As early as 2020, Qualcomm proposed the concept of 5G+AI, at that time, AI was far less hot than it is now, and AI was also mainly used in AI application scenarios such as image shooting on the Qualcomm Snapdragon platform. This year, with the popularity of generative AI, Qualcomm has significantly accelerated the promotion of 5G + AI, and the application scope of AI in the field of mobile communication has also increased significantly. At the 2023 World 5G Conference held earlier, Li Yan, vice president of technical standards at Qualcomm, shared many of Qualcomm's efforts in the field of 5G + AI in his keynote speech on "Towards AI-native communication system design", from which it can be learned that Qualcomm is actively promoting the introduction of AI into 5G standards and the progress of popularizing generative AI on the device side.
Qualcomm applies AI to multiple aspects of 5G network operation. For example, Qualcomm introduced AI technology in the planning and design stage of 5G mmWave networks, generating digital twins of real-world scenarios through machine learning Xi, and quickly iterating and optimizing the network in large models to find more efficient deployment locations. For example, AI technology is used to control 5G virtual private networks in real time to avoid interference and improve the reliability and performance of 5G industrial networks. Qualcomm is also pushing to bring more AI technologies to the 3GPP Release 18 5G standard, with a focus on three key wireless AI use cases: channel feedback, beam management, and precise positioning.
Qualcomm is also actively promoting the large-scale development of generative AI on the terminal side, and the popularization of generative AI is currently facing some difficulties, such as the supply of cloud computing hardware cannot keep up with the market demand, and cloud computing power has become one of the main bottlenecks in the development of generative AI. Qualcomm proposed the concept of hybrid AI, which can greatly accelerate the large-scale development of generative AI through the integrated development of device-side AI and cloud AI, and through the support of massive terminal computing power. In February this year, Qualcomm exhibited a generative AI model running 1 billion parameters on a smartphone based on the second-generation Snapdragon 8 platform, which can complete 20 steps of inference in 15 seconds and generate a **. And in the third-generation Snapdragon 8 platform released in October, the time to complete the same task has been reduced to 06 seconds. At present, many Chinese smartphone manufacturers have launched 5G smart flagship phones based on Qualcomm's third-generation Snapdragon 8, such as Xiaomi, Redmi, OnePlus, iQOO, Red Devil, Meizu, etc., with the efforts of Qualcomm and partners, generative AI is expected to be integrated into people's daily lives more quickly.
The combination of AI and 5G mobile communication technology can trigger a "technological revolution" in mobile communication, and 5G communication and AI have complementary advantages, which can greatly improve the service quality of 5G communication systems, thereby comprehensively improving the performance and user experience of 5G. Looking ahead, Qualcomm is also looking forward to integrating AI-related technologies into 6G system design.