We're on the cutting edge of using social media and near-real time collaboration tools (free) to learn and use GAE for free, and then blog on the process (also free). The price is obviously right, and the ease and transparency of sharing and witnessing are just about friction-free.
As Dion points out (and Dan Farber makes note), there are trade-offs between GAE and Amazon (AMZN) Web Services. And there are concerns to be evaluated and vetted over the application lifecycle remaining in the Google cloud, as Garett Rogers makes note.
But the process I'm witnessing here on Twitter is nothing short of breathtaking for its rapid, agile and productive online team approach (we are located all over) to web app development. Other Google services could be used, too, like Groups. And, of course, developers are well acquainted with other forms of collaboration such as CollabNet.
If even for minor apps, services, or for prototyping development of subsets of large projects, this is all very compelling. I'm fascinated by how developers will use GAE within existing projects and processes. GAE will not be used in isolation, I suspect, but will be a powerful tool in the WOA quiver. And that may also prompt more use of GAE as the end-all, be-all for more and more apps.
I know a lot of people use Amazon as a test bed for their apps. Google App Engine will be very attractive for that too.
But what Google can soon bring to the table is an ability to put these apps and services in front of a ton of other developers and huge potential audiences of end users and consumers. Google has clout of scale, metadata and reach that Amazon does not.
Like others, I hope that Google adds more tools to Python on GAE, like Ruby. I also hope they find a way to port parts or all of the apps off of GAE. Perhaps for a cost, you could choose to not only deploy via the Google cloud, but perhaps get the basic script and code for extraction and use elsewhere, or for mixed-purpose development.