Edge computing (also called “fog computing” or “on-device computing”) moves computing from centralized clouds out to decentralized devices (such as smartphones, cars, machines, etc.). This holds big advantages for applications such as autonomous driving, for instance: when a car “sees” a stop sign, it should break right then, and not only once it has received the answer from some centralized cloud-based system.
In order to get a 360° view on edge computing, I would probably ask questions such as these:
- What venture-funded companies are there, what do they do, who invests in them, and how much?
- What are applications and markets for edge computing?
- Who are relevant “big players” in industry?
- What is happening in edge computing R&D?
I can use Mergeflow to address these questions (because luckily, my boss bought all of us subscriptions to Mergeflow).
So far, FogHorn has received the most funding. Their “flagship product” is called EdgeML. EdgeML is a software solution for running complex event processing, machine learning, and other analytics “on the edge” (i.e. on-premise, on machines, vehicles, etc.).
Swim.ai is the latest edge computing company that received funding. On July 18, they received $10M in Series B funding. Swim’s product, SWIM EDX, is an edge-based software solution for performing data stream analytics in real-time.
As edge computing company investors, Mergeflow identified, among others, Almaz Capital, Cambridge Innovation Capital, Intel Capital, Lux Capital, and Robert Bosch Venture Capital.
Applications and markets
Next, I used Mergeflow’s Market Information Extraction tool to get at applications. For my search “edge computing OR fog computing”, this tool extracts relevant market information from news and other sources. Here is the result, sorted by estimated market size in 2018:
I excluded “edge computing” and “fog computing” market estimates because I wanted to get markets where edge computing plays a role, i.e. applications of edge computing. As you can see, there are a number of applications that make sense once you see them. But they are nontrivial if you had to identify them upfront. For example, “IoT in oil and gas” has obvious edge computing applications (just think of all the robots involved in oil and gas), but I would not have thought of this application upfront. Same goes for applications such as “fleet management” or “warehouse management”.
Relevant “big players” in industry
Are there any, given that the topic is relatively new? Well, here are the top 40 identified by Mergeflow:
Overall, if I had expected to see big players, it probably would have been the ones shown in this tag cloud. So, no big surprises here.
But then I also looked at patents. And there I was a bit surprised to see that patents go back quite far, to the early 2000s. For example, in 2004, IBM filed a patent titled “Application splitting for network edge computing”.
Research and development
I mostly looked at applied research. In order to bridge research and applications, I used Mergeflow’s Patent Class tool. This tool assigns patent classes to non-patent documents (such as research papers), and I can then use the patent classes to zoom in on particular applications. For example, one was “aircraft”. This turned out to be about cellular-connected UAVs, e.g. this paper, which just came out a few weeks ago.
“Meteorology” was another application. This turned out to be a paper about applying edge computing to precision agriculture.
Yet another application I found interesting was “power network operations”. There, I found a recent paper that applies edge computing to smart grid architectures.
How long did this take me now?
Except writing this article, getting my 30,000ft overview of edge computing took me about 20-30 minutes. Without analytics, just relying on conventional web and database searches, I think that it would have taken me a few days at least.