Also number of people using Azure isn't the metric that's really interesting, a better metric would be something that shows how many server hours were bought in

Also number of people using Azure isn't the metric that's really interesting, a better metric would be something that shows how many server hours were bought in a way that you could compare with EC2. I know Microsoft goes around university campuses trying to get students to use Azure for projects, and hand out a lot of test accounts in the process, inflating their numbers.