It is all marketing BS, there is no need or use for collaboration like that.
Hehe, a little blunt, but has elements of the truth ....
Modern CADs are reasonably good at [manually] merging parts of schematics and layout, so different people can work on them at the same time. There still must be a manual merge process, no one will trust a complex design to a fully automated tool.
...
EDIT: I'm not excluding that there may be a place for a tool like that, but it must be absolutely seamless. And it must handle complex corner cases, like merging off line changes and a good way to detect and resolve conflicts. And nothing like this exists and nothing comes even close.
Even there, I'd stop short of "Modern CADs are reasonably good" in merge.
Mentor PADS can ASCII in on top of existing data, but the process is quite brain-dead.
It cannot see a "Same part, new XYRS", for example and apply that.
It barfs on routes replacing connections.
KiCad is a little smarter, their Append Board
can see "Same part, new XYRS", and it
can apply routes over connections, but still falls short in user control of this flow.
kicad cannot renumber on merge, for example & append is not there in all launch cases. (see below)
There is plenty of scope to improve even manual merge operations.
Both Mentor and Kicad lack any user prompts during the ASCII in merge, and one wrong choice can domino.
Some great replies here.
As I originally mentioned I was doubtful of the use case for collaborative PCB design at all, but now I see some people actually doing it (at least, with the SVN/Vault method).
There is a whole usage area you missed, which is "collaborating with yourself".
Before they let Marketing departments run away with the Enterprise buzzword fluff, EDA vendors should
first get the simple basics right.
A great litmus test, is
can you collaborate, even with yourself.This means good design reuse, and good cell append and renumber merge.
I'd say ASCII databases are also important here, as you certainly want SVN/Vault methods checking this, and there should also be NETLIST sanity checks built into the flows, along with smart enough back-annotate to the SCH, of any automated changes.