The number I posted was off the top of my head with rounded math, but it's probably close to the truth. There is definitely voltage sag. You can't just multiply things out to get a kW number. Doesn't work like that in the real world, especially when it comes to batteries.
The HV wiring is only 2/0 gauge. at 1500A or even 80% of that there is going to be significant sag there, plus more importantly the internal resistance of the cells under load. The internal resistance of the cells cause a significant voltage drop under load.
I've found the voltage drop under load, in testing, to be a hair better than the Panasonic NCR18650B, which has an internal resistance of about 55 miliohms in testing. I haven't made a definitive measure of the Model S cell internal resistance, but we'll be conservative and optimistic at the same time and say 40 miliohms.
So, at 1500A that is paralleled among 74 cells, so about 20.27A per cell. 85kWh pack is 96 sets in series of 74 cells in parallel. Remember amperage is constant across cells in series and split among cells in parallel.
Ohms law says that at 40 miliohms at 20.27A I would see a voltage drop of about 0.8108V. That would drop a fully charged pack's voltage from 403.2V down to 325.36V. 325.36 * 1500A = ~488kW.
For comparison, let's work the 1300A number from Elon. 1300A would be 17.57A per cell. Voltage drop of 0.7028V per cell @ 40 miliohms IR, so 335.73V. 335.73 * 1300A = ~436.5kW. Pretty close to reality considering this is optimistic and *only* accounts for internal resistance losses and not wiring, inverter, etc losses, which are bound to be appreciable.
So, being optimistic, I'm going to stick with saying about 500kW is the ceiling @ 1500A.
I'll go a little further and work the 420kW number (that we see currently from the P85D at full charge) backwards using the 1300A number from Elon and extrapolate.
420kW (~563HP) @ 1300A = 323.08V = 3.37V per cell after voltage drops (~0.8346V drop). That would give us 47.5 miliohms of resistance, which seems like a very reasonable full system number.
Using 47.5 miliohms at the 1500A level we get this:
0.96283V drop @ cell level, so 0.96283 * 96 = 92.43168V drop at the pack level = 310.76832V at the motor = ~466.1kW = about 625HP.
Basically there seems to be no way to spin this to look good.