Through-life growth and adaptation is essential, I agree.
Unfortunately, the F-35B differs from most aircraft in having no margin for weight growth at all, absent a major uprating of not only the engine, but also the lift fan and transmission, which would in turn be constrained by jet effects on the ground. It will also be costly because much of the work will be confined to 400-some B models.
There is also a problem with growing stealth platforms, which is that you can't add capability through pods, or even add antennas or apertures without installing cavities and stealth-compatible radomes/windows. This can be done, but is expensive. Of all the stealth aircraft deployed so far, the F-117 is the only one to have had major upgrades, and that mostly to replace stuff that was initially borrowed from other aircraft.
Now consider that the F-35 architecture is just not the way you'd do things today, being built around two massive central processors. A problem with this is that you can't just port technology over from other programs (newer AESA tech for instance).
And, sure, you can do a lot with software - but S/W and hardware go hand in hand.
Even the modest upgrades to the F-22 (not even AIM-9X yet) have cost a bomb.
As to JTO's point, Raytheon is now preaching that line on-the-record. RCS is locked-in, but detection is progressing quickly.
Last edited by LowObservable; 18th Nov 2013 at 11:47.