Hash Distribution

If each row belongs to one particular distribution then how can the data present inside one distribution vary from other distribution. Both will contain just only 1 row in them. Let's suppose I have 100M rows. So according to the logic, I'll be have 100M distributions. These distributions will further br divided across the compute nodes. Can you please explain this ?