As I mentioned in the past, it is simplifying the process for the narrative and depends upon the semantics of the stair and Monty is expressing the output (which is fair enough but context is critical);
Case in point if you explain the line pattern of quantisation noise/error, it showed in the analyser as stair-step but Monty is not putting weight on this because the context of his narrative is the end output).
What about the output of a NOS DAC with no filter and the pattern of the aliases (you will be suprised but it is the filter that is critical for removal aliases that can create a similar behaviour to stairs that then becomes smoother at higher sampling rates - very specific to the filter or lack of one).
What is the difference between a full represented sinewave and one that is at -90dbfs when comparing 16-bit and 24-bit (hint they can start to look very different but again dithering resolves and hides this, although technically it is another separate process added to the digital chain just like quantisation-dither (now days inside the workstation)-oversampling/noise shaping-DAC-filter.
While these are not really/semantically stair-stepped they are closer to that than actual discrete sinewave points as described above in the Monty video.
But as Monty and myself discussed in another thread we are approaching this from different angles and narratives; the video being the end result (although it should consider and discuss the effect of dither and the filter to remove said stair type effects that would otherwise show).
Cheers
Orb