way it will play all existing Switch games without compatibility concerns, and then they can simply offer a new high clock profile for devs to take advantage of the new speeds.
The truth is the technology exists, or could exist, to do the things heavily speculated in this thread like with DLSS and such. But this is Nintendo. They maximize hardware profits first and foremost. They won’t adopt expensive new silicon when a retread of their existing chip will “do”.