It reminds me a bit of when people worry about collisions in random uuids
I once had to convince someone much smarter than me that UUID collisions (for a messaging system we were developing) would not happen. It was a surreal experience.
Are you sure they were talking about UUID collisions in a idealized/theoretical system? If you generate UUIDs using, for example, a random number generator with only a 16-bit seed, you absolutely will see collisions, for much the same reasons as in TFA.
I am regularly concerned about using UUIDs in a public-facing way, since that would encourage people to make "vanity" UUIDs somehow. Requiring some kind of value to be hashed to create the UUID would sidestep this a little.
Those are more subtle, and often well-founded, due to the birthday paradox.
(and sometimes weak entropy sources)
It is but probably though new math and not faster machines.