Current GTX970 speed:
GPU 0 found share: 7-ch type 2
Share accepted.
GPU 0 found share: 7-ch type 1
Share accepted.
[GPU 0] T=-1C A=-1% E=0 primes=0.108847 fermat=303492/sec cpd=6.12/day
(ST/INV/DUP): 2606x 7ch(6/0/8) 265x 8ch(0/0/0) 29x 9ch(1/0/0) 2x 10ch(0/0/0)
10CPD is too optimistic.. I think, maximum possible speed of GTX970 is ~8 CPD, GTX980 - 9.xx (same as R9 280X, but with less power consumption).
I am interested to know if it will be using CUDA or OpenCL for Nvidia? My guess is OpenCL as it is a nightmare to recode everything in CUDA and then maintain two versions, but CUDA should be quite a bit faster than OpenCL on Nvidia platform.