IBM has announced a new line of data center chips and servers, introduced on Tuesday, designed to enhance power efficiency and streamline artificial intelligence integration within business operations. This marks the initial significant update to IBM’s “Power” chip line since 2020.
These new Power11 chips are specifically engineered for data centers, historically competing with offerings from Intel and Advanced Micro Devices. IBM’s Power systems, similar to Nvidia’s AI servers, constitute an integrated package of both hardware chips and proprietary software. Tom McPherson, general manager of Power systems at IBM, stated that this integrated approach facilitated a focus on system reliability and security. The Power11 systems will be commercially available starting July 25.
The Power11 systems are designed for continuous operation, eliminating the need for planned downtime associated with software updates. Unplanned downtime for these systems averages slightly over 30 seconds annually. IBM also detailed that these systems are equipped to detect and respond to ransomware attacks within one minute, a mechanism to counter data encryption attempts by malicious actors seeking ransom. Further integration is planned for the fourth quarter of this year, as IBM intends to combine the Power11 with Spyre, its AI chip introduced in 2023.
McPherson clarified IBM’s strategic positioning regarding artificial intelligence, indicating that the company does not aim to directly compete with Nvidia in the domain of AI system creation and training. Instead, IBM’s focus is on simplifying AI deployment for inference. Inference, in this context, refers to the practical application of an AI system to accelerate various business tasks. “We can integrate AI capabilities seamlessly into this for inference acceleration and help their business process improvements,” McPherson affirmed in a recent interview. He added, regarding the Power11’s capabilities, “It’s not going to have all the horsepower for training or anything, but it’s going to have really good inferencing capabilities that are simple to integrate.”
