The reason the Bitcoin network requires a large amount of computing power is to enable decentralization, with a proof-of-work being the tool employed.
But that computing power is for a specific purpose, ... to make it so that the decentralization is immune to corruption. That's the only reason there is this large amount of computing power.
But the proof-of-work computation that is performed is useless for anything else.
Now that doesn't mean the computing resources that Bitcoiners will abandon can't still have use for scientific purposes, but that has nothing to do with currency.
CoinLab was making inroads to doing this but got stuck on how distributed scientific computing is a hard problem to solve with little opportunity to monetize it.
But if someone wants a "currency" based on computing power, essentially any exchange market for computing time would be that currency. But that's a security, not a currency.
Incidentally, Bloomberg needs to update their B-roll. It was recorded back when Bitcoin was $64-ish.