AMD CEO Lisa Su announced that in 2026, for the first time, the world is utilizing more AI compute resources for running models than for training them. This shift reflects a broader trend within the AI sector, where there is a growing focus on inference workloads as AI models achieve widespread production use. Semiconductor providers are adapting their offerings to meet the differing demands of training versus operating AI systems at scale.

AMD: AMD is a semiconductor company that designs processors and accelerators for high-performance and AI computing applications. Its leadership recently highlighted a pivotal change in how AI resources are allocated globally. This positions AMD as a supplier of hardware supporting both model development and operational deployment.
Lisa Su: Lisa Su serves as President and CEO of AMD, guiding its focus on advanced computing technologies including those for artificial intelligence. She recently noted the point at which running AI models began to consume more compute than training them. Her perspective underscores AMD’s stake in the evolving AI hardware landscape.

Hardware Demand: Semiconductor providers are adapting offerings to address the distinct requirements of training versus running AI systems at scale.
AI Workload Transition: The AI sector is experiencing a notable pivot toward greater emphasis on inference workloads as models reach widespread production use.