The Power11 family introduces an AI-ready infrastructure with on-chip acceleration for inference and the capability to scale toward more demanding AI workloads through the IBM Spyre accelerator. Combined with a broad open-source ecosystem, Power11 delivers the flexibility, performance, and cost efficiency needed to operationalize AI workloads close to existing data on Power systems.This session will begin with a concise overview of key AI concepts, focusing on the distinction between training and inference, particularly in the context of large language models (LLMs). It will then examine current trends in the GPU market, highlighting the major challenges customers face—such as availability, cost, and scalability limitations.Within this context, IBM Spyre will be introduced as a new generation accelerator designed to enhance inference workloads. The session will explain how Spyre complements CPU and GPU infrastructures, providing an efficient, scalable, and cost-effective alternative to traditional accelerator architectures. Attendees will gain a clear understanding of Spyre’s positioning, its integration with software solutions, and the tangible business and technical benefits it offers for AI deployment and optimization.
