The fabled cliff is still nowhere in sight; 2020 might not even have been the peak.
I think 2020 was the peak. Things look great right now in relative terms because January & February 2020 were before AC and the pandemic. And in absolute terms, the Switch is doing very well. It's on track to be a contender for best fifth Q1 ever (the DS is the current #1 with 1662k units sold in Q2 2009). But the March-October period will provide a gigantic obstacle for the Switch's 2021.
The Switch sold over 1.8M units in just March & April alone last year, March because of AC (with a boost from the start of the "COVID bump") and April because of the first round of stimulus checks (with maybe some residual effect from AC). While there is a new round of stimulus checks starting to go out right now, larger than the January checks, there is nothing near the scale of Animal Crossing to help boost sales, so the likelihood of the Switch's March+April period this year being anywhere close to what it was last year is slim. Unless the Switch does at least 1.4M total for the March+April period, it's going to fall behind 2020 in YTD terms. Looking beyond April, the May-October 2020 period saw 3145k units sold, a weekly average of 121k units over that six-month span. Only the DS has managed to maintain a baseline well over 100k/week for a period of more than a year, and the Switch has been struggling to keep pace with the DS. Overall, the Switch will need to sell at least 4.5M units from March to October just to keep pace. Then it'll have to sell at least another 3.4M during the holidays. While March is probably going to be a another pretty big month, if April and May (typically the slowest months of the year) are seeing sales closer to 300k, that could be a sign of a slower year, at least relative to 2020.
I still stand by my prediction that demand for consoles will soften once we're on the other side of the pandemic, which at this rate could be sooner rather than later (and that also entails a high likelihood of the current round of stimulus being the last). If so, then without something, either internal or external, to keep demand high for most of the year, selling another 9M or more units this year is going to be extremely difficult. Not impossible, but highly unlikely. I think sales will drop this year, but not off a cliff. Something far more modest. My current prediction is 7.5M, only a 16.7% drop, though I'm willing to go as high as 8M.