The power required will in fact only have increased to twice it original value because the density will be about 1/4 of its msl value. Under these circumstances it would be reasonable to say that power required is proportional to Cd x density x TAS cubed. But it would not be at all accurate to say that power required is proportional to Cd x density x CAS cubed.
Looks like we agree on the equations, Keith. We'll just have to agree to differ as to how we describe those in words.