I too am seeing a general trend away from Ada for new product development.
It may be due to a confluence of skills availability and the drive towards
OTS software (from commercial, govt., and public sources).  Most military
aerospace systems do not have stringent certification requirements, or if
they do, it applies to only a fraction of the total software.

There is a tremendous desire to raise productivity and reduce costs.  Some
may cite a lower rate of defect injection for one language over another.
However, by their actions, I conclude that most companies do not believe
that defect rates are a significant cost driver.  The popular hypothesis is
that using OTS software is a key to achieving the desired improvements.
Theory would be too strong of a word;  I have seen no data to suggest either
improved productivity or reduced lifecycle cost from using COTS, but that's
today's silver bullet.

The drive to use OTS software has created a need for a new kind of domain
knowledge, that of OTS products and the technology to bind it together with
custom code.  I know from experience that this is often dependent on
specific combinations of compilers and hardware.  Most of the OTS products
are written in C++/Java, and those are the languages favored by the OTS
domain experts.  Furthermore, large companies do not like to be dependant on
a few individual with "esoteric" skills.  They are more comfortable with
large labor pools from which to draw on.  Large labor pools give more
flexibility to quickly ramp up for new contracts.  Conversely, in an age
where companies keep telling us that our careers are our responsibility, not
theirs, most engineers look for the largest labor pool in which to swim.

Just my two cents, and certainly not those of my employer...

        - John Harbaugh