Edge computing will see new workloads, strong growth, IDC predicts

Even as old and new IT workloads migrate to cloud, enterprise IT shops are seeing more computing resources moving closer to where the data is created. 

Under the banner of edge computing, this trend promises greater distribution of computing power, and it’s become an area of interest for telecoms, established IT houses and others, including the top cloud providers whose machine learning products are increasingly appearing in “edge AI” versions.

Edge computing shows aspects both familiar and new, analysts told attendees at this week’s IDC Directions 2022 event in Boston. It has a likeness to client-server computing that preceded cloud architecture. But many of the workloads involved are straight out of advanced computing’s future playbook.  

In precision farming, drone deployments, public safety applications, freight monitoring and other front-line use cases, data is the driver that moves intelligence and storage to the edge, according to Jennifer Cooke, research director for edge strategies at IDC.

“One of the common guiding factors in use cases is that there are massive amounts of data. We need edge to really make sense of it,” she said in an IDC Directions presentation that described edge computing as “the runway to digital-first operations.”

While there have always been applications outside of the data center, new challenges await, as operations today are far more digitized, Cooke indicated. As well, edge applications are far more impactful than in the past.

“Edge today is not like the edge we used to have. It’s supporting workloads that are not just about making it easier for the bean counters to do the bean counting,” she said. 

Building out new networking infrastructure, monitoring edge workloads and ensuring resilient and low-latency operations are among the challenges ahead for edge, but growth is on tap. In January, IDC said it expects worldwide edge spending to reach nearly $274 billion in 2025, growing at 18.7% CAGR over that period. 

Meanwhile, back on the cloud

Most of the growth IDC foresees for edge will take the form of services. Such projections have clearly gained the attention of the companies that have been gladly taking IT data loads up to their clouds for some time. For example:

• Amazon offers AWS IoT Greengrass as an IoT edge runtime and AWS Outposts for edge and on-premises infrastructure and services.
• Microsoft has released an Azure Stack Edge gateway, Azure IoT edge devices and, recently, an Azure MEC (multiaccess edge compute) platform.
• Google’s efforts have centered on a Distributed Cloud Edge infrastructure and services offering. 

Underlying their different products is a common drive to push their own cloud architecture out to the edge, according to Charles Fitzgerald, a Seattle-area angel investor and former platform strategist at Microsoft and VMware, who recently spoke with Venture Beat.

“Microsoft has had an embedded business for a long time, and they went there first. Amazon poo-poohed it — then they got religion as you see with Outposts and Greengrass. Google’s further behind, without the full breadth of services to match what Amazon and Microsoft can offer – but to the degree that they want to compete for the same enterprise business, they will have to do that too,” he said. “They all want their architecture to be everywhere.”

Edge computing looms as an important next step for AI and analytics applications that have been a target for cloud migration, but which have clogged up some of the data pipes to and from the cloud.

“There are many AI use cases where there’s just so much data that it doesn’t make sense to round-trip it,” Fitzgerald said, while noting that people are still in the process of figuring out  which of those AI use cases promise the best return on investments.  

It is not yet clear that the big cloud players’ software architectures can be packaged appropriately for edge use cases. Other “non-hyper-scalar cloud players” such as Cloudflare, Fastly and Akamai have  software-based content delivery networks that can become competitive platforms for building edge applications, Fitzgerald suggested. 

Edge computing covers a lot of ground

Today, edge nodes take many forms, Ian Skerrett emphasized in an email conversation with Venture Beat. Skerrett is vice president for marketing at HiveMQ, which provides a messaging platform for connecting edge devices to the Internet.

Skerrett views edge as including intelligent sensors connecting directly to the Internet; edge gateways, that are similar to industrial computers and controllers; edge data centers that aggregate and process local data from edge gateways; and network edges that telcos focus on now as part of high-profile 5G roll outs. 

“The main point is that there is a huge change coming in corporate data and most of it will be coming from the edge,” said Skerrett.  “This will dramatically change how we collect and process data.” 

He marked V2X, or Vehicle to Everything communication, as a major new edge use case, and said industrial firms, telcos and cloud players will work to define this space, with no clear leader yet emergent. 

Many views mark the edge 

In a way, edge computing looks a bit like other hot-button memes of the day, such as the metaverse and Web3. Each offers greater distribution of computing power, while complementing now-established cloud computing architecture. All these technologies are multifaceted, and can carry different meanings in different contexts.

In fact, diversity in edge architecture is likely to be a fact of life, Jennifer Cooke told her IDC Directions’ audience. 

She said edge computing choices could be quite different for different parts of the same organization. Moreover, the definition of what is best in edge architecture will change as firms work in different parts of the globe, or as they become dependent on different business partners over time.