Artificial Intelligence is rapidly moving closer to where services are delivered. As gateways and video devices become increasingly powerful, operators have a new opportunity to leverage intelligence directly on customer premises equipment (CPE). But deploying AI models is only the beginning. To unlock long-term value, operators need a way to manage AI services at scale while maintaining performance, security, and service continuity. John Blackford shares his perspective on what it will take to operationalize Edge AI across broadband and video environments.
Edge AI has been discussed for years. Why is it becoming a real opportunity for operators now?
Several technology trends are converging at the same time. Modern gateways and video devices increasingly include dedicated AI acceleration capabilities, such as NPUs and advanced GPUs, that were simply not available in most deployed devices a few years ago. At the same time, AI models have become more efficient, making it possible to run meaningful inference workloads directly on consumer devices instead of relying entirely on cloud infrastructure.
For operators, this opens up new possibilities. Some AI tasks can now be performed locally, closer to the user and the service itself. That can reduce latency, improve responsiveness, and help keep sensitive data on the device rather than sending it elsewhere for processing. It can also reduce cloud processing costs by ensuring only the most resource-intensive functions are handled centrally.
What’s important to understand is that Edge AI works alongside the cloud rather than replacing it. The future is a hybrid model where the cloud and the edge work together. The cloud remains essential for orchestration, analytics, and large-scale AI operations, while the edge becomes the ideal place to execute real-time tasks and deliver immediate value to subscribers. The industry has moved past asking whether AI can run on CPE. The real question now is how operators can deploy and manage these capabilities at scale.
Many discussions focus on AI models and use cases. Why is managing AI at scale the bigger challenge?
Deploying a model is only a small part of the story. Running AI consistently across millions of devices introduces an entirely different set of challenges.
Unlike cloud environments, CPE devices operate within strict resource constraints. AI workloads must coexist with mission-critical functions such as broadband connectivity, Wi-Fi management, and video delivery. Operators need assurance that introducing intelligence to the device will never compromise the services customers depend on every day.
There is also the challenge of managing the lifecycle of AI. Models evolve rapidly. They need to be deployed, updated, monitored, optimized, and sometimes replaced. Operators need visibility into performance, resource consumption, and operational health across their entire base of installed devices.
In many respects, the challenge sits with operationalizing AI rather than with AI itself. Without orchestration, governance and lifecycle management, AI remains a collection of isolated proofs of concept. Operators need a framework that allows them to treat AI services the same way they manage any other critical service in their network environment.
What does “operationalizing Edge AI” actually mean in practice?
Operationalizing Edge AI means turning it from a technology experiment into a managed service that can be deployed, monitored and evolved over time.
From an operator’s perspective, that means being able to remotely provision AI services, manage versions, collect telemetry, monitor performance, and enforce operational policies across the entire fleet. It also means understanding exactly what resources an AI application is consuming and ensuring that those resources are allocated appropriately.
Resource orchestration becomes particularly important. Many edge devices have limited compute and memory resources, so operators need mechanisms to determine which AI services run, when they run, and how they interact with other applications on the device. Maintaining service continuity must always remain the top priority.
Just as important is observability. Operators need insight into the performance of AI services in the field, not only to troubleshoot issues but also to continuously improve service quality. Operationalizing AI is ultimately about creating a predictable, reliable, and scalable operating environment that can support future innovation.
How important are standards and interoperability in making Edge AI scalable?
They are absolutely fundamental.
Most operators manage a highly diverse installed base that spans multiple generations of gateways, video devices, operating systems, and access technologies. If every AI service requires a separate management framework or a different integration model, complexity grows very quickly and limits the operator’s ability to scale.
A standards-based approach helps create consistency across that environment. It allows operators to leverage familiar management practices while introducing new AI-powered capabilities. It also enables greater flexibility because operators are not tied to a single AI model, application provider, or technology ecosystem.
That’s an important point. The AI market is evolving extremely quickly. New models and new services are appearing all the time. Operators don’t want to redesign their architecture every time an innovation emerges. They need a foundation that gives them control and allows them to adopt new technologies while maintaining a consistent operational framework.
Ultimately, interoperability moves past being solely a technical requirement and turns into a business requirement that gives operators the freedom to innovate without sacrificing control.
Where do you see the most promising opportunities for Edge AI in broadband and video services?
There are opportunities in both domains, and many are already becoming tangible.
In broadband environments, Edge AI can help optimize Wi-Fi performance, analyze local traffic conditions, and identify network anomalies before they affect subscribers. That enables a more proactive operational model and can help reduce support costs while improving customer satisfaction.
In video environments, AI can support capabilities such as real-time transcription, content translation, and other advanced media experiences. By processing these functions locally, operators can deliver highly responsive experiences while reducing dependence on centralized resources.
What excites me here is the platform that enables these capabilities. Once operators establish the ability to manage AI services effectively across their device fleets, they gain the flexibility to continuously introduce new services over time. The same operational foundation that supports a Wi-Fi optimization application today could support entirely different AI-driven experiences tomorrow.
Aside from operational efficiency the value of Edge AI is it’s ability to deliver more intelligent, responsive, and personalized services to subscribers while maintaining the control and scalability operators require.
What should operators be doing today to prepare for the next phase of Edge AI?
Operators should start by thinking beyond individual AI applications and focus on long-term operational strategy.
The industry is moving toward a future where AI becomes a standard capability of connected devices, much as cloud connectivity and software platforms have become standard today. Success will not be determined by who deploys the most AI models. It will be determined by who can deploy, manage and evolve those models efficiently across their entire device population.
That means investing early in the foundational capabilities required for large-scale AI operations: orchestration, lifecycle management, telemetry, security and governance. Those building blocks may not always be the most visible part of an AI strategy, but they are what ultimately make innovation sustainable.
The future of Edge AI bringsis not simply about bringing intelligence closer to the user and. It’s about givesing operators a practical way to manage that intelligence across millions of devices, while continuing to deliver the reliability, performance, and customer experience their subscribers expect. When that foundation is in place, AI becomes far more than a technology trend. It becomes an operational capability that can drive innovation for years to come.