IBM has released the latest version of its mainframe hardware, including new updates designed to accelerate AI adoption.
Hardware and consulting firm announced the IBM Z17 on Monday, the latest version of its mainframe computer hardware. The company says the fully encrypted mainframe is powered by IBM Telum II processors and is designed for more than 250 AI use cases, including AI agents and Generative AI.
Mainframes may seem like old hats, but today 71% of Fortune 500 companies use them According to a source. According to the future of consulting firm market research, the mainframe market is estimated to be worth $5.3 billion in 2024.
The Z17 can process 450 billion inference operations a day, a 50% increase over its predecessor. IBM Z16The company was released in 2022 and runs on the company’s original Tellum processor. The system is designed to be fully integrated with other hardware, software and open source tools.
Tina Tarquinio, vice president of product management and design at IBM Z, told TechCrunch that this mainframe upgrade has been in operation for five years – Openai’s Chatgpt November 2022.
Tarquinio said IBM spent more than 2,000 research time to get feedback from the Z17 from more than 100 customers. She thought it was interesting that five years later, they were in line with the feedback that the market would eventually go.
“It’s crazy to know that we’re introducing AI accelerators and then seeing, especially in late 2022, all the changes in the industry involve AI,” Tarquinio told TechCrunch. “It’s really exciting. I think the biggest point is [that] We don’t know what we don’t know, right? So, in terms of what AI can help us do, this possibility is truly limitless. ”
Tarquinio said the Z17 was set up to adapt and adapt to the position of the AI market leader. Mainframes will support 48 IBM Spyre AI accelerator chips upon release, and plan to increase that number to 96 in 12 months.
“We built it intentionally in clean air,” Tarquinho said. “We built it intentionally in AI agility. So when new models are introduced, [we’re] Make sure we have built larger, larger models in the clearance – these models may require more local memory to communicate with each other. We build on this because we know that this is actually a way to change, right? New models come and go. ”
One of the highlights of this latest hardware — though she joked that it was like being asked to pick a kid she liked — is that the Z17 is also more energy-efficient than its predecessor and so-called competitors.
“On-chip, we’ve increased AI acceleration by seven and a half times, but that’s five and a half times less energy than you’ll need to do on another type of accelerator or platform in the industry,” Tarquinio said.
The Z17 mainframe will usually be available on June 8.
