Qualcomm has announced its renewed entry into the data center CPU market, unveiling plans to develop custom processors that leverage Nvidia technology for seamless integration with Nvidia´s leading Artificial Intelligence chips. This strategic move marks a revival of Qualcomm´s earlier ambitions in the sector after having previously scaled back its data center processor operations in the 2010s due to high costs and legal disputes.
The new CPUs are specifically designed to work optimally with Nvidia GPUs, which are pivotal in driving Artificial Intelligence workloads in modern data centers. These custom chips will incorporate Nvidia connectivity technology, facilitating extremely rapid communication between the CPU and GPU components. This close integration aims to enable the high-speed data transfers required for large-scale Artificial Intelligence applications, boosting performance and efficiency compared to more generic system architectures.
Historically, the data center CPU market has been dominated by Intel and AMD. However, the exponential growth of Artificial Intelligence applications is opening doors for alternative architectures and suppliers. Alongside Qualcomm´s renewed initiative, Nvidia itself has entered the CPU space with its Arm-based ´Grace´ processor, highlighting the sector´s shift toward heterogeneous computing architectures. Qualcomm´s recent progress follows its acquisition in 2021 of a team of ex-Apple chip designers, further accelerating its data center ambitions. In addition to confirmed talks with Meta Platforms regarding a potential collaboration, Qualcomm has already established a letter of intent with Saudi Arabian Artificial Intelligence firm Humain to develop a specialized CPU. According to CEO Cristiano Amon, the goal is to deliver high-performance, energy-efficient data center computing through deep integration with Nvidia´s rack-scale architecture, further aligning both companies´ visions for the future of the data center market.
