Some new and emerging markets are very hard to quantify in terms of growth. “Edge computing,” for example, includes a range of market segments, from devices and sensors to apps, computers and servers to remote “computing as a service” purchases.
And since many existing categories of products can plausibly also be considered to engage in edge computing, one has to separate functions. Content delivery networks are an existing form of edge computing. But so are personal computers and smartphones.
We face similar problems trying to quantify the growth of the “internet of things.” Even restricting the definition to include only things that communicate over the internet, many “things” already do so: smartphones, PCs, gaming devices, TVs, printers and so forth.
Presumably the big nearer-term growth of IoT comes from newer use cases such as industrial sensors. Eventually all sorts of “things” might connect using the internet.
The issue with market forecasts for edge computing and IoT is that so much existing economic activity and sales are for products that have internet access functionality. By that standard the IoT market--just for devices--is substantial. In 2022 alone possibly 1.3 billion smartphone units will be shipped.
Some forecasts even include instances of devices connected to any LAN and WAN as part of the IoT market. The methodological issue with that approach is that it simply takes a lot of existing markets and essentially rebrands them, often mixing such rebranded activity with incrementally new use cases.
Edge computing is going to pose the same sorts of questions. On-device computing is probably not what most people think of when evaluating edge computing. Whether it makes sense to rebrand premises computing of any sort as “edge” computing also is an issue.
By definition, enterprise computing facilities “on the premises” are a form of edge computing. Perhaps most would agree that edge computing centers on the use of more-distributed computing someplace within a metro area.
But even there, cloud computing service suppliers are developing “edge computing” systems that can be deployed on any enterprise’s property. Such systems tend to run the same protocols as the remote hyperscale data centers provide, but execute code locally.
How we account for such systems is a question any forecaster has to answer. What counts as “edge” computing?
No comments:
Post a Comment