August 11, 2025
5 min read
Stephen M. Saunders MBE
The AI Opportunity Moves to the Edge: How 5G and Edge AI Agents Transform Networks
The AI revolution is entering a new phase — and that’s good news for carriers and service providers. Phase 1 of the global shift to AI focused on training large language models (LLMs) in giant data centers packed with expensive GPUs. In Phase 2, which is currently underway, the focus shifts to autonomous agentic AI agents installed at the network edge. As with so many “new” trends in communications, we’ve seen this before — back in the 1980s, with the transition from centralized mainframes to client-server computing. That shift prompted the need to increase network capacity at the periphery of the network from coaxial cables supporting a few hundred kilobits per second to twisted pair cables and protocols capable of megabit speeds, which in turn spawned a new and profitable local area network (LAN) industry.The Need for Agentic AI Agents at the Edge of the Network to Draw Down Inference
And that, right there, is the solution to the existential question that service providers have been asking themselves for the last two-and-a-bit years: how do we get a piece of the AI revolution? (And, of course, it also finally answers another carrier question: “Why the heck do we need 5G?” Because only 5G, not 4G, can deliver the AI-native, cloud-native, network-sliced, low-latency network that inference requires). So, the answer for carriers is to stick to their knitting: upgrading and improving the edge networks providing the essential architecture in anticipation of ubiquitous agentic AI. Equally importantly, the new edge phase of AI also provides carriers with clarity on what not to do. Service providers have spent much of the last couple of years arguing internally over whether their future lies in building AI factories or enabling an ecosystem to launch an AI-enabled OTT-style consumer app. Those are terrible strategies (unless you had the foresight to see this whole AI trend coming a decade ago and start a new company with a strong, wide moat — like Vultr or Cerebras). This is a shift in the AI market that is bigger than AI itself — or carriers, or telecom, for that matter. The combination of agentic AI and inference-capable edge networks provides the infrastructure for a new global digital economy in which all businesses and industries become smarter, more efficient, and more sustainable. Carriers now have an essential role in building it.Op-eds from industry experts, analysts or our editorial staff are opinion pieces that do not represent the opinions of Fierce Network.
Source: Originally published at Fierce Network on August 11, 2025.