Since you wrote SHA256 you're not in the right loop here. Shor's algorithm is the looming threat (ECDSA), not Grover's algorithm, even though Grover's does offer quadratic speedup and there are potential threats there too.

As for this YouTube video, you can find countless such proclamations about the impossibility of the AI we have today from 5 years ago (i.e. the time we called it machine learning because we were to ashamed to use the term AI). And from very reputable people. You can even find Yan LeCun 2 years ago saying that what LLMs can now do LLMs will never be able to do down to physical limits. This is how it goes.

I looked at the YouTube video, it was posted a few months ago but is very out of date. The main arguments are going back to the time before we made a number of breakthroughs specifically in error correction and the sustainability of logical qubits. These breakthroughs, proven by multiple experiments, undo many of the key arguments. If that person remade that video today they'd have to scrap half of it.

The other thing I'll add is that unexpected breakthroughs are, by nature, *unexpected*. We are on the potential cusp of a number of them in quantum computing, from many different directions. This is how it goes, this is just the age of technology we are in, not only for quantum but for gene editing and all kinds of things. Betting against process is silly.