That's quite unlikely. If Dropbox could save a decent amount of their engineering effort using LLMs with 20% less people, then they could also increase their engineering effort by a decent amount using LLMs and keeping the same workforce. Assuming the team aren't completely useless, being able to go faster would be significantly more valuable than any cost saving made here.
I dont know… i felt like the eng facing llm work at my company was distracting and not useful. I can only see Dropbox and others obsessing over LLMs as clinging to relevancy. At the end of the day we will just get a chat bot and a handful of niche projects AI investments paid off for
