Lesson 5 adding a column of ones as bias instead of random intialization

I feel that its a really dumb question but i can’t seem to answer it so here goes:
When jeremy talks about bias in lesson 5, [https://youtu.be/MkCAiRw7o34?t=2696], he says that you dont actually need bias and can add it for free through an additonal column of ones; BUT that is pretty inefficient. Could someone explain to me why is that inefficient as compared to the random initialization?