I have no great worries about the power architecture disappearing but I have large worries about the stability and reliability of a large system written in C++. I have worked on such things and they tend to be absolutely riddled with bugs.
It is extremely hard to write code in languages like C++ that is truly correct in all ways. e.g. most programmers effectively assume that they will have infinite memory or infinite disc space. There are a huge combination of ways in which some complex procedure can fail and programmers hardly ever cover them all properly nor can one ever be 100% sure that one has managed to test for them all.
If you have magnificent testing, static analysis and everything is perfect, then you still end up with a huge codebase for which each area is only understood by the people who wrote it. In some reliable area, people spend very little effort so the creators leave, or move and later when it needs an update, new people have to relearn it and now they discover that it's not as well documented or designed as everyone thought. It gets to the point where the existing system is complicated and insufficiently well-known by one person that each update is hard and slow - another risk. Especially horrible is when some major assumption about the design constraints ceases to be true or where one needs to re-architect to add a new capability that was not thought of originally.
Such a lot depends on having wonderful programmers and a management that understands and all of that seems very hard to reproduce to me - spending lots of money guarantees nothing.
Their advantages (over making phones which was my area) are that they can limit the number of hardware types they have to support at one time and that the number of "third party" modules (apps) is going to be limited so the code can be fixed properly and you don't have to trying to keep compatibility with your own old bugs.
Last edited by t43562; 17th Dec 2012 at 01:52.