Over the past decade, edge computing has been a recurring and prominent topic at various conferences.

Initially, the industry grappled with defining 'what edge computing is' and 'why it is necessary,' engaging in debates over terms such as far edge, near edge, and network edge. As time progressed, edge computing became closely associated with 5G, the industrial internet, and connected vehicles, with the focus shifting to core keywords like 'real-time performance' and 'local decision-making.' Operators and equipment vendors started promoting MEC (Multi-Access Edge Computing) architectures. Subsequently, as concepts became clearer, AI inference began to shift towards the edge, enabling cameras, robots, and industrial devices to analyze data locally and respond in real time, rather than merely collecting it. Today, with the advent of generative AI and agents, industry discussions have transitioned from 'computing power migration' to 'distributed intelligent collaboration.' Many theoretical edge computing concepts are now rapidly being implemented in real industrial scenarios.

Recently, at an international symposium, experts from JLL, Intel, Ericsson, Qualcomm, and American Tower Corporation discussed the latest trends in edge computing. Participants represented various segments of the industrial chain, including real estate, chips, communication equipment, and tower infrastructure, offering numerous insightful perspectives. This article aims to collate, organize, streamline, and share these insights.

The 'Delayed' Killer Application: AI Inference

Rather than attempting to establish a single standard definition for 'edge,' participants gradually reached a consensus around a 'continuum'—a flexible and programmable execution environment that spans central clouds, regional edges, local edges, and even enterprise edges. In this system, workloads are dynamically deployed to different locations based on latency, privacy, security, and cost requirements.

One of the most noteworthy viewpoints from the discussion is that AI inference is emerging as the true 'killer application' for edge computing.

Jim Poole of American Tower Corporation used a vivid metaphor to summarize a decade of MEC development: 'MEC was like looking for a hammer while holding a nail.' Multi-Access Edge Computing (MEC) is a network architecture that provides cloud computing capabilities and IT service environments at the network edge, aiming to reduce latency, ensure efficient network operations and service delivery, and enhance customer experience.

From the definition of MEC, it is evident that the industry long anticipated the need for a distributed computing layer, prompting operators and infrastructure vendors to deploy numerous edge nodes in advance. However, the challenge lay in the lack of business scenarios that genuinely required these nodes. In other words, infrastructure development outpaced demand. This observation also explains why edge computing remained more theoretical than practical for years. Whether in the industrial internet, VR/AR, or connected vehicles, none of these scenarios generated sufficiently large-scale or sustained demand for computing power.

The current difference lies in a fundamental shift on the demand side. With the proliferation of generative AI, regardless of the answers users seek, they must provide requests and data sources upstream. Raw data such as high-definition images, audio, and video streams generated locally must be uploaded to the cloud in real time for processing, leading to a significant increase in upstream data volume over the past year. Dr. Koymen of Qualcomm noted that user behavior is shifting from downstream-centric video consumption to upstream-centric AI-generated traffic, with agentic data expected to surpass human-generated data in the coming years. Joe Constantine of Ericsson cited data from the Ericsson Mobility Report to further support this point: global data traffic is projected to triple by 2029, while upstream traffic will grow tenfold by 2035.

This new 'upload-infer-respond' model imposes unprecedented requirements on network latency and bandwidth—precisely where edge computing excels. Sean Farney of JLL asserted, 'Edge AI inference is making the infrastructure sector exciting again.' After two decades of pursuit, the industry has finally encountered its true killer application: AI inference. It possesses two critical characteristics: sufficiently high computational density and sensitivity to latency, both of which 'force' computing power to diffuse outward from centralized cloud data centers.

AI Is Driving a 'Rewrite' of the Data Center Ecosystem

So, what will a true edge node for the AI era look like? Poole presented striking data: over the past 25 years, approximately 95% of global data centers have been designed around a power density of 5–10 kW per rack. Today, however, next-generation AI systems require power densities of 150–200 kW per rack, with Google even demonstrating a 1 MW per rack configuration.

This is no longer a problem that can be solved through minor airflow optimizations. Two direct consequences have emerged. First, 'enterprise self-built data centers' are rapidly becoming infeasible. Poole noted that for the past two decades, the data center industry's greatest competitor has been in-house enterprise data centers. However, with the soaring complexity and power density of AI infrastructure, the self-built model is increasingly untenable. 'You can no longer solve the problem by simply building your own data center,' he said.

Second, liquid cooling is transitioning from an 'advanced solution' to an industry standard, while on-site power generation is evolving from an 'option' to a regulatory requirement in some U.S. states.

Simultaneously, the geographical distribution of computing infrastructure is undergoing significant changes. While most computing resources in North America remain concentrated in around 15 core metropolitan areas today, Poole predicted that the industry will rapidly expand into 30–50 second- and third-tier markets within a development cycle far shorter than the past 25 years.

The truly unresolved question is whether future edge infrastructure will trend toward 'centralization' or 'decentralization': Will the future feature 300 national 10 MW-class data facilities, or 2,000 60 kW edge cabinets deployed beside every communication tower? Tower companies clearly favor the latter.

As a representative of chip vendors, Qualcomm offered an additional perspective: not all AI inference tasks must rely on GPUs. Dr. Koymen argued that while GPUs excel at model training, their cost and power consumption are prohibitive for inference scenarios. In contrast, NPUs optimized for inference and deployed on end devices and far edges are better suited for lightweight inference tasks within the edge continuum.

In a sense, AI is restoring data centers to their 'heavy industry' roots. This also means that competition in edge computing is shifting from software capabilities to energy, real estate, and infrastructure capabilities—i.e., who can more quickly secure land, power, cooling, and deployment resources. This explains the presence of real estate and tower infrastructure firms in the discussion.

2028–2029: Where Will the Industry Be Headed?

Facing impending transformations, experts offered specific predictions for the industry landscape in 2028–2029.

Qualcomm's Koymen linked his predictions to the 6G roadmap: pre-commercial devices will emerge in 2028, with global commercialization synchronized with operators by 2029. By then, edge infrastructure will support applications such as AI Recall, 'what you see is what you get' AR glasses, and robotic distributed computing.

Ericsson's Constantine provided three more quantitative judgments: First, by 2029, 75% of global data traffic will run on 5G networks. Second, the industry will shift from debating 'what the edge is' to arguing over service-level agreements (SLAs) and TM Forum Level 4/5 automation—a sign of maturation. Third, sustainability pressures in data centers will become the primary design constraint.

Intel's Agarwal focused on industrial implementation, predicting that by 2028, industry conferences will feature retailers, mining companies, and port operators discussing actual ROI from deployments rather than equipment vendors discussing architectures. He warned the industry against repeating the mistakes of private wireless networks, which were built but failed to produce truly successful cases.

Farney offered the most forward-looking prediction: humanoid robots will begin appearing in data center operations to help address severe labor shortages. This concept of 'physical AI' is becoming a reality—NVIDIA and T-Mobile have announced a partnership to deploy AI-RAN infrastructure at the 5G network edge, enabling AI agents to perceive and respond in real time at urban intersections and industrial facilities while significantly reducing terminal device requirements through edge computing power.

Epilogue

However, beyond the technology itself, the discussion ultimately highlighted several more practical issues: power, talent, and data.

First is power: AI inference is driving explosive growth in computing demand, but global grid construction lags far behind. Poole even stated bluntly that the U.S. grid was not designed for such localized high-density loads. Second is talent: JLL alone currently has thousands of unfilled data center-related positions. Finally, there is data: Constantine argued that future winners may not be companies with the best models but those with the highest-quality data systems. As model capabilities converge, data quality, structure, and governance will likely become the true competitive advantages in future AI competition.

Taken together, a clear picture emerges: edge computing has moved beyond the 'why we need it' theoretical phase into the 'how to build it well' engineering phase.

References: Why the edge finally has its killer use case – RCRWireless; Times Have Changed: 'Massive Uplink' Becomes the Focus of Communication Network Upgrades – moomoo; What Is Multi-Access Edge Computing (MEC)? – Redhat