At the Hot Chips conference in 2024, IBM made a stunning announcement of the core components of its next-generation mainframe technology - the Telum II processor and the Spyre AI accelerator. This innovative combination aims to greatly enhance the artificial intelligence (AI) processing capabilities of IBM Z series systems to meet the urgent need for efficient, secure, and scalable AI solutions in the enterprise sector. The Telum II processor, as the powerful heart of future IBM Z and LinuxONE platforms, boasts impressive specifications. It features eight high-performance cores with a clock frequency of up to 5.5GHz and a significant 40% increase in on-chip cache capacity. Notably, this processor incorporates an AI accelerator that delivers four times the computing power compared to its predecessor, specifically designed for low-latency, high-throughput AI inference tasks. Of particular interest is the addition of the Data Processing Unit (DPU) in Telum II. This on-chip feature, designed for I/O acceleration, significantly improves data processing efficiency and increases I/O density by 50%. The introduction of DPU not only simplifies system operations but also brings a significant performance leap for data-intensive applications and AI workloads. Working alongside Telum II is the Spyre accelerator, a solution carefully crafted by IBM to meet the growing demand for large language models (LLM) in the mainframe environment. Each Spyre chip embeds 32 computing cores and supports various data types such as int4, int8, fp8, and fp16, specifically tailored for AI applications. The Spyre accelerator and Telum II work closely together to implement IBM's "AI integration approach," which seamlessly combines traditional machine learning models with LLM to achieve more precise and robust AI processing effects. For IBM's mainframe customers, this technological innovation means they will enjoy an unprecedented AI application experience on the Z platform. Telum II effortlessly handles large-scale AI workloads and data-intensive business demands, while the Spyre accelerator focuses on processing complex AI models and generative AI scenarios. All of this is achieved without compromising the security and reliability of the IBM Z environment. It is worth mentioning that both the Telum II processor and the Spyre accelerator will be manufactured using Samsung's advanced 5-nanometer process technology. IBM expects these two products to be officially available to customers in 2025, bringing even more powerful AI capabilities to enterprises worldwide.