IBM Unveils New Chips and Servers to Streamline Enterprise AI Deployment

IBM has launched a new generation of data center chips and servers, intended to consume less energy than rival products, and simpler, for companies to adopt AI in their businesses.

On Tuesday, it launched the Power11 chip, its first major upgrade to its Power range since 2020. Power chips are mostly used in enterprise settings like banking, healthcare, and manufacturing where performance, reliability, and security are very important, and multiple cores are used to enable a chip to operate in a variety of ways.

Like Nvidia’s tightly integrated hardware and software AI systems, IBM’s Power servers pair hardware and software.According to Tom McPherson, general manager of IBM Power Systems, this close integration allows IBM to deliver enhanced reliability and security.

The Power11 systems, which will be available starting July 25, are engineered for minimal downtime. They require no planned interruptions for software updates, and average just over 30 seconds of unplanned downtime annually. Additionally, the systems can detect and respond to ransomware attacks within a minute, offering a new level of defense against data breaches.

Later this year, IBM plans to pair Power11 with Spyre, its purpose-built AI chip introduced in 2024. However, IBM is not attempting to go head-to-head with Nvidia in training large AI models. Instead, the company is focusing on AI inference—the deployment of AI to automate and enhance business processes.

“We’re focused on seamlessly integrating AI for inference acceleration, making it easier for enterprises to improve operations,” McPherson said in an interview."It's not about having the most power for training", but "simply having efficient inferencing capability that businesses can implement in as little time as possible" said IBM.

IBM said it has received positive initial customer feedback on their Power11 systems, especially in industries that are trying to increase speed in decision making, automate workloads and increase operational resiliency.