Rushing To The Edge

Virtually every major tech company has an “edge” marketing presentation these days, and some even have products they are calling edge devices. But the reality is that today no one is quite sure how to define the edge or what it will become, and any attempts to pigeon-hole it are premature.

What is becoming clear is the edge is not simply an extension of the Internet of Things. It is the result of the IoT, a way of managing and utilizing the vast and exploding amount of data produced by sensors from connected devices everywhere. But it also is its own distinct segment, even though it hasn’t been fully articulated.

The big challenge at the edge, at least at this point, appears to be more about figuring out how to parse large quantities of data intelligently. Whatever system or action requires an immediate response needs to be processed as close to the source, while longer-term trend information can be sent to the cloud. And whatever data is deemed useless needs to be discarded quickly, because it costs money to move, process and store that data.

This sounds straightforward enough, but it’s creating extreme angst among tech companies of all shapes and sizes because at this point no one owns the edge. In fact, the edge could upset the balance of power among the largest tech companies in favor of more localized processing and pattern recognition, using as-yet undefined instruction-set architectures and customized AI training and inferencing algorithms. Put simply, this is a brand new and apparently vast market opportunity, but it’s also one that can grow rapidly at the expense of existing markets and investments.

There are a couple of key reasons for this. First, processing closer to the source of the data is faster and cheaper. In fact, there are efforts underway by startups and established companies to process that data in-memory or much closer to memory, starting with battery-powered devices. This is easier said than done, because data must be tagged somehow so that local processors can grab only what is needed and identify what needs to be sent on and what can be purged.

While the EDA industry has focused on designing, verifying and debugging a chip, this adds a whole new level of complexity to the process. Understanding how to pattern data is an AI issue, and AI systems today are basically black boxes. How these systems ultimately behave isn’t always obvious. This presents a problem, because for the edge concept to work—adding precision wherever it is necessary and less when it is not needed—will require transparency. Behaviors need to be predictable to automate them, and at this point that’s difficult because there is no clear strategy or architecture and so much churn on the algorithm side.

Second, processing more data locally helps alleviate growing concerns about privacy and security. There is a growing groundswell to keep data private, driven initially by Europe and followed closely by North America, particularly in states like California and New York. Pushing that data to a public cloud where it can be mined for a variety of unknown reasons, or where data may leak or ultimately be hacked, has unnerved enough people to the point where this is now becoming a political issue. Keeping data local also makes it more difficult for hackers to collect financial and personal data on nearly 150 million people with one attack, like the 2017 Equifax breach.

There are a lot of moving parts in the edge equation, and so far there are few rules other than the laws of physics (and even then there are sometimes workarounds, like immersion lithography). Still, no matter how it ultimately shakes out, the edge is a massive opportunity—and one that will require deep knowledge of everything the semiconductor industry has learned so far, and an enormous amount of new technology that still needs to be developed. This is system-level design at a much broader level, and it should keep the design industry very busy for the foreseeable future.

Related Stories
Edge Knowledge Center
Edge top stories, videos, white papers, and blogs
Designing For The Edge
Growth in data is fueling many more options, but so far it’s not clear which of them will win.
Racing To The Edge
The opportunity is daunting, but so are the challenges for making all the pieces work together.