AI has become pervasive across enterprises, and the drive is now on to push this potentially powerful intelligence out to the edge and network, where it can deliver insight and operational performance in real time.

These days, no discussion of edge and IoT is complete without weighing the implications of AI. “If people think AI will stop at data centers, they have an unfortunate blind spot, because AI is likely to be deployed in some form at the edge,” said Kevin Brown, SVP at Schneider Electric. “In my opinion, it will bring with it a host of new challenges regarding data privacy concerns, an increasing demand for power, and an infrastructure that will be even more broadly distributed. Undoubtedly, we need to get ahead of these challenges.”

Though edge and IoT have been on the scene for well over a decade, we’re “just beginning to see data at the edge being put into action,” said Dan DeBacker, SVP of products at Extreme Networks. “When organizations can convert all that available data into useful information, they can dig into business insights about user activity that can help them make better data-driven decisions around investments and improving user experiences.”

Adding more AI capabilities will help convert edge data into useful business insights, DeBacker added. “AI will enable organizations to add a layer of intelligence into a process that would previously require a lot of manual hours spent digging into data, costing significant time and money. With AI, that process becomes much faster and easier.”

AI will help bring more intelligence and interactive experiences at the edge across homes, factories, healthcare, and connected devices. “AI and inferencing can help realize the promise of edge computing, like fully seamless integration with IoT devices across all industries, ultra-low latency in every application, and widespread adoption in sectors such as healthcare and manufacturing,” said Rahul Pradhan, VP of product and strategy at Couchbase. “By enabling data processing closer to the source of data generation, edge computing reduces latency, enhances data security, and supports real-time analytics, crucial for time-sensitive applications in manufacturing and autonomous vehicles.”

Edge and IoT Use Cases Percolate

Consider the capabilities AI potentially brings to manufacturing, which has been one of the strongest adopters of edge and IoT. “The most common or successful use cases come from industrial verticals such as manufacturing and energy,” said Saurabh Mishra, global director of IoT product management at SAS. “These revolve around condition-based monitoring, process optimization, product quality improvements, and predictive maintenance.” One client, a $20 billion consumer-products company, is “applying AI and IoT analytics to the huge amounts of operational data it collects. This is approximately a terabyte of data each day from sensors and edge computing devices at 140-plus manufacturing facilities.”

“The application of edge AI and machine learning algorithms, combined with multi-modal large language models, is enabling real-time-decision-making on industry floors,” said Rahul Bajpai, principal and U.S. connected edge leader with Deloitte Consulting. “Plus, with the proliferation of IoT devices, enterprises are increasingly focusing on security at the edge—a critical move to safeguard data and systems from potential breaches.”

The growing convergence between AI and edge and IoT “enables real-time data processing and analysis where the data is generated, minimizing latency,” said Mishra. “This is fundamental for scenarios like computer vision and predictive maintenance, where a large volume of data requires real-time analysis with low latency.”

Edge and IoT are key in tracking and ensuring predictive maintenance as manufacturers expand their service arms. For example, CNH, which designs, produces, and sells agricultural machinery and construction equipment, has “massive point clouds being generated by Lidar laser sensors which we filter and cluster for our tracking algorithms, advancing our sensing and perception capabilities on our tractors,” said Philip Corio, senior director of engineering for Raven Industries, a brand of CNH.

The company also employs “general telematics data and data that is specific to agronomic operations, such as machine and implement state,” said Corio. “They have different requirements that can be best addressed with a simple microcontroller. For these reasons, we typically design custom boxes that utilize multiple edge computing and IoT technologies.”

Another aggressive adopter of edge AI is the retail sector. “Electronic shelf labels are now table stakes in retail environments, most venues today use connected point of sales systems for cashless transactions, and nearly every organization uses IoT for connected security cameras and video surveillance,” said DeBacker.

Travel and transportation also benefit from edge AI. “An airline can utilize edge computing to facilitate the management of its flight and employment schedules, customer communications, and quality control of equipment,” said Pradhan. “They can receive real-time information from edge/IoT devices. The travel and transportation industries require 100% uptime and real-time updates to ensure time-sensitive data is processed where it is collected, regardless of internet connectivity.”

There has also been significant growth in hyper-local use cases as well, with companies exploring generative AI small models that can be deployed at the edge for a variety of tasks, such as “chatbots and voice assistants,” said Mishra.

Architectural Implications of Edge and IoT

The AI-edge convergence is accelerating the growing movement from centralized cloud computing to the decentralized intelligence of edge computing. “This transition is not merely a change in location for data processing, but a fundamental evolution in how we interact with and leverage AI technologies in real time,” said Pradhan.

“For edge AI to be successfully implemented, a persistent data layer is required for local and cloud-based management, distribution, and processing of data,” Pradhan continued. “Additionally, with multimodal AI models, organizations need to have a unified data platform capable of handling various data types to ensure it meets edge computing’s operational demands. Distributed inferencing, where models are trained across multiple devices holding local data without actual data exchange, can address current data privacy and compliance issues.”

Uptime and business continuity are also important parts of the equation. “As IT infrastructure continues to sprawl geographically and increases in complexity, resiliency takes on renewed focus to prevent outages,” said Brown. Security—both cyber and physical—is also cast in a different light. “Every device on a network is a potential attack surface for hackers, and every one of those devices should have up-to-date firmware and protocols. Also, how do you monitor who is on-site when there may be thousands of sites?”

Edge and IoT networks are also bringing more power to the end user. “As more people and devices move to the edge, it’s easier to manage all these endpoints via a single network,” said DeBacker. “With the massive amount of data being collected at the edge, organizations must have some kind of edge solution in place that can separate what’s most important. We will not see a decrease in people, applications, or devices at the edge anytime soon.”

The edge and IoT just don’t stop at Earth’s surface—they also extend skyward. “Satellite connectivity for IoT will continue to grow in popularity, especially low-Earth orbit satellite networks that tout more extensive coverage and reliability than their geostationary orbit constellation counterparts,” said Ian Itz, director of the IoT line of business for Iridium Communications. “Positioned closer to the Earth’s surface, LEO networks can deliver real-time data processing through reduced latency and faster data transmission, ensuring more consistent connectivity across the globe. The ubiquitous connectivity offered by LEO satellite connectivity provides the ideal platform for chipset integrations, especially as service providers look to increase IoT connected offerings.”

Search across your past Websets

Articles: real-time translation, edge computing, border patrol

Search Criteria

Enrichments

Find more results