Well i'm sure there are usecases where it matters, in general sha-256 is already really fast. Why bother with something faster?
Have you never sat there waiting for a hash to finish computing? I regularly do and I'm not even dealing with large data. It's very slow, and worse than that it's both very slow and also entirely sequential! (So is BLAKE2 - BLAKE3 is parallel.)
Worse than that - it's an astronomical waste of energy if we could be doing the same thing in less time. I wonder how many hashes humanity runs a day?
We need cryptographic hashes for many legitimate applications completely outside crypto-currencies, many of which protect critical things like national security and personal privacy.
And BLAKE is faster even on a single core, so it's strictly less compute time using the same instruction set.
For that matter, if your file is on disk, i would expect that sha-256 already works faster than reading from disk, so its not like blake3 is going to improve on that.