It's pretty much guaranteed at this point. The staking network has been running since Dec. 1, with over 3% of all ETH staked so far. The initial migration is simple to implement and a high priority for the devs. Community support is very high, and once everybody switches to the new fork the miners have no influence.
The miners could keep running the old PoW chain but it's unlikely to have significant value. Exchanges will support the PoS chain and some are already offering staking services. Tokens collateralized by off-chain assets will use the PoS chain. Etc.
With the PoW chain having little value, miners will be deeply unprofitable and most of them will have to quit. Moving to other chains isn't much of an option because other GPU chains have little aggregate value.
The more people join, the lower the rate of return[1]. The max inflation is about 1.71% annually, if 100M ETH stakes. Total supply right now is only 115M.
Net issuance will be lower than that because some transaction fees will get burned. It's possible that the supply will actually decrease.
There's no limit to the amount of stake other than the ETH supply, but they're talking about setting a limit of 32M ETH, just because it'd be more efficient and they don't really expect it to get that high anyway. Under that plan the validators get automatically rotated if >32M stakes.
There is a high likelihood of a fork, but forks aren't necessarily harmful to Eth. The "rebelling" miners can't actually do anything to the proof of stake system.
The real people who decide what fork wins are the exchanges and users, and those people have no real reason to stick with PoW.
Miners will rebel and continue mining, some will mine on an ETH fork and others switch to other coins. GPUs will not be returned to gamers anytime soon.
Theoretically the ETH PoW fork will drop in value relative to the original ETH. That would decrease the mining reward and result in less mining (and more GPUs returned to gamers).
The goal isn't to "kill" the coin entirely, the goal is to reduce the amount of mining by 99% or more. And that's an easily achieveable goal. See Ethereum Classic which only uses 1% of the resources of Ethereum because of how low its price is.
The miners could keep running the old PoW chain but it's unlikely to have significant value. Exchanges will support the PoS chain and some are already offering staking services. Tokens collateralized by off-chain assets will use the PoS chain. Etc.
With the PoW chain having little value, miners will be deeply unprofitable and most of them will have to quit. Moving to other chains isn't much of an option because other GPU chains have little aggregate value.