IBM has unveiled the Z17, its latest mainframe system designed specifically to manage the workloads of artificial intelligence on a business scale, including important languages and generative AI applications. The new iteration is represented as a significant evolution of the previous models. It incorporates AI’s capacities into hardware, software and operational management. IBM positions its mainframe Z17 design as a complete solution for companies requiring high performance, secure and scalable AI infrastructure.
Key progress in the Z17 is its Telum II processor, which includes an AI accelerator on a chip capable of processing 450 billion inference operations per day with a response time of a single millisecond. It is. Improvement of 50% compared to Z16.
In addition, IBM plans to introduce the Spyre accelerator at the end of 2025, a PCIE card that will improve the generating performance of the AI. This upgrade will allow companies to execute AI assistants and agents directly on mainframe without compromising data security.
Image credit: IBM