He has never said so!
He has said that f''(x)>0 implies that f'(x) that is the derivative of f is increasing. But that does not imply that f is increasing, too. Think e.g.f(x)=(1/2)x^2 ,so f'(x)=x, where f'(x)<0 for x<0 and so f is decreasing there. But, still, f''(x)=1>0. Thus, f'' determines curvature, that is , affects f
indirectly, and f' directly.