There's a lot more to running a production database than simply installing the software. At a minimum you have monitoring, tuning, backups, failover/replication, and security. These and more are all tasks that some people look to outsource as it is outside of their core competency/business (for better or worse).
Amazon Web Services itself has three distinct services [1] for database as a service: SDB (2007), RDS (2009), and DynamoDB (2012). If there wasn't demand or money in it, I'm sure they would have stopped at their first, simplest iteration.
The over-simplistic response to that argument is that setting up a website is as simple as 'sudo apt-get install nginx && service start nginx', which delivers in principle -- now you have a website. In practice, we all know there's much more involved.
With some of the more spectacular failures we've seen as a result of Mongo (I'm specifically thinking of FourSquare's downtimes), I think that their offering is quite timely and useful.
We're currently mulling over the options between switching our local Mongo installs to either MongoHQ or engaging 10Gen support to manage it for us. Both are significantly cheaper than hiring a dedicated person to manage it.
Many years ago, people asked the same questions about food, clothing, shelter, farming, weapons, mining, steel, oil, gas, electricity, sewer systems, telephone networks, radio, television, and the internet.
