If I understand your argument properly, then I believe you to be wrong about the idea that tenably factoring 1024 or 2048 bit moduli implies a solution to the elliptic curve discrete modulus problem, yes.