I’d imagine chip design is sufficiently complex that you could both be competent and not have a fucking clue what’s causing this. A recall is bound to be cheaper than the impact this is going to have on customer trust. Not only are they the lower performance chips, they’re buggy lower performance chips.
They have a clue…probably had a clue before they shipped. I chat up the engineers where I work and they always figure out the what, how and why somethings fucked when more than expected failures pop up on a wafer or God forbid in the customer’s production units like a car or radar or medical equipment.