Based on static conditions - 300 Ph/s network hash rate, 0.1 USD/kWh, $500 USD/BTC and immediate mining start, AM's new Prisma overtakes the S3+ and closes the gap on the SP35:

SP35 costs 4,035 USD inc. PSU + shipping for 6 Th/s at 3.5kW, it would take 147 days to break even,
10 x AM Prisma costs 9,250 USD inc. PSUs + shipping for 14 Th/s at 10.5 kW, it would take 158 days to break even, and
S3+ B9 costs 320 USD inc. PSU + shipping for 453 Gh/s at 355W, it would take 173 days to break even.

Also, here's how much profit they'd make each day (mining income - power costs):

SP35 = 4.6 USD/Th
AM Prisma = 4.2 USD/Th
S3+ = 4.12 USD/Th

I'll leave you to your own estimations with regards to changes in BTC price, changes in difficulty and pre-order waiting times but I will remind you that Prismas will be shipping a full month at least before SP35s and there's no availability date regarding S3+ B10s or S4s. A full Prisma set would make 1,764 USD (after power costs) in 30 days under the static conditions.

Here's a question for you RoadStress. Given the 1 month head start for the Prisma, how long would it take for the SP35 to overtake the Prisma in total mining profits? Show you're working as well. Oh yeah, and where's that 0.1 BTC you said you were going to send me.  

Edit: A mistake was pointed out where I was using the fixed 500 USD/BTC price for initial purchase costs and break even time, which I've now changed to $400 USD to be more reflective of the current conditions.

10 x AM Prisma costs 7760 USD inc. PSUs + shipping for 14 Th/s at 10.5 kW, it would take 185 days to break even,
SP35 costs 4,035 USD inc. PSU + shipping for 6 Th/s at 3.5kW, it would take 198 days to break even, and
S3+ B9 costs 282 USD inc. PSU + shipping for 453 Gh/s at 355W, it would take 214 days to break even.

Mining Profit/Th/day:

SP20 = 3.25
SP35 = 3.4
S3+ = 2.92
AM Tube = 2.1
AM Prisma = 3