This is where I see the "staking makes a cryptocurrency a security" fall apart. There are a handful of definitions for "staking" and some of them definitely meet the criteria of "a common enterprise with profits to come solely from the efforts of others" however many do not.
1. Ethereum's staking at a protocol level requires you to run a staking node and you are paid for what is basically an SLA between you and the network and you only get paid if your node maintains a certain uptime, is kept up to date, and operates correctly. That I don't believe meets the criteria as it requires direct, sustained effort from you the operator (even if it's generally low effort).
2. Cardano's (or Tezos') staking is similar. Stake pool operators are effectively the same as Ethereum's stake node operators. But even delegators (who aren't required to stay online) still provide a service in that they are picking the stake pools who then fill the SLAs. If they pick pools that can't meet the requirements then the delegators don't get paid until they can find one that does.
3. Meanwhile you have networks like Algorand where participation in consensus does not effect staking rewards and you never have to perform a service to get paid out by the network. Those would meet the criteria by my understanding.
4. And then you have all the DeFi "staking" which is better described as lending or liquidity pooling. You aren't doing anything proof of stake related but are just lending out capital as an investment.
I can't speak on other networks but generally I found that networks fell into one of those 4 categories. The first two only pay people who provide a service back to the network while with those like the third, even if you can provide a service back to the network, you aren't required to to be able to get paid. And then the fourth category is just a security outright.
Edit (because I forgot to mention it): With regards to Coinbase's staking program, it's in a weird spot. With those you aren't directly staking but you are outsourcing the responsibility to a 3rd part (coinbase) to stake for you. I wouldn't be opposed to considering this type of custodial staking as meeting the criteria to be a security but I don't think proof of stake at a protocol level constitutes "a common enterprise with profits to come solely from the efforts of others" as they require work on behalf of the participants to get paid out.
Ethereum was funded as a public ICO.
Which is weird. In my humble layman's opinion, an ICO looks an awful lot like a security offering, while a staking protocol doesn't.
Compare the simplified statements "Buy my token! It's going to be great in the future!" with "Let's pool our money and execute an agreed-upon protocol where those who have the most get more". Why is the SEC wasting time by going after all the weird ones before the easy pickings?
ICOs, based on specific facts, may be securities offerings, and fall under the SEC’s jurisdiction of enforcing federal securities laws.
If you pay money in now and get more money out later, and someone else runs the machines, then it's just a loan.
Then there's a question of where to draw the line with cloud hosting, and I have no opinion about that. Someone will sell a service that's barely on the right side of that line, wherever it is.
"The test is whether the scheme involves an investment of money in a common enterprise with profits to come solely from the efforts of others. If that test be satisfied, it is immaterial whether the enterprise is speculative or non-speculative or whether there is a sale of property with or without intrinsic value."
I think it is plausible that all pure staking initiatives will go. I don’t think you have much of an argument here but I could be wrong. It will definitely be an interesting case to watch.