Here is my thought though. I think decentralized storage is the future. It can be done at incredibly cheaper prices than the big companies can, and to get the infrastructure built now would give a huge advantage over the years as newer and better technology come out. Burst is probably the most profitable currently,
BURST isn't decentralised storage, that's STORJ or MAID (and SIA? Not sure there).
Storage isn't that hard to shrink, it's just EXPEN$IVE to do so right now.
Intel showed off 60 TB (yes, SIXTY TERRABYTES no typo) in a 3.5" form factor last year - issue is that it was a SSD and probably WAY pricey.
I don't think it will happen real soon, but SSD will probably replace magnetic HD storage eventually - it's already there in niche usage, just not quite price competative on a per-TB basis yet.
My current estimate on ROI for a Seagate Archive 8TB (which is usually the lowest price per TB on new drives) is around 20 months, unless the price starts climbing faster than the network total TB does (price has been DROPPING like most altcoins the last 2-3 weeks, while network capasity keeps climbing).
There are calculators, but they are ALL on the optimistic side as due to how BURST pools work, only *one* machine pointed at the same "address" gets counted for total capacity on that pool.
There are times the pool thinks I only have a few hundred MEG of capasity, even though my total among all of my machines is probably 20 TB or close.
Up side - BURST doesn't use a drive particularly hard, so they should have good longevity in BURST usage.
Another up side - I have to buy a drive anyway when I'm building a new mining machine, and the cost of buying a 3TB HGST refurb isn't much more right now than buying the cheapest drive available, so the difference in price will pay off pretty quickly.