AI, 5G and IoT spur edge data centers

The volume of data generated at the enterprise edge by IoT devices and other sources continues to grow at a rapid pace. By 2025, IDC estimates that there will be 41.6 billion IoT devices, capable of generating 79.4 zettabytes (ZB) of data.

As data multiplies, enterprises are recognizing that it’s preferable to process that data at the edge for real-time decision making, rather than the battle latency and congestion issues associated with routing that data back to cloud or on-premises data centers. PricewaterhouseCoopers (PwC) says the edge data center market will grow to $13.5 billion in 2024, up from $4 billion in 2017. Looking ahead, MarketsandMarkets predicts the edge data center market will grow to $29.6 billion by 2028.

The popularity of edge data centers has attracted a number of vendors, including startups such as DartPoints, EdgeConneX, and Zella DC. Incumbents American Tower, Dell, IBM, and Schneider Electric also provide edge data center solutions.

What are edge data centers?

Shifting from a centralized to a distributed data center strategy runs counter to the trend among hyperscalers, who are building massive data centers that can house 10,000+ racks and have a capacity of more than 80 megawatts (MW). By comparison, edge data centers are tiny, with a capacity between 500 kW to 2 MW. What they lack in size, they make up for with versatility. They can be deployed practically anywhere: on rooftops, in extreme locations, and in rural, underserved areas.

Modular edge data centers can range in size from a closet to a shipping container to a garage. Their turnkey design means that things like security and cooling are built in. They can be up and running quickly, and their standardized design means an enterprise can order in bulk and deploy compatible, interoperable infrastructure across its global footprint.

Three key drivers of edge data centers

Edge data centers are catching on for a number of reasons, but the key drivers align with several other big tech trends: IoT, AI, and 5G.

1. Internet of things

In industries such as health care, telematics, and manufacturing, both the number of endpoints creating data and the need for real-time processing make sending data back and forth to centralized data centers a costly proposition. According to Gartner, roughly 45% of IoT data is already processed at the edge, and that number is expected to continue to rise.

“As the volume and velocity of data increases, so too does the inefficiency of streaming all this information to a cloud or data center for processing,” said Santhosh Rao, senior research director at Gartner.

As more data is generated and processed at the edge, machine-to-machine (M2M) communications threaten to overwhelm traditional networks. Thus, edge data centers not only boost real-time IoT use cases, they also protect traditional networks from a flood of M2M traffic.

In manufacturing, for instance, the number of connected sensors and machines is expanding rapidly, and those devices are generating huge volumes of data. With edge data centers, manufacturers can bypass latency and congestion issues by processing data locally, which enables real-time decision making for production optimization, preventative maintenance, and the proactive correction of problems that could undermine just-in-time supply chains. Other verticals that are early adopters of edge data centers include education, health care, financial, and smart cities.

2. Artificial intelligence

AI at the edge enables use cases such as facial recognition in crowds, autonomous vehicles, and real-time health monitoring. However, according to research from JLL, the rapid adoption of AI, along with ongoing cloud migrations, is outpacing the ability of data center providers to keep up.

AI applications require significant computing resources to generate and consume massive amounts of data. The trouble is that the latency involved in transferring data back and forth between centralized data centers and the edge eliminates real-time capabilities, and the bandwidth needed for this constant flow of data increases costs.

According to JLL’s 1H 2023 North American Data Center Report, current and near-term data center capacity in major markets is not keeping up with demand, as industries such as finance, health care, and manufacturing gobble up available data center capacity.

Edge data centers could serve as a way for not just enterprises, but also service providers and colocation providers to roll out capacity quickly wherever demand emerges, while also using them to plug coverage gaps in smaller markets.

3. 5G

At first glance, it may appear that 5G could compete against edge data centers, but it’s more likely that the two will complement each other. According to PwC, 5G is a key driver of “decentralized small cell network of edge data centers.” This network of edge data centers provides low-cost, low-latency support for use cases with large numbers of devices, such as sensor networks in factories and smart cities.

The same issues that bedevil real-time IoT use cases (latency, congestion, backhaul costs) apply to 5G, and edge data centers could serve as a convenient way to overcome those issues.

Will RedCap supercharge the demand for edge data centers?

The arrival of 5G RedCap (reduced capacity) could add further incentive to deploy data center resources at the edge. RedCap was originally introduced in 3GPP Release 17, targeted at constrained devices that currently rely on LTE Cat-4 or other legacy connections. While full-blown 5G includes a ton of advanced features, it’s overkill for many edge devices, and those features come at a cost.

While RedCap has not yet been approved (it’s expected to be approved sometime this year), pre-standard RedCap gear is already starting to hit the market, with AT&T, Ericsson, and Nokia, among others, already testing RedCap-compatible gear. (See also: Will RedCap 5G spark IoT gold rush?)

RedCap offers better peak data rates, broader coverage, and lower latency versus 4G LTE and especially 3G, which is still prominent in many parts of the world. It is also designed to accommodate constrained devices with simpler radio designs and lower power consumption needs. According to research from Ericsson, relying on RedCap (vs. 5G) could bring the cost of IoT devices down by 65%.

Another advantage of RedCap is that it could help unify connectivity at the edge and beyond. “Today, we almost have too many ways to connect,” said Jason Leigh, Research Manager, 5G & Mobility Research for IDC. “WiFi, LPWAN, Bluetooth, LTE, 5G, and so on. What’s the right connectivity option and when?”

Many devices do just fine with 4G and even 3G, but they aren’t robust enough to shoulder the added cost of full 5G support. With 3G already sunsetted in the U.S., vendors are hoping that RedCap will change that equation, pulling laggards along. “RedCap could help consolidate connectivity, serving as an on-ramp to an eventual all 5G world,” Leigh said.

Early movers gain the edge

In Japan, the 5G-edge data center connection is already underway. The country’s first ISP, IIJ (Internet Initiative Japan), has deployed edge data centers to help it deliver end-to-end 5G services. IIJ was having difficulty providing end-to-end 5G services because relying on centralized data centers created coverage gaps. End-to-end 5G necessitates the deployment of infrastructure near 5G antennas, but many antenna sites are impractical for full-blown data centers because they are in a remote location or located on a constrained site, like on a rooftop.

This means that traffic must travel back to central data centers or infrastructure must be deployed at the edge in smaller-footprint data center packages. IIJ started building a hybrid 5G-MEC (Mobile Edge Computing) network based on hardware from Zella DC, an Australian edge-data-center provider. 5G-MEC now provides the connectivity and infrastructure that IIJ says will enable applications that require low-latency, reliable communications, including use cases in telemedicine, smart factories, and even AgTech.

In Australia, Afterpay, a global fintech company, was expanding so rapidly that they needed to scale their data center infrastructure to keep pace. The company provides a buy-now, pay-later payment platform that relies on a proprietary decision-making engine to determine credit worthiness of retail customers in near real time. As the business scaled up, Afterpay worked with Digital Realty to deploy edge computing services to support approximately 450 employees located around the globe with the low-latency, high-availability connectivity needed to process more than 2 billion transactions annually.

In China, the operators of the Pingtan Strait Bridge are deploying an edge data center from Vertiv to manage intelligent infrastructure. The bridge is a 10-mile, two-layer span designed as a six-lane expressway on the top layer with a two-track Grade I railway on the lower layer. Because this is a cross-sea, dual-use bridge, operators must contend with issues such as strong winds and waves, meaning constant monitoring is a must. The Pingtan Strait Bridge project deployed Vertiv’s Smart Aisle edge data center to help it monitor conditions, manage bridge operations, and automate things like an intelligent gate system.

The edge data center market is still in its early stages, but with tailwinds that include high-growth technologies like AI, IoT, and 5G, this market should continue on a fast-growth trajectory for the foreseeable future.