structural flows and vertical integrations

Just like there is a deep relation between the network structure of neurons and the subsequent dynamical flow of information around them, there is a deep relation between market structure and profit flow. 

Think of each supply chain as a vertical stack where each stack manufactures the inputs used at one higher stack. For instance, one stack may have ten different type of inputs from the stacks below itself (in different supply chains) and each input may be generated inside either a monopolistic, an oligopolistic or a competitive market conditions. (Supply chains intersect at stacks where more than one input are used to generate a single output.)

Profit of the whole supply chain is capped above by the consumer value assigned to the final product and below by the total cost of production of the entire chain. Now think of the thickness of each stack as the amount of profit captured by that particular stack. Visually, what happens over time is that the whole chain dynamically flexes as the total amount of profit and its internal distribution evolve. These dynamics are mainly shaped by the (ever-changing) market structures at each stack. For instance, a monopolistic stack can easily suck profits from the stack above by charging a higher price for its output. (Of course, a higher price also results decreased demand but the monopolist has the luxury to optimize this trade-off.) A monopolistic stack can flex its muscle downwards as well if the output of a stack below is not utilized elsewhere in the economy.

Each company exists inside a supply chain and one reason why companies try to become vertically integrated is to minimize the uncertainties entailed by the ever-evolving profit distribution across the chain. Of course, in practice, most vertical integrations are done for the wrong reasons. For instance, companies often extend themselves into neighboring stacks that have long become commoditized and effectively stabilized into a no-profit equilibrium due to extreme competition. Also, generally speaking, the uncertainties are best addressed at the financial (investor) level rather the operational (manager) level. That is basically why investors like to conduct portfolio optimization with pure plays only. (Part of the uncertainty that can never be eliminated by portfolio optimization methods is called systemic risk in finance jargon.)

Companies routinely launch attacks against monopolistic stacks by funding open source projects, antitrust lobbying activities, meanwhile merging with other behemoths in order to become monopolies themselves. They also occasionally reach out horizontally across to the neighboring supply chains. For instance, in order to increase how much the stack above can pay for their output, they try to find ways to decrease the prices of the complements of their output. (Remember the stack above needs to combine several inputs, including your output. The less it pays for the other inputs, the more it can pay to you.)

When a pioneer startup is building an entire new supply chain from scratch, it has no choice but to own the entire chain since everyone else is far behind it in terms of understanding the contents of what is emerging. Of course, no single company can do a high-quality job at each stack in the chain. Sooner or later, other (more-focused) startups join the game by claiming certain stacks. The pioneer startup has the first-mover advantage of having visibility over which stacks are worth defending. (Of course, it is never easy to focus an already over-extended company.) As the new supply chain matures, the number of players at each stack proliferates and the robustness of the whole supply chain increases.