IBM announces the latest version of the latest main frame hardware that includes new updates to accelerate AI adoption.
On Monday, hardware and consulting firms announced the IBM Z17, the latest version of the main frame computer hardware. This completely encrypted main frame is driven by IBM Telum II processors and is designed for more than 250 AI use cases, including AI agents and production AI, the company says.
The main frame may look like an old hat, but 71%of the 500 Fortune 500 companies are used today. In 2024, the main frame market was worth about $ 5.3 billion depending on the future of the consulting firm market research.
The Z17 can handle 45 billion inferences per day, and was released in 2022 and increased 50% compared to the predecessor of the company’s original Tellum processor. The system is designed to integrate with other hardware, software and open source tools.
Tina Tarquinio, the vice president of product management and design of IBM Z, said that the TechCrunch’s againe frame upgrade worked for five years, and it worked for five years before the AI craze when Openai’s CHATGPT began in November 2022.
Tarquinio spent more than 2,000 hours of research time, receiving feedback from more than 100 customers as IBM destroyed the Z17. Five years later, she believes it is interesting to see feedback, which has become a feedback that is matched with the mayor’s departure.
Tarquinio told The Techcrunch, “I know that after introducing the AI accelerator, especially in the late 2022, we are seeing all the changes in the industry.” It was really interesting. The biggest point is that you don’t know what’s coming. Therefore, the possibility is actually unlimited that AI can help. “
Tarquinio said Z17 was set to adapt and accept where the AI market is, Tarquinio. The main frame supports 48 IBM Spyre AI Accelerator chips when it is released, and plans to raise the number to 96 within 12 months.
Tarquinio said, “We are intentionally building a headroom.” We intentionally build AI agility. Therefore, as the new model is introduced, we will check if we have built a headroom for larger and larger models. It is a model that may require more local memory to talk to each other. We will come to a new model because we know how to change. “
Tarquinio said that one of the highlights of this latest hardware was to choose her favorite child, but Z17 is more energy -efficient than predecessors and competitors.
Tarquinio said, “Chip, we are increasing the AI acceleration by seven and a half, but it is 5.5 times less than multiple models for other types of accelerators or platforms.
The Z17 main frame is usually available on June 8th.