Artificial intelligence is rapidly becoming a critical component of modern digital systems. While much attention has focused on model accuracy and large-scale training in centralized data centers, an equally important factor is often overlooked: latency. In many real-world applications, the value of AI depends not only on how accurate the model is, but also on how quickly it can produce a decision. This talk explores how low-latency AI inference is emerging as a new workload class for Internet infrastructure. We begin with a simple observation: in the age of AI, outcomes increasingly depend on both accuracy and response time. This principle can be seen across domains. In the public sector, modern unmanned systems combine remote human control with elements of autonomy and AI-assisted targeting, where decision speed can be critical. In the private sector, applications such as visual product search in mobile eCommerce rely on machine vision models that must return results quickly to maintain user engagement. These requirements raise important questions for network operators. If AI inference must operate with very low latency, where should it run? While model training will continue to reside in centralized data centers, inference workloads may increasingly move closer to users at CDN points of presence or ISP edge facilities. From an operator’s perspective, this shift introduces several new considerations. Edge environments that historically relied on CPU-based workloads may begin to incorporate GPUs as new infrastructure components. Power consumption, while manageable for inference workloads, becomes an operational factor. Network architecture—such as IPv6 deployment and latency optimization—directly influences AI performance. In addition, edge software stacks designed around CPUs must evolve to coordinate scarce GPU resources efficiently. Finally, as AI becomes embedded in operational and decision-making systems, traditional security principles remain essential. AI models are only as reliable as their training data, and infrastructure security continues to play a critical role in ensuring trustworthy outcomes. This talk discusses the emerging Age of AI Inference at the Edge and its implications for network operators, CDN platforms, and Internet infrastructure.