Google, Facebook and Microsoft weren't invited, except for a contingent from Microsoft Research, because there's simply nothing those entities can add to the discussion at this stage. The situation is somewhat comparable to engineers often not being invited to the conferences of materials scientists.
It's generally rather pointless/inefficient to mix the application of research with pure research. You've got to have someone define the properties, characteristics and behaviors of a thing then 'box it up' for use (that'll be the boffins). Then you've got to have someone that the stuff in that box and see what they can do with it (that'll be the engineers). Some overlap is inevitable, but keeping research and application on seperate sides of the aisle was long ago determined the best way to go about it all.
Without that separation you get boffins trying to play engineers and businesspeople. You end up with a Graphene scenario where the research dies in its crib because the boffins have less than zero understanding of commercial application dynamics. Suffice it to say, crossing the streams is ususlly a bad idea.
Microsoft* et al, the engineers in this example, aren't going to be helpful at this stage simply because there's nothing for them to apply. Any input they did provide is at least a few rungs down the ladder from broad research and will be tainted by their current needs, not the needs of an entire sector.
*Not Microsoft Research, which is actually a legitimate research group.
Biting the hand that feeds IT
