You seem a little fixated on absolute hashrate numbers. Remember that if the PoW changed, the hash function is inherently different, and so comparing SHA256 hashrate with CuckooCycle hashrate is pretty meaningless.
You're also forgetting that PoW change will see Bitcoin mining explode in popularity, so the actual purpose of the hashing will be far better satisfied with such an unavoidably more decentralised hashrate.
You forget too that Coinbase and Bitpay are not the only providers of their respective services. And in fact, the recent news about Atomic swap transactions with BTC/LTC trading pairs is a paradigm shift in the making when it comes to cryptocurrency exchanges. Sure, Coinbase could provide it, but what would be the point in having a centralised point of trust when atomic swaps transcend the need? Who's going to trust Coinbase after all this strongarming anyway?
You should stop forgetting things
You're also forgetting that PoW change will see Bitcoin mining explode in popularity, so the actual purpose of the hashing will be far better satisfied with such an unavoidably more decentralised hashrate.
You forget too that Coinbase and Bitpay are not the only providers of their respective services. And in fact, the recent news about Atomic swap transactions with BTC/LTC trading pairs is a paradigm shift in the making when it comes to cryptocurrency exchanges. Sure, Coinbase could provide it, but what would be the point in having a centralised point of trust when atomic swaps transcend the need? Who's going to trust Coinbase after all this strongarming anyway?
You should stop forgetting things
Comments are based on the assumption that current miners hold their signaled position.
Agree that hashpower in terms of hashes per second can be much less, on a new hash algorithm, if there are no current asics available, and still be strongly credible.
However,
People are going to have to have the new Core selected hash scheme for POW quickly or the Core-ALT-Bitcoin will take quite a while to get going. If Core selects a new POW Hash scheme with the due diligence that it normally follows that would seem to take a fair amount of effort, test deployment, yada yada. When are you thinking this is going to happen?
I see the current Bitcoin wonderful success as a very large ocean liner, used by a lot of people. Useage patters like oceanliners turn slowly. So while there may indeed be longterm ways around the existing exchanges (and damn I hope their are), I don't see that as having much effect in 2018 while Bitcoin-from-segwit2x rocks on and Core-ALT-Bitcoin is a low volume altcoin trying to get listed.