Specialization and increased overall efficiency due to improved technology that you are talking about here is handled in Aurora by the techs that improve research efficiency.
If you have the same level of technology, same level of equipment, then having 10 labs working on the same tech instead of 1 should not really be 10 times more effective.
CERN today has around 15000 employed (also including visiting scientists & engineers), compared to 1 million for a single research lab in Aurora. It is hard to imagine any situation where 1 million is not enough for even the largest research projects and experiments needed. Also remember that we already are very specialized due to the nature of Aurora ship building divided into components. Instead of researching an entire Carrier with everything on it you already divide it into dozens of projects. One research project can for example be researching a new missile engine component that will be used inside a missile, inside a launcher, inside a fighter, inside a carrier. It's hard to get more specialized then that.
It also seems you are mixing up alot of concepts here. Economy of Scale and Factory production then sure of course you are right that having one big assembly line instead of 10 small workshops employing the same amount is more effective.
But it doesn't work that way for Research, RnD and Product development. Every day you can see big companies buy up small upstarts with innovative technology and ideas, simply because they can't come up with those ideas themself, no matter if their RnD department is 1000 times larger. In a big company waste and as you put it C3 prevents you from seeing the big picture and making the big science breakthroughs.
If you ever see a big company buy a smaller rural workshop, it is because of their ideas, patents and innovative approaches, not because of their production assets.
I would be leery of assigning numbers too closely to how work is done. The in-game numbers might include significant portions of the rest of the supply/research chain, such as a copper refiner which gets copper to a superconductor manufacturer which gets superconducting elements to the engineers who actually build and set up everything. I don't know the source of your number but I doubt it accounts for such things, while the in-game workforce very well could (and reasonably so, considering you can pick them up and move it around it has to be mostly self-contained).
In any case, I would be especially hesitant to try and predict the future sizes of large research collaborations. I doubt anyone can predict future research project sizes now (although we are rapidly approaching Aurora's default start date), much less once we throw in trans-newtonian materials, jump technology or alien artifacts. Can anyone guess how many people it would to support a long-term manned expedition to study jump points, including planet-bound support staff, ship manufacturers, probe constructors, launch facilities, upkeep and researcher training? That's just to collect the data, much less analyze and utilize it. For comparison, the Apollo program employed about 400,000 people at its peak according to NASA, and that was in the 60s/70s without new TN materials involved.
If research is so much better with small firms, why do DuPont and 3M exist? Why is NIST around, or Argonne, or CERN? I don't think these monolithic companies/labs grow large and then completely ossify, feeding themselves solely through acquisitions? You should check out Bell Labs especially, a longtime research area owned by various companies, responsible for such modern items as transistors and CCDs plus fundamental physics like the fractional quantum Hall effect. It's a shame Alcatel-Lucent shut down their fundamental research division considering how much came out of there. There are plenty of other examples if you would like more.
Research is not a zero-sum game. The fact that companies acquire others does not mean the big company failed, only that they saw something they needed in the smaller lab. I have never stated that a small lab could not produce useful research. I merely stated that large labs are at least as good as small labs on average. I think there is a wealth of evidence and specific examples to support that assertion, including many projects that would be outright impossible without large teams of researchers.
*edit* I should add that it might be best to split a research discussion off so as not to clog the suggestion thread, especially if we keep up the debate.