* Long term I have to imagine this is just "cheap" because there's got to be tons of industries that can use intermittent free energy.
Usually clients want consistency in the output. (they depend on the product/service)
Business owner usually wants consistency in employee/machinery output. (They cost money)
To me it seems that only low priority computation (incl. con-coins) is such industry. (Investment kind of reused, and relatively few employees)
There will be some shift to adjust to intermitency (e.g. Worker hours might be shifted to the time when solar generates the most energy to take advantage of the price)
Like I'm not an expert on fertilizer production, but it seems pretty likely to me that you could draw that box around ammonia production and build out capacity that relatively cheaply, while leaving what I imagine to be the more operationally complex and capital intensive steps of extracting phosphates (via sulphiric and phosphoric acid) and potassium (which has nitric and sulphiric acid as biproducts), and combining that nitric acid with the ammonia (into ammonium nitrate) alone.
In some cases if you have excess production capacity (because you have unpredictable or seasonal load and you aren't currently at a peak) you don't necessarily even need more production capacity at all, just the ability to store a bit more of the inputs and outputs in a buffer.