Ericsson Americas Chief Strategy and Technology Officer Joe Constantine stated that the industry’s earlier struggles with mobile edge computing were due to timing rather than fundamental flaws. He argued that recent advancements in artificial intelligence and network technology have created conditions where edge computing can finally succeed.
Constantine described multi-access edge computing as a supply-side concept when it emerged in the 2010s, lacking the application demand needed to justify its deployment. Today, he said, AI inference provides that missing driver. He cited Ericsson’s Mobility Report, which projects that global mobile traffic will triple between 2023 and 2029, with AI serving as the primary catalyst for this expansion.
The executive highlighted a specific shift in data flow patterns. Ericsson predicts that uplink traffic will increase tenfold by 2035 as AI changes how information is shared across networks. Constantine asserted that routing this volume of data to centralized cloud servers is both slow and economically unsustainable, making local processing at the network edge essential.
Network capabilities have also evolved significantly since the initial push for edge computing. Early mobile networks were designed for best-effort connectivity, whereas modern 5G infrastructure supports time-critical communication with 15-millisecond latency and five nines of reliability. The maturity of 5G standalone networks, combined with network slicing and cloud-native cores, provides a more stable foundation for edge applications.
Constantine identified physical AI and robotics as key use cases requiring this infrastructure. He explained that autonomous systems, including drones, vehicles, and robots, need networks that can process contextual awareness in real time. Using an autonomous car with LIDAR as an example, he noted that onboard sensors cannot see around corners, necessitating network-based sensing to operate safely in society.
