This document describes the use case requirements and technical evaluation criteria for remote GPU services for generative AI and LLM. After discussing GPU computing in green data centers capable of handling large-scale AI computations in recent years and the potential need to connect user data, we discuss the use case of connecting large volumes of AI training data to GPUs using IOWN technology to reduce training time, power consumption, and cost. Next, the technical evaluation criteria are specified to verify the technology readiness of IOWN technologies required for these use cases.