Sprint have recently announced that they will be implementing Qualcomm's BREW on some of their mobile devices. Sprint were one of the few CDMA based network operators to choose Java ME over BREW and even went so far as to invest heavily in their own Java application execution platform so this is a surprise move. In my eyes this is another nail in the coffin of Java ME in its current configuration.
While BREW has many shortcomings compared to Java ME there is one factor which really distinguishes it and that is the fact that it provides an entire end-to-end architecture with a built in application distribution mechanism (aka "application store"). At the moment it seems that solitary application execution environments just can't complete with an full end-to-end infrastructure. These ecosystems whose features include an SDK, application execution environment, search and discovery mechanism, robust deployment and provisioning solution, payment mechanisms, fragmentation management and administration are far more appealing to network operators than investing in relatively autonomous 3rd party application execution environments like Flash Lite, Java ME, .net CF, Python etc.
Basically if I was a JVM vendor I would be working hard to extend the on device Java Application Manager to become a sophisticated application distribution environment which includes the requisite backend infrastructure necessary to offer a full ecosystem. This would also provide large software vendors like Aromasoft, Aplix and Esmertec with new revenue streams beyond simply selling JVMs. Thoughts?
Update: I wrote something similar to this at the start of the year.


What got J2ME in trouble was its "silo" approach. This approach was due to Sun's philosophy of creating specs and APIs and JVM and let the ecosystem/vendors define the rest. In theory it sounds like a good "open" philosophy, but without E2E strategy it results and resulted in fragmentation. Isn't that ironic? Sun's idea was to be open, and openness meant way too many people involved and slowing the process down and also meant no unified E2E solution, hell, no E2E at all. I saw this happening in front of my eyes. I remember time ago going to some Sun folks about the importance of application discovery (and security) as the 2 top issues to address, but they didn't get it and were stuck on "letting others define that". That is when I had the idea to do YouApp. In any case, sometimes even having all the E2E pieces is not sufficient - this must be done at the right time with the right attitude or it will be ignored.