Qualcomm (QCOM, Financial) has announced plans to develop custom central processing units (CPUs) for data centers using NVIDIA's (NVDA) technology to enhance connectivity with its AI chips. NVIDIA's chips, while dominant in the AI market, still require CPUs, traditionally led by Intel (INTC) and AMD (AMD). NVIDIA has also ventured into the CPU market with its ARM-based "Grace" processor.
Qualcomm is re-entering the data center CPU market. Although it previously developed an ARM-based CPU and tested it with Meta Platforms, the project was halted due to cost cuts and legal challenges. Following its acquisition of a team of former Apple chip designers in 2021, Qualcomm quietly resumed development and is in discussions with Meta again.
Last week, Qualcomm confirmed signing a letter of intent with Saudi Arabian AI company Humain to co-develop a custom data center CPU. Qualcomm's upcoming chips will leverage NVIDIA's technology to improve communication speed with NVIDIA's GPUs, central to its AI chip offerings. Qualcomm CEO Cristiano Amon stated that this collaboration aims to promote high-performance and energy-efficient data center computing.
