My lay impression is that AlphaFold (and now AlphaFold v2) solves the same problem F@H solves, but better, faster, and cheaper (at least, if compute power wasn't all donated).
Is that not true? Are they somehow complementary? Just how much did AlphaFold eat F@H's lunch?
AF did a tremendous amount of work in patching up our spotty structure database, but those hairy, novel regions are still going to require extensive computation to model, which is better done by simulation.
They are both working towards the same goal, but with different methods. Sort of like DSL vs Cable internet - both deliver the internet but how it gets there and what each technology allows you to do due to the difference between telephone lines and coaxial cable are quite different, and can even run side by side for extra coverage in case of an ISP outage :)