Intel CEO Visits Elon Musk’s Data Center, Praises Rapid Xeon Deployment & xAI Team

Intel’s CEO, Pat Gelsinger, recently expressed his admiration for the xAI team following his visit to Elon Musk’s Memphis Supercluster. Sharing his thoughts on X (previously known as Twitter), Gelsinger highlighted that xAI has incorporated Intel Xeon processors to manage its AI head node — the primary server overseeing a massive cluster of 100,000 GPUs. He remarked, “It’s incredible what’s been built in such a short amount of time!”

Intel has been facing financial hurdles, with losses surpassing $1.6 billion in its data center and foundry sectors. Additionally, the company has lagged behind in the AI industry, particularly as its Gaudi 3 AI accelerator continues to experience performance issues.

This year, Elon Musk has invested approximately $10 billion in AI training infrastructure, managing to deploy 100,000 Nvidia H200 GPUs within just 19 days — a task that Nvidia’s CEO Jensen Huang says typically takes four years. While Nvidia supplies the GPUs for the AI processing, a robust CPU is essential for directing the operations of such an extensive AI cluster. Gelsinger’s post confirms that the xAI team selected Intel Xeon processors for their AI head node. In September 2024, Intel unveiled its latest 128-core flagship CPU, Granite Rapids, although Gelsinger stopped short of confirming its use in the xAI setup.

Gelsinger also commended Michael Dell, founder and CEO of Dell Technologies, which currently supplies the servers for xAI’s head node — at least those visible in the shared images. While Musk has previously acquired systems from Supermicro, it remains unclear if the Dell servers are fully replacing other Supermicro units or if they are being used alongside them. The preference for Dell head nodes may fuel speculation that legal troubles at Supermicro have influenced xAI’s vendor choices, although this remains unconfirmed.

While Gelsinger’s visit to the Memphis data center was primarily to observe xAI’s vast AI cluster, other motives remain speculative. Musk has plans to expand the GPU count to 200,000 soon, with aspirations to reach up to 300,000 in subsequent phases. These expansions will likely continue to utilize Nvidia’s AI accelerators, possibly including Blackwell GPUs, making it improbable that Gelsinger was pitching Intel’s Gaudi 3 chips to Musk.

However, the substantial investment in GPUs implies a significant need for CPUs, which might be why Gelsinger is maintaining a favorable relationship with Musk, aiming to secure more sales for Intel’s Xeon processors. Despite xAI not currently purchasing Intel’s AI chips, securing CPU sales could significantly aid Intel’s recovery from its recent financial downturn announced this past August.