AA was waaaaaay late and so was AAA/Nyx.
I read that AA was finished earlier than expected, IIRC the project was started in late 1990 and the first working silicon was in early 1991.
If engineering had it's way (and the resources) the A3000 would have gotten AA from the start (or shortly after), about 2 years before it went into the market with the A4000.
When the A3000 was started AA didn't even exist on paper, in the talk Dave says they couldn't get any custom chip changes due to all the chip guys working on AAA. The A3000 was purely supposed to get money in to pay for AAA development & it couldn't be too good or it would compete with their coming soon AAA machines.
AA could probably have made it out a year earlier if the A3000+ had not been canned. But we're talking about late 1991 instead of 1992. That does seem to have been a management issue, however company politics are never simple and it wouldn't surprise me if there were people in engineering that were throwing in hand grenades because they didn't want AAA resources to be cut.
We know that engineering wasn't cohesive. The C65 project was one engineers dream to build the ultimate 8 bit computer, because they didn't like the Amiga. There were a few people who did all the stuff that got sold and there were a ton of others who spent more time creating things that were never going to happen. The view that management sucked and engineering were perfect is overly simplistic and not true.