Advancing the capabilities of intelligent robotic systems fundamentally requires intelligent design of their embodiment, to provide the necessary topology, morphology, morphology, actuation and more. This is an interdisciplinary challenge and a bottle neck to robotic implementation which is leading to the robotics community exploring computational methods and AI for robot design. Foundation models, and in particular LLMs, can play many beneficial roles in the robot design pipeline from ideation, design, optimization and deployment. In this talk I will present different methodologies and means by which foundational models can be used, and also the associated ethical and technical issues.