IBM Launches New AI-Powered Mainframe

IBM Unveils z17 Mainframe with Enhanced AI Capabilities

IBM has announced the release of its newest mainframe system, designed to facilitate the wider implementation of artificial intelligence technologies.

Introducing the IBM z17

On Monday, the technology and consulting giant revealed the IBM z17, representing the latest iteration of its mainframe computer hardware. This system boasts full encryption and is powered by the IBM Telum II processor. It’s engineered to handle over 250 distinct AI applications, encompassing AI agents and generative AI solutions.

Mainframes Remain Relevant

Despite their long history, mainframes continue to be crucial for many large organizations. Data indicates that 71% of Fortune 500 companies currently utilize them. The global mainframe market was valued at approximately $5.3 billion in 2024, according to Market Research Future.

Performance Improvements

The z17 is capable of executing 450 billion inference operations daily, marking a 50% performance gain compared to the IBM z16, released in 2022 and based on the initial Tellum processor. The architecture is designed for seamless integration with existing hardware, software, and open source tools.

Years in Development

Tina Tarquinio, VP of product management and design for IBM Z, shared with TechCrunch that the development of this mainframe upgrade spanned five years. This process began prior to the surge in AI interest sparked by the launch of OpenAI’s ChatGPT in November 2022.

Customer-Driven Design

IBM invested over 2,000 research hours and gathered input from more than 100 customers during the z17’s creation. Tarquinio noted the alignment between the feedback received and the current direction of the market.

Adapting to the Evolving AI Landscape

“We’re introducing an AI accelerator at a time of significant industry changes regarding AI,” Tarquinio explained to TechCrunch. “The possibilities are truly limitless in terms of how AI can assist us.”

Scalability and Future-Proofing

The z17 is designed to adapt to future developments in the AI market. Initially, it will support 48 IBM Spyre AI accelerator chips, with plans to increase this to 96 within a year.

“We are intentionally incorporating capacity for growth,” Tarquinio stated. “We’re intentionally building in AI agility, ensuring we can accommodate larger, more complex models – those potentially requiring increased local memory for communication.”

Energy Efficiency

Tarquinio highlighted the improved energy efficiency of the z17 compared to its predecessor and competitors.

“We’re achieving a seven and a half times increase in AI acceleration on-chip, while consuming five and a half times less energy than comparable multi-model operations on other platforms,” she said.

Availability

The z17 mainframes will be generally available starting June 18.

This article has been updated to accurately reflect IBM’s stated timeline.