How greatly are transaction/switching costs affected when segmenting projects in 6-12 month intervals? I would guess, that when you're throwing senior engineers at projects, the switching costs are non-trivial, but low. It's probably an acceptable tradeoff to pay when developing specialized projects you don't expect to carry over to future ones.

There's also the argument of developing a labor force to compete with other enterprises. However, if the best engineers are in sufficiently high demand, it may be more economically rational for them to move between projects every 6-12 months, gradually increasing their pay-rate and improving their portfolio after completing each one. If this is the case, then companies would be forced to hire top engineers on a contractual basis (or pay them a significant markup annually), purely to compete in the labor market for top talent.

Or even, at the very least, by switching projects internally.