Storage workloads aren't uniform. Some need to have localized, reliable, consistently performing persistent storage with room to absorb sudden demand spikes (like customer data on cloud clusters). Others can be stored anywhere and moved with ease, with few requirements on latency, redundancy on a global scale, and read-mostly workloads, which means immediate loss of a single local copy is harmless (like YouTube videos). FAANGs have long since figured this out and how to optimize their storage utilization.

There is zero chance one of them is going to up and decide that Chia is a better use of their "spare" (not really) storage capacity if they have anything resembling competent engineering teams. I know Google does.

Source: Xoogler.

While I don’t think too many businesses will buy this idea of using spare storage for “cryptofarming”, unused resources are pretty common.

The price of a cryptocurrency will tend to roughly follow the cost of generating it. In storage, it’s safe to assume the cost will only go down for the same amount of space. And disk space has been growing faster than transistor density for some time now.

So in the end this is going to be just like every other cryptocurrency: people largely deploying exclusive mining (sorry, "farming") capacity and consuming resources doing so.