The cause of transit price drops is another common discussion at peering fora so I will share some of the theories put forth.
As you can see by the graph, the trend is unmistakable, and there are at least five fundamental forces at play here:
As the price of peering ports decreases, the cost of traffic exchange drops. This leads to the ability for smaller ISPs to exchange traffic directly with one another at a lower cost. Therefore, they will able to offer a lower price to the market. We see this working across regions as well, where ISPs build into an adjacent country and establish a peering and transit purchase presence there. In this way, a foreign ISP can also offer lower transit prices to customers in the home market.
You can see this effect in the graphic below.
Internet Region A is a well functioning Internet Peering Ecosystem, with a strong regional IX. Everyone is peering there, both domestic traffic and international traffic. As a result, the locals get really cheap transit pricing.
Internet Region B has pricing pretty close to the price of Internet Region A because it is cheap and easy to build in and pick up that pricing across the border. And so on.
With long-haul wave division multiplexing costs going down, and metropolitan area dense wave division multiplexing costs going down, the cost of hauling traffic across a region allows ISPs to offer lower transit prices to its customers. The economies of scale kick in here as well, where those who have very large amounts of traffic and transport across the region can leverage the huge economies of scale. For these ISPs and large scale network savvy content providers who have deployed wave division multiplexing equipment, the price per bit is pretty low
Within the last mile, cable and DSL technologies enable consumers to consume much more traffic than in the past, enabling last mile providers to purchase much larger quantities of traffic at a lower cost per bit. The eyeballs are driving innovation and economies of scale here, and the end result is that the eyeball networks are much more attractive peering candidates.
Today, Paid Peering is offered at a price that is similar to the price of Internet Transit. As a result, more companies are looking at paid peering as a complement to their Internet transit purchases. As paid peering prices decrease, this may apply additional downward pressure on the price of Internet Transit. Others may argue that the paid peering approach actually stabilizes the price of Internet Transit, since the handful of eyeball network control the price of access to that which they and they alone control. Why would they drop the price of paid peering?
We have seen over a decade of decline in the price of Internet Transit. There is a expectation that the price of Internet transit will continue to decline by order of 20 to 30% per year. Many suppliers have attempted to hold their ground, and not compete on price. These companies lose business. When the salespeople go back to the office and say that they lost the business to a more agile competitor, these companies invariably buckle.