A handful of reasons (disclaimer: painting with very wide brush here, in the hope that you get the strategic picture)
* The desktop market is fragmented. Pick your poison: walled garden with crappy devtools, OSS fanatics who will not pay for software in a million years; or warezing pirates. Note that whichever you choose, your app will still end up on torrent sites ~2 days after release anyways.
* The cross-OS development tools' end results are usually sub-pair, even compared with javascript.
* Desktop usage is, and has been stagnating for a while now; while mobile usage is growing at an incredible rate. Mobile native market is also highly fragmented.
* The lowest common denominator across all of these markets is the web: develop it once using responsive design, run it on everything for at least the next decade, or so.
* People have been voting for this platform with their wallets for a while now; specifically by being unable to pirate web apps.
* The stakeholders in selling AWS are: developers (for whom command line & APIs are the critical part), and business decision makers (who are looking for easy leverage). Neither target group is particularly interested in desktop apps.
* Specifically regarding to their propriatory tech (eg dynamoDB), developing a third-party desktop tool for it is a high-risk-no-reward endavour: you either get traction, in which case Amazon will gladly oblige with an app of their own; or it won't, in which case you're out with $lots$.
Joel Spolsky put it best: "Filling little gaps in another company's product lineup is snatching nickels from the path of an oncoming steam roller." ( http://www.joelonsoftware.com/items/2009/06/10c.html )