Oh and we runned a Tensorflow Lite image processing pipeline on that chip. It is probably a few times too powerful for fiddling pins and an IP stack.
In some cases they've never been. It depends on the series. A lot of their portfolio is being made in external fabs like TSMC, whose allocations they gave up when Corona started as they were thinking the chip demand will go down. (Oh boy were they wrong.) On the other hand they still had allocations (in their "own" fabs), so depending on what you want they deliver. Problem is basically anything "mainstream" was in external fabs, e.g. the entire STM32F0 to STM32F4 series. In contrast, if you want an STM32L0 oder STM32L4, you can still order like nothing happened.
The G series seems to have good availability at the moment, if I was going to bet on a MCU for a new design that would be it. L series has only just gotten better, couldn’t get them a few months ago.
Some F3 chips have been looked over (didn’t go out of stock) despite being more advanced than their F4 counterparts. That’s marketing I guess.
There are some variants/footprint combos you can get. For example, I've been able to get G431s for a while in the right footprint, which is similar to the one I'm looking for , but with less flash and ram. Or, I can can an H7 in a way-too-big footprint.
Other MCUs like Nordic and ESP went through the shortage with no unavailability. Not sure if STM's comparatively large product lineup (as opposed to there only being a handful of Nordic and ESP variants) played into this.
Then it'll definitely do the job!