Qualcomm Reenters AI Compute Chip Market, Partners with Saudi Arabia’s Humain to Challenge Nvidia’s Dominance
On May 17, the Kechuangban Daily reported that smartphone processor giant Qualcomm has officially joined the competition against Nvidia in the AI compute chip market. Qualcomm recently announced a partnership with Saudi Arabia’s emerging AI company Humain to jointly develop high-performance AI compute chips aimed at AI data centers.
According to the memorandum of understanding, Qualcomm and Humain plan to build cutting-edge AI data centers in Saudi Arabia, leveraging Qualcomm’s edge and data center solutions to provide efficient and scalable cloud-to-edge hybrid AI inference services for both local and international clients. They will also collaborate with Saudi Arabia’s Ministry of Communications and Information Technology (MCIT) to establish a Qualcomm semiconductor technology design center.
Notably, Humain has also recently formed partnerships with Qualcomm’s main competitors Nvidia and AMD, utilizing their chips in its AI data centers, showing a diversified cooperation strategy.
In fact, Qualcomm first ventured into the AI compute chip market in 2019 by launching its first data center chip, AI 100, leveraging its low-power, high-efficiency technology from the mobile chip sector to enter the AI inference computing market. Early customers included Meta. Although Meta acknowledged AI 100’s impressive performance per watt during testing, concerns over Qualcomm’s software maturity and stability in long-term tasks led to the collaboration falling through.
Nvidia’s leading position in AI chips is largely due to its robust software ecosystem — the CUDA platform and associated tools provide developers comprehensive support, enhancing AI model training and deployment efficiency. In comparison, Qualcomm’s AI chip software platform is still in its early stages and faces significant challenges.
However, AI 100 has seen some success, with Industrial Fulian (Foxconn Industrial Internet) being the first public customer, using AI 100 in servers analyzing traffic and security monitoring video streams, demonstrating the chip’s real-world application potential.
AI inference computing focuses on making real-time decisions based on trained models, such as recommendation systems, voice recognition, and image analysis. This requires high computational efficiency while maintaining strict latency and power consumption constraints — areas where Qualcomm’s expertise in low-power mobile chips is advantageous.
According to Gartner’s report, the global AI chip market is projected to reach $119.4 billion by 2027, representing enormous market opportunities but also intense competition. While Nvidia remains strong, it must continuously face challenges from traditional giants and emerging players.
