It would be a terrible idea for Intel to return to DRAM. Just like it was a terrible idea to enter the flash market. They are commodity technologies subject to massive supply and demand shocks. Intel made the most money by leading the process node race and shipping the best silicon for servers, and they lost their way through short-sighted financialization games, long bets on losing technology like 3D XPoint, and forcing out their best engineering talent.
I think some of this is true and some of this is secondary to their bad choices.
If they hadn't fucked up their process and gotten complacent from the absolute lack of competition in the space, they'd be fine today. And, if they were fine today, they'd have had money to continue to support more money-losing/commodity endeavours that could have had good payoffs as long as the fraction of their expenses weren't meaningful to the business (e.g. mobileye, altera, 3dxpoint, Intel arc).
The main issue is they squandered the golden goose and then all other endeavours that were worth doing but riskier also suffered.
It was just a lot of bad management and bean counters going unchecked for a very long time. Their board is equally at fault for signing off on the destruction of the business for so long.