At Mobile World Congress (MWC) Barcelona 2026, SoftBank Corp. unveiled its transition from a traditional carrier that moves data – raw, uninterpreted data packets – to an AI-native infrastructure provider enabling distributed AI workloads across edge and cloud environments.
SoftBank Corp. makes the point that, traditionally, telecommunications networks have been built to carry data, not comprehend it; static infrastructure designed to move information from origin to destination without understanding what it contained.
In the AI era, SoftBank’s Telco AI Cloud vision evolves the network into a central nervous system: an active computation platform that operates AI models directly within the infrastructure. Through AI-RAN-based MEC (Multi-access Edge Computing), SoftBank can now orchestrate and broker AI workloads across this distributed edge, offloading GPU compute to deliver real-time, reliable inference where it is needed.
By embedding this intelligence from the core to the edge, SoftBank states that it is creating a distributed platform that delivers meaning, not just data, enabling immediate decision-making for robotics, autonomous systems and smart cities.
For Physical AI, this means resource-constrained robots can now perform complex, scalable behaviors that would otherwise be difficult to achieve independently; powered not by what they carry but by the network intelligence that surrounds them.
Enabling the era of physical AI
That vision is already taking shape. A key highlight of the announcement was SoftBank’s focus on ‘Physical AI’: the convergence of AI with the physical world of robotics. Unlike traditional centralised clouds, Telco AI Cloud brings intelligence to the edge, enabling robots to make split-second decisions based on sensor data, performing complex behaviors that their onboard hardware alone could not independently support.
Following a collaboration with Yaskawa Electric Corporation focused on deploying robots in real-world environments, SoftBank demonstrated a joint proof-of-concept with Ericsson. The demonstration showcased how AI-RAN networks can optimise connectivity for robots, ensuring the stability required to work safely alongside humans in dynamic environments.
