Qualcomm to build Data Center CPUs linking Nvidia chips
Qualcomm announced on Monday its renewed entry into the data center CPU market, revealing plans to develop custom processors that integrate seamlessly with Nvidia’s dominant artificial intelligence (AI) chips. This move reflects a strategic pivot as AI applications increasingly demand high-performance, energy-efficient computing at scale.
Historically, CPUs from Intel and AMD have powered data centers, often working alongside Nvidia’s graphics processing units (GPUs) to run AI workloads. Although Nvidia has begun designing its own CPUs like the Arm-based “Grace” the broader ecosystem still relies on strong CPU-GPU cooperation. Qualcomm aims to enhance this synergy by ensuring its chips communicate efficiently with Nvidia’s rack-scale architecture.
The announcement highlights Qualcomm’s use of Nvidia’s interconnect technology, which enables faster data exchange between CPUs and GPUs. This capability is critical for AI performance, especially in complex data center environments.
Renewed Efforts After Previous Setbacks
This isn’t Qualcomm’s first attempt at data center CPUs. During the 2010s, the company experimented with Arm-based server chips and even tested them with Meta Platforms. However, those efforts were scaled back due to cost concerns and legal hurdles. Despite the setback, Qualcomm did not abandon the vision entirely.
Following its 2021 acquisition of a team of former Apple chip engineers, Qualcomm began to quietly rebuild its server chip ambitions. As a result, conversations with Meta resumed, suggesting ongoing interest in Qualcomm’s capabilities. Additionally, the company recently confirmed a letter of understanding with Humain, a Saudi Arabian AI firm, to co-develop a custom CPU tailored for AI workloads.
A Strategic Vision for AI Infrastructure
Cristiano Amon, Qualcomm’s CEO, emphasized the broader goal: delivering high-performance computing with optimal energy efficiency. “With the ability to connect our custom processors to Nvidia’s rack-scale architecture, we’re advancing a shared vision of high-performance energy-efficient computing to the data center,” he said on Monday.
This partnership could reshape the competitive landscape. While Intel and AMD continue to lead in traditional CPUs, Qualcomm is banking on a different formula: specialized design and strategic collaboration. By leveraging Nvidia’s strengths and its own renewed capabilities, Qualcomm hopes to carve out a space in the increasingly AI-driven future of data center technology.
