I suspect the latter is much easier and cheaper than the former? You can port a lot of software with cheap (or even local) models if you're tenacious whereas finding all the bugs is both very very expensive (if it's even possible) and potentially never ending (there's always new code and bugs!).

Maybe these big ai labs will uses their own devices to find and fix bugs up and down their stack and contribute that back.