I had some additional commentary to make, specifically regarding "peak years" in general.
Systems peak at different times, the specific reasons being different for each one. Some peak early, while others peak later. Some peaked in their first year, while the 360 didn't peak until its sixth year, the all-time record late peak for a home console. Even a single system will usually peak at different times in different regions. For example, the DS peaked in 2006 in Japan, but not until 2009 in the U.S.
Before I move on to the "why do they peak when they do" part, I wanted to address the arbitrariness of using a calendar year to define a peak. Why not any 12-month period? For example, while the PS3's peak calendar year in the U.S. was in 2011, its best 12-month period was the period from Sept. 2009 to Aug. 2010, where it sold 4914k, which was 8.7% higher than the 4519k it sold in calendar year 2011. That case is especially notable considering that the 2009 to 2011 period saw each of those three calendar years being extremely close to each other, to the point where it was pretty much flat for those three years.
Even a single bad month can prevent a year from being a system's peak calendar year. The PS4's 2018 was brought up earlier, and while its best calendar year in the U.S. was indeed 2015 and not 2018, its best 12-month period was from Nov. 2017 to Oct. 2018. The Jan.-Oct. period of 2018 saw the PS4 sell 3040k, 16.6% better than the 2607k sold in the same period in 2015. While Nov. 2018 was a slower month than Nov. 2015, it was only by 4%, not enough to seriously dent 2018's lead over 2015. But December was enough by itself to take away that YTD lead over 2015. Dec. 2015 saw the PS4 sell 1582k, while in Dec. 2018 it sold only 797k. What was a 372k lead over 2015 going into December turned into a 413k deficit in the span of just five weeks. This was due to there being absolutely no big holiday deals in the Dec. 2018 sales period.
So, which is it? Did the PS4 peak in 2015 or 2018? If you arbitrarily restrict things to calendar years alone, it would by definition be 2015. But it had its best non-holiday period in 2018, and only a single month prevented that calendar year from being the first. Using less restrictive criteria, one could easily argue that, while 2018 wasn't its peak calendar year, its overall sales peak was in 2018.
I also have to ask, why can't we define a system's peak with other criteria? While using a multi-month period would be better than a single month as it gives us a good running average, why not a best quarter, or best non-holiday period?
In any case, why do they peak when they do? While a lot of people are talking about software lineups, there are other factors, some of them far more important than any games. That's not to say software can't be a deciding factor. In the PS4's case, that peak Jan.-Oct. period in the U.S. in 2018 was due mainly to three major system-sellers (God of War in March, Spider-Man in September, and Red Dead 2 in October). But sometimes it's a hardware revision that's the deciding factor. In the case of both the PS3 and 360, it was their slimline models, while in the DS's case, it was the Lite in Japan and the DSi in the U.S. Sometimes it's a specific price cut that puts it over the edge, such as the PS2's first price cut in the U.S. in 2002 or the 3DS's big price cut in its first year. On rare occasions it can be other factors. The Switch's best year in terms of absolute numbers and YoY growth was 2🦠2😷 *ahem* 2020 despite it having no new hardware revisions, no price cuts, and only two major blockbusters released that year, noticeably fewer than any other year to date, and that's because of [REDACTED] ... I mean, reasons we won't discuss.
TL;DR: How we measure a peak can be rather arbitrary, and wherever we determine that peak to be, the primary cause of it can be any of a number of things.
Visit http://shadowofthevoid.wordpress.com
In accordance to the VGC forum rules, §8.5, I hereby exercise my right to demand to be left alone regarding the subject of the effects of the pandemic on video game sales (i.e., "COVID bump").