Pros
1. It's in Seattle. 2. You get to support the O.G. Cloud. 3. Ultimate mobility - if you see a team you want to join, just start hanging out with them. If you're invaluable to your present team, it just takes a little longer. If you've been around, you've got a great shot at getting on one of the S3 ops teams -- only dev teams get resumes from recent grads. It's really hard, but it's all there, simply a matter of reading. Nothing at all off the shelf any more beside the hardware, all either open source or Amazon proprietary. Best integration of incident management and knowledge sharing, ever -- just read old tickets, all are heavily annotated. I strongly recommend this to cuspy olde UNIX admins, especially self-taught. Most important skill of all == log diving. Enormous and entertaining library of videos from principal engineers to teach the unique architecture. New model to scale to the always-increasing size of the cloud every few years -- you'll know when probability equations start showing up on the whiteboards in the hallway. Nothing wrong with developing or doing ops for other teams - my experience is limited to the S3 team.
Cons
Not just a time-sucker, can be a life-sucker. No one will force or even suggest putting in more time, it just happens because it's completely and utterly fascinating.