The Sunken cost fallacy would be to retain those people because of the costs irrevocably incurred (sunk) recruiting them, employing them to date, etc. Apple are (apparently following a theoretical economics position) considering the prospective costs going forward of keeping them, when they no longer have an obvious future revenue stream, and would still generate additional redeployment/retraining costs
In this particular case, GP posits that fallacy is itself a fallacy, insomuch as there is no greater opportunity the company could pursue which required laying those employees off. The reputational harm done, among other things, is a hidden cost not reflected in the numbers of their salaries, and surely they could have participated in other R&D efforts, or something along those lines.
Classic instance of the sunken cost fallacy fallacy (not a typo). They prefer to cut their losses and forget about how much they spent recruiting those people.
Keeping them might seem wasteful on the short term, but such skilled workers could likely be made profitable in a big R&D powerhouse.
But the real cost is damaging their reputation among future candidates and current employees, and lowering their perceived value as an employer.
For the most skilled and efficient engineers, choosing to work for big tech is always a tradeoff, a company like Apple is usually very appealing for the most risk-averse.