Foundation Models

Foundation models offer a way to build shared intelligence across multiple wireless tasks instead of solving each one in isolation. They can support transfer across settings, absorb multimodal inputs such as traffic measurements and topology information, and provide a general reasoning layer for network management.

In wireless systems, this raises important questions about what large models should represent, how they should be adapted to domain-specific constraints, and how they can operate efficiently in real-time environments. The opportunity is to create reusable models that accelerate innovation across planning, optimisation, automation, and network operations.