And I do agree with this as in TODAY, the math is simple, our most powerfull supercomputers calculates in 30sh PFlops that's about 30x10^15 Flops Time in year = 3600x(24x365+6) = 31557600s and 2^256 ~ 1.14x10^77 so it will take to crack it with the usumption that it will require 100Flops per combination = 1.14x10^79/(31557600x30x10^15) =~ 1.20x10^55 years !
BUT THAT'S NOT THE POINT! My point is if you consider only classical computing in the last 30 years we've moved from KiloFlops to PentaFlops or 10^15Flops in terms of processing power, it is easy to assume that in the next few decades, we can easly achieve 10^30 / 10^40 (we've already gone past the point of cracking 2^128 or 128bits in a few seconds) and it will reach eventually 10^70+.
The advance of computing hardware is irrelevant relative to the strength of a 256-bit key. Seriously, irrelevant. If you rely on hardware improvements, you get to the point where you can flip a bit with one electron volt of energy, and there is NO FURTHER IMPROVEMENT TO BE MADE BEYOND THAT POINT. And that point, is still not relevant - to the tune of taking billions of years of the total output of the sun. And no matter how fast the computers get, that isn't relevant either, because they can't run faster than they can get the energy. We're not talking anymore about circuits and design - we're talking about fundamental limits of physics.
Now, if you're anticipating that we're going to get through it by learning better algorithms, that's essentially saying that the code isn't secure *now*. Which, in theory anyway, could be. Right now we don't know any way to attack the problem short of brute force. In fact, that's pretty much the definition of cryptographic algorithms... but there are a lot of cryptographic algorithms that turned out not to hold up to the mathematical insights of later generations. ECDSA could be another of them.
BTW, back in the 80's I was the guy saying "56 bits? That's maybe enough for most things now, but not for anything worth a million dollars or more, not for national security, and not for more than a couple more years at most..."