The benefits of edge computing, low latency and backhaul efficiency are well understood. However, there is a cost to realize these benefits through small data centers. Hyperscale data centers are the backbone of the cloud. The cost structure and operation of these data centers is well understood. On the other hand, different requirements govern the edge cloud leading to different economics. Understanding these economics and monetization potential is fundamental to evaluating the business case for edge computing. The question becomes: is latency worth the cost of edge computing?
The economics of Cloud and Edge Cloud
The cloud players – AWS, Microsoft, Google and others – have been driving to reduce the cost of their services through data centers that scale upwards in performance while they optimize for power consumption. These hyperscale data centers are located where power and real estate are cheap. The proximity of these data centers to end users is not at the top of the list of requirements.
In the case of edge computing, or the edge cloud, proximity to users is the governing requirement. The location of the data center shoots to the top of the list of requirements. The cost of power and real estate become secondary consideration – relatively, depending on the level of distribution. Thus, the design parameters for edge computing are very different from the large data centers that run the cloud.
No single ‘Killer App’ for edge
Edge cloud economics on aggregate will trend to be more expensive than those for the large data centers. The cost distribution is highly variable depending on location. The requirements for edge computing are also variable, for example, there’s a need for different types of hardware acceleration depending on the applications that will run at the edge. This increases stress on the business case for the edge cloud. The revenue side of the profit equation is highly uncertain. The search for the edge ‘killer app’ is ongoing, but in reality none exist: it is a whole set of applications that will ultimately drive the market and justify the investment in the edge cloud.
The literature is full of examples of applications that will benefit from edge computing. Today, CDNs, IoT and AI are the leading applications. In this, the global cloud players have a head start over the Telcos who operate on national and regional basis. Applications such as autonomous driving and AR/VR will come one day, but it is still unknown how to monetize such applications. How far to distribute the edge cloud is a major question that has to balance the benefits, including monetization, with costs.
Organic growth of edge services
The uncertain monetization potential and high cost variability of edge computing, points towards organic growth of edge services. This will favor well financed firms who can weather the growth ramp for latency-intolerant applications to mature and be in demand. Incumbents have a play at this. A category of incumbents – the telcos – has been participating in trials to help define the business case. But the telcos have their work cut out for them.
One area of challenge facing the telcos is that their subscribers have a limited willingness to pay for differentiated services. Subscribers don’t value much additional services that they consider should be part of the connectivity package to begin with. That’s one reason why there’s revived focus on private enterprise networks, but the prospects for edge computing in this deployment scenario have their own constraints that will determined how fast they can be realized.
Edge computing is an open game
We are still at the early days of the edge cloud era. Multiples players are vying for leadership, each with its set of advantages and challenges. The cloud players are driving towards edge from their centralized, hyperscale data centers. New entrants are active in carving a niche by addressing emerging applications and challenges that the cloud players are not chasing. They have been leading on defining new technologies and processes for the edge. The telcos are also there with their extensive real estate assets. But while real estate is important, the telcos need to move quicker into this space along different fronts, for example software/cloud expertise and their corresponding operational processes.