Do investments in specialized knowledge lead to composite good industries?

There are more and more industries in which firms are specialized in the production of a component of the final good. This is especially true in high-tech industries. The basic question is why don't these firms merge ? We paradoxically show that industries which are typical candidates for composite good industries, are those in which high levels of investments in specialized and component based knowledge are a major source of benefits. In this important case, strictly complementary assets should be separately owned. The basic argument is linked to imperfect competition which changes the ability to extract payoff (power) and the effect of specialized investments on the quality of the composite good. Separate ownership in the case of at least 3 components both powers the individual incentive to invest and is stable with respect to unilateral deviations.