Pat Gelsinger, CEO of Intel, recently visited Elon Musk's Memphis Supercluster, offering high praise for the impressive AI infrastructure built by Musk's xAI team. In a post on X (formerly Twitter), Gelsinger highlighted the use of Intel Xeon processors in xAI’s AI head node, a dedicated server tasked with managing a massive 100,000-GPU AI cluster. Gelsinger marveled at how this system was constructed in an exceptionally short amount of time, emphasizing the remarkable engineering feat behind it.

Intel, once a dominant force in the data center and semiconductor industries, has faced significant challenges in recent years. After a substantial $1.6 billion loss in its data center and foundry businesses, the company has been struggling to regain momentum. Furthermore, Intel has notably missed out on the AI revolution, with its Gaudi 3 AI accelerator still grappling with performance issues. In contrast, companies like Nvidia have surged ahead, with Musk’s xAI project spending an estimated $10 billion this year alone on AI training hardware, including a staggering 100,000 Nvidia H200 GPUs.

Despite Nvidia's dominance in AI accelerator chips, the role of CPUs remains critical for managing and directing these vast processing power networks. Gelsinger's post confirms that xAI has selected Intel Xeon processors to power their head node, a key part of their data center’s AI operations. Although Intel recently launched the powerful Granite Rapids 128-core CPU, Gelsinger did not specify which Xeon model is being used by xAI.

The collaboration between xAI and Intel also highlights the ongoing relationship between Musk and other major tech players. Gelsinger took the opportunity to praise Michael Dell, the founder and CEO of Dell Technologies, for providing the head node servers used by xAI. This raises questions about whether Dell servers have replaced those originally purchased from Supermicro. While rumors circulate about Supermicro's legal issues potentially influencing xAI’s hardware choices, it’s unclear whether Dell has fully replaced Supermicro or if it is simply a case of xAI diversifying its suppliers.

Looking ahead, xAI’s massive AI infrastructure is poised for further expansion. Musk has announced plans to increase the number of GPUs on-site to 200,000, with potential growth to 300,000 GPUs in the future. While the bulk of these additional GPUs will likely continue to rely on Nvidia’s AI accelerators, it is also expected that xAI will require additional CPUs to support this scale. Gelsinger’s visit could be an effort to maintain strong ties with Musk in the hopes of securing more Intel CPU business.

Despite xAI not yet adopting Intel’s AI chips, the growing demand for CPUs in AI clusters presents a promising opportunity for Intel to recover from its recent financial setbacks. The Memphis data center serves as a critical juncture in Intel’s strategy to reposition itself within the rapidly expanding AI sector. With competition from Nvidia and others intensifying, Gelsinger’s visit may signal Intel’s determination to leverage its Xeon processors in an increasingly AI-driven world.

TOMSHARDWARE