There is no new "more efficient" node anticipated to BUILD miners on for years.
Is that hard for you to understand?
In fact, there is some question how much longer semiconductor node CAN continue to progress - GMO's over-optimistic projections to the contrary.
Intel and IBM in particular have both stated that "pure silicon" has reached the end of the road - IBM's current "next gen" process design work uses a MIXED silicon/germanium die while Intel has been quite mum on their plans for anything past their current 10nm node.
The old "new miner model every 6 months" routine died when ASIC mining hardware caught up with semiconductor state-of-the-art at the 14nm node (S9 and such), and CAN'T ever return.
Your plan is nice on paper, mine long time, but unfortunately right now if we assume a flat price and over 7 cents per kwh you're not going to ROI before the halving
Of course, prices might go up but at that point just buy directly
BTC.
The s15 might be efficient but it has a price tag that is dragging it down and this is not a normal industry, I doubt there is something out there where you have to plan a halving in global markets every 4 years.
With an s9 you're going to ROI faster and start making a profit faster, the s15 is indeed for the long run but it might take way too much time to make some bucks on it.
Slash the price at least 25% and yeah, it looks attractive but at its current price...nope. Just no!