For the people who criticize as to why they've taken so long in shifting to a PoS model: consensus takes time, so does developing and testing something that definitively shouldn't go wrong. Also, there's the inherent issue with Proof-of-Stake that Proof-of-Work doesn't have: the initial distribution of the coin has to be wide enough before it could feasibly self-maintain a PoS shift without being immediately vulnerab…
I wonder if it's already on on a testnet
What remains is to change the legacy clients to use the PoS network for choosing blocks, instead of miners. That has a working multi-client testnet.