> How would Bitcoin ever move over to PoS like this? Would it just be a Bitcoin hard fork?
I believe it would be possible for Bitcoin to migrate to proof of stake with a soft fork, similar to how they migrated to SegWit. A majority of nodes (both non-miner full nodes and miners) would have to upgrade to a version which rejects blocks which do not contain a proof of stake together with the current proof of work; non-upgraded nodes would ignore the proof of stake but would still accept the blocks as valid since the proof of work is valid (this is the key difference between a soft fork and a hard fork; on a hard fork, non-upgraded notes would reject the new chain). And to get rid of the proof of work, the block intervals would have to be tweaked to make the network appear to be running slow (for instance, making every block take one minute longer, either on reality or through clever manipulation of the block timestamps), so that the difficulty adjustment gradually reduces the difficulty, until the (now vestigial) proof of work becomes trivial to compute.
I believe it would be possible for Bitcoin to migrate to proof of stake with a soft fork, similar to how they migrated to SegWit. A majority of nodes (both non-miner full nodes and miners) would have to upgrade to a version which rejects blocks which do not contain a proof of stake together with the current proof of work; non-upgraded nodes would ignore the proof of stake but would still accept the blocks as valid since the proof of work is valid (this is the key difference between a soft fork and a hard fork; on a hard fork, non-upgraded notes would reject the new chain). And to get rid of the proof of work, the block intervals would have to be tweaked to make the network appear to be running slow (for instance, making every block take one minute longer, either on reality or through clever manipulation of the block timestamps), so that the difficulty adjustment gradually reduces the difficulty, until the (now vestigial) proof of work becomes trivial to compute.