Right. Assuming the Prismas' are $1500 cheaper and you'll mine ~$800 before the SP35 ships (start mining Oct 15th, end mining Nov 21st, 10c/kWh, 15% inc. per jump with a 5% lowering per jump), that gives you $2300 in hand with the Prismas vs the SP35.
From Nov 21st onward the difference in power is 11kW vs 7kW, so 4kW. There's 8760 hours in a year, so at 10c/kWhr it would take 2.62 years for the cost of powering the units to make up for the price differential.
At 15c/kWh, the numbers go down to 1.4 years of running before the difference in price is made up.
At 20c/kWh, don't bother. At the current price and next estimated difficulty (~35B), the network only has to increase 50% in size before you're at breakeven for power with $400/BTC which it very likely will be by the time you get your Prismas.
I got slightly different numbers.
10 AM prismas = ~$7000/15TH
2.5 SP35 = $10,500/15TH
According to
https://bitcoinwisdom.com/bitcoin/calculator at 15% increase the AM Prismas will earn about 4btc before SP35 begins shipping. (if they start mining on November 1st)
So the $3500 cost difference + $1500 mined means the SP35 needs to recover $5000 in electricity savings to be a better deal.
2.5 sp35 is 8.75kw so it would save 2.25kw.
At $0.05/kwh the 2.5 sp35's would save $2.7/day and at that rate it would take more than 5 years to make up for the price difference.
In my numbers the AM Prismas (at 14TH/s) would earn 4.4BTC between Oct 15th (~1/3 of the way through their delivery window, coincidentally also an estimated diff change) and Nov 21st (1/3 of the way through the SP35 late November delivery window, also an estimated difficulty change). However, in those 37 days the Prismas would burn 11kW*24hrs*37days= 9768kWh of electricity, which you have to account for. At the 10c per kWh I was using, that's almost $1000 of electricity you need to pay for. Even at 5c, it's $500.