But even in a market where you can easily switch one supplier for another, there often remains a helpful factor for grouping - everything costs less in large quantities ;) And many administrative costs can be reduced. Not to mention that perfect competition would, strictly mathematically speaking, drive profits towards zero...
It's clear that some markets are like this, and some aren't. It's not at all clear to me that most markets aren't. In fact, i believe the opposite - most markets are like this. I believe that in most markets, economies of scale, cost of entry, and the coercive power of large participants outweigh the incremental efficiencies of new innovators.
If there is any logic or data which supports the contrary position, i would be interested to hear it.
My desktop OS example requires some clarification- there were plenty of things Microsoft did in the '80s and '90s where they exploited their position in the market to increase market share. Per-processor licensing fees are a prime example of that- a PC vendor that wanted to license a Microsoft OS would have to pay a licensing fee for every computer they sold, regardless of whether or not that computer had an MS OS on it. However, one could argue that this was a consequence of MS already having a monopoly (50% market share can be considered a monopoly in some markets), and they were exploiting that monopoly to become even more deeply entrenched. I honestly think that the desktop OS market in the pre-web days had very strong tendencies towards a natural monopoly regardless of whether or not any given player in that market was engaging in unfair business practices to achieve that monopoly.
Any time discussion about competing products revolves around which product is ultimately going to "win", there is an implicit assumption that the market those products compete in tends towards natural monopolies. It may not be true, but the people participating in those discussions think so even if they're not fully aware of it.