E.g. if you adopt the mindset that your CAPEX is a sunk cost, which is reasonable if you already own something nobody wants to buy (unprofitable oil or mining rig), your break even becomes much lower (just has to beat cost of operations/electricity).
With mining it's even more difficult, since mining at an "unprofitable" point in time could later actually become a very profitable venture if the price goes back up, assuming that the mining difficulty also goes down in unprofitable times.
Edit: ninjaed by mkempe.
I know it's just "one analyst", but I don't mind investing (gambling) on something that may give me a x10 yield.
Say your [marginal] production cost is USD 100, but BTC market price is USD 50. You could spend USD 100 to make 1 BTC, or you could turn off your mining equipment and spend USD 100 to buy 2 BTC...
Added: and when the market price goes back up to say USD 1000, would you rather have the 1 or 2 BTCs in hand?
But there are (perhaps edge) cases where your operation cost is essentially zero. A couple of friends are using their gaming rigs for altcoin mining, and simultaneously as space heaters (we live in cold climate). If they stopped mining, they'd have to turn on the actual space heaters, no electricity cost saved. And they already have the hardware, so no expense there either.
The reason why mining rigs aren't sold in the space heater aisle is that people usually don't buy $2000 space heaters.
You just aren't trying to make a return on the cost of your initial investment (CapEx).
discussed 6 days ago in https://news.ycombinator.com/item?id=16120062
Money quote from the top comment: "The break even is 1 BTC = $922". Personally I have my doubts, but even at the $6,925 the bloomberg article claims is still very profitable.