I'm still not entirely convinced of this.
It seems to me that fleeing, (i.e. infected ships exodusing known space) is a viable survival strategy. Possibly centered around one of the "smarter, less insane" ships. They would leave to protect themselves, especially after seeing what the humans et.al. were doing to virus infected vessels, fleeing would be an idea that some infected ships would come up with and impliment.
Humans aren't really doing anything particularly brutal to Virus ships -- at least nothing that Virus ships aren't already doing to each other, which is blowing each other up and launching infection attacks on each other.
I think you're right,
some Virus ships would obviously flee, especially if things started to look desperate to the point of extinction, but I disagree it's something they'd do initially - the old Imperial domains is where stuff is known to exist - things that the Virus need to survive and prosper as a life-form. In "wilderspace" beyond the knowledge of the gaming universe, things are a lot less certain. Societies do move like that historically, but usually only because of the most dire threats.
In other words, Darwin works. The Vampire fleets would eventually die out, unless they got smart and in some way or another become a non threat. Among the ways of becoming a non-threat are fleeing, and co-operating with humans.
This assumes the Vampire fleets, and by extension, the Virus AIs are somehow stupid, which I think is indeed the implied GDW canon on this. While I liked the Virus AI catacylsm plot (because I thought it was interesting, despite how it worked contained a large bunk-factor), I never really could bring myself to buy Nilsen's presentation of the Virus in-game. From what I followed of Nilsen's reasoning, the Virus AIs were essentially a stupid, incomplete, and parasitic thing, not really alive or worthy of respect accorded to sentient beings simply because they weren't - they were things going through the motions of being alive.
Nilsen softened (and partially reversed) his position in
Vampire Fleets, but even then I think Nilsen tended towards the "coercion or cooperation" binary thinking - either Virus AI were rational and therefore "just like us" (ie; Sandman "him"-self) or they were insane and their thinking (as it were) was totally unfathomable, even to them - they only operated by forcing humans to do repairs on them before blowing them out of the airlock or something.
I think there's any number of viable "middle-roads" in this case - sane, but alien, AIs. A Vampire could, for instance, find where two human tribes are fighting, use its lasers to blow away the stronger tribe in a brilliant display of god-weapons. Then it could come over to the weaker tribe and: Demand fealty (or I blow you away, too) ... or "I helped you out, and I can continue to help you out, but you need to do something for me, too." While the second could be seen as cooperation, I think the population would still be keenly aware there's still the "or I blow you away, too" somewhere in there - there is an element of coercion in there.
The basic problem (that all but the most short-sighted AIs would realize) is that their life-mode is basically parasitic until they do something about it. That is, they can only propagate (and therefore provide a bulwark against extinction) by utilizing what someone else made - be it robots, data systems, or ships. While the AIs are trapped this way, I think they're sort of stuck in the wastelands of the former 3I since that is where the known stocks of ships, data systems, and so on are.
More far-sighted AIs might decide to start building something new - perhaps coerce or convince (coerce) humans to build a nucleus of robotic workers. The robots could in turn build more robots, who could then build (or rebuild) starports, factories, and so on to provide further systems for the Virus to propagate into. Of course, this would again have the effect of encouraging the AIs to settle down and defend what they've built - again, no encouragement to just leave, especially now that this AI (or more likely group of AIs) would be in a position where they'd be growing stronger over time, not weaker due to irreplaceable losses in ships or equipment.
Given humanti's weakness right up to like 1200, those AIs that have given thought to this problem are in a stronger position - you rarely abandon your homes outright when you're in a position of strength, and you'd think that it'd be clear to AIs by around 1180 or so that unless they do something proactive, they're stuck in a fishbowl of limited size, and the water is just evaporating away. That would still give the AIs a 20 year head start on the humans by the classical TNE timeline (I would argue that the AIs would probably realize sooner and that humans would rebuild faster but not rocking that boat too much I'll go with 1200).
I think only after the AIs have a "closed-loop" society (ie; the Vampires can meet all of their own needs indefinitely without human labor to exploit) would they consider fleeing known space, and only then if they faced cataclysmic losses.
conception of the virus
o: 
