It seems to me that anyone wanting to make a profit from maintaining such a cloud of hardware would be best served by just doing the mining themselves rather than renting it out to people who could make more in mining than they pay in rent.
In an ideal market (e.g., with perfect knowledge of utilities to be realized by all decisions from all participants) this might be true. In real markets where outcomes are in fact uncertain and different participants both have different perceptions of the expected utilities and risks and, even to the extent those perceptions align, different sensitivities to risk, its quite possibly optimal for one participant based on their perceived utilities and risk sensitivity to provide minining hardware for rent while also optimal for some othe participant to rent that hardware for mining.
(Note that essentially the same consideration applies to making and selling mining hardware vs. making it and keeping it to use to mine.)
Mining contracts happen when two parties have different predictions about future difficulty and thus profitability; the buyer believes that mining will be more profitable than the seller believes. I suspect most of the buyers have lost money because people who run mining data centers for a living probably understand it better than anyone else.
There's also risk distribution: if you rent your hardware for 80% of the value, you're sure to get 80% of the value, while the risk is portioned up among many smaller investors, all of whom could reasonably stand a loss; if you don't, you're stuck holding the bag entirely if something goes wrong (eg, the has rate speeds up more than predicted).
There is a value in risk distribution that can cause mining hardware to be a good idea to rent out pieces of even if you're likely to make money on it, as long as the risk of losing money isn't trivially small.
