Cool. I'll have to check out WASABI.
I recently signed up with Azure for a new project of mine. It's a software as a service model where other website owners will install a snippet on their website to gain specialized insights and uncover problems.
So of course its popularity is unpredictable at this point, but it has the potential to see an enormous number of requests every day and may even see drastic bursts of traffic. So scaling has always been at the back of my mind as I'm building it.
Furthermore, it's absolutely true that Azure makes you think about efficiency. Since you're essentially paying for usage, it encourages you to make your application as efficient as possible. I'm always thinking about this as I'm writing the code, designing the database, etc.
I recently incorporated an optimization technique where I'm storing a temporary list of updates in-memory (I'll move this over to Azure Cache soon for better reliability). Then I send these updates to SQL Azure and Azure Table Storage once every minute. Due to my schema design, I'm able to greatly reduce the total number of inserts/updates that have to be performed by using this technique.