It would be interesting to see how they count a "user" too. If a good portion of those 2 billion people don't use YouTube very much then the % of users who are using it regularly and subscribing might be a lot higher.