From June 28th to 29th, the 8th Jiwei Semiconductor Conference 2024 with the theme of "Crossing Boundaries and New Quality Future" was held in Xiamen International Convention Center Hotel. At the main forum held on the morning of the 29th, Meng Feng, chairman of Qualcomm China, gave a keynote speech entitled "5G+AI: Bringing "Chip" Opportunities".
China's 5th anniversary of commercial use, 5G has entered a new stage
Meng said that Qualcomm, as an important member of the mobile communication industry chain and the mobile Internet industry chain, has been promoting the application and popularization of mobile technology in all walks of life. As the world's largest consumer electronics, mobile phones are closely related to people's daily life and work. Although the annual shipment of mobile phones worldwide has dropped to around 1.2 billion units in recent years, it is still a huge market. More importantly, China has formed a complete industrial chain from design to production in the field of mobile smart terminals, especially mobile smartphones, and has made great contributions to the global mobile smart industry. In 2023, the Chinese smartphone market will be about 270 million units, making it one of the most important smartphone markets in the world. At the same time, about two-thirds of the world's smartphones are manufactured in China every year, including smartphones made in China by foreign brands.
"It can be seen that the smartphone industry is very important for China's development, especially for semiconductor companies. Whether we are engaged in terminal SoC design or other semiconductor component manufacturing, our common goal is to serve the end OEMs of the industry chain and help the sustainable development of China's consumer electronics in the global market. Meng Feng said.
Meng pointed out that in the past few years, 5G technology has made significant progress on a global scale, especially in China. Since 2019, 5G has been commercially available in the global and Chinese markets for five years, and is now in the second phase of 5G development, 5G Advanced. Earlier this month, the mobile communications industry celebrated the fifth anniversary of China's 5G license issuance, with the Ministry of Industry and Information Technology, major mobile operators, and local governments launching new network construction and technology deployments of "5G Advanced (5G A)" in several cities across the country. This stage will bring lower latency and more IoT terminal support for many industrial applications, thereby promoting the rapid development of the mobile communication industry chain in all walks of life.
Terminal AI has advantages such as personalization and privacy
When it comes to artificial intelligence (AI), Meng said that as early as the 2021 Shanghai International Import Expo, Qualcomm proposed "5G + AI empowers thousands of industries" - based on years of technology research and development and follow-up, Qualcomm recognizes that AI and 5G will be more and more closely integrated in many aspects.
"In particular, AI has made significant progress in the past year or two with the advent of large models and the popularization of cloud-based AI applications. However, for Qualcomm, especially in the field of semiconductors, the final product landing is crucial. Whether it is a large model in the cloud or a variety of AI applications, they ultimately need to be implemented on terminal devices. That's why we've been pushing the boundaries of hybrid AI. Meng Feng said.
Meng pointed out that the AI on the terminal side not only helps the implementation of cloud AI, but more importantly, can make up for the shortcomings of cloud AI - because cloud AI is usually based on a general large model, it can handle some tasks that are not real-time and have little to do with personal characteristics. However, everyone's daily applications and mobility involve a large amount of personal data, which is often stored on mobile devices, and people may be reluctant to upload it to the cloud for machine learning and sharing.
In addition, from the perspective of mobile terminals, with the development of built-in sensors, they can sense location, activity, and needs to provide more personalized services. Therefore, in order to provide timely personalized applications and protect personal privacy, the implementation of AI technologies, including generative AI, on the device side has become crucial, which is also a concern of Qualcomm in the process of promoting the development of AI.
Generative AI presents a huge opportunity for the semiconductor industry
Meng emphasized that the evolution of generative AI has a profound impact on the semiconductor industry, whether it is in cloud data centers or terminal equipment, whether it is the demand for hardware such as CPU (central processing unit) and GPU (graphics processing unit), or the demand for bandwidth and various applications, which is a huge opportunity for the semiconductor industry.
Taking Qualcomm as an example, Meng introduced two important products that were announced at last year's summit in Hawaii. The first is the world's first third-generation Snapdragon 8 mobile platform launched for the application of generative AI in devices, which has now realized the application of generative AI in mobile phones with up to 10 billion parameters. The second is the release of the new Snapdragon X Elite for PC, which supports generative AI applications with up to 13 billion parameters.
More than half a year has passed, and we have seen dozens of flagship phones equipped with the third-generation Snapdragon 8 chip launched around the world. In the future, more smartphones with generative AI capabilities will appear on the market.
Meng Feng further introduced that in May 2024, Qualcomm and Microsoft jointly released the concept of artificial intelligence computer (AI PC), which will be equipped with a new Windows 11 operating system, which reflects the new technical requirements and changes brought about by the AI era. According to Windows' definition of an AI PC, it needs to have at least 40 TOPS of AI processing power in order to run various large models built into the PC. Chips such as the X Elite and X Plus released by Qualcomm last year have AI processing power of up to 45 TOPS, and their performance is at the world's leading level. Up to now, more than 20 AI PC products released by major PC manufacturers around the world, including Lenovo, Dell, HP, Acer, Asus, Samsung, and Microsoft, have used Qualcomm Snapdragon X-series chips.
Join hands with industry partners to promote the development of intelligent and connected vehicles
In Meng Feng's view, with the development and deployment of 5G, mobile terminals are no longer limited to traditional smartphones, but include PCs, intelligent connected cars, various XR devices, intelligent terminals of industrial manufacturing, and various robots, all of which represent the development of mobile terminals. Therefore, with the parallel development of new energy technology and intelligent network technology, the development of intelligent network vehicles and new energy vehicles in China is also very fast. In terms of intelligent networked vehicles, Qualcomm's product portfolio, the Snapdragon digital chassis, mainly includes intelligent connected platforms, digital cockpits, intelligent driving, and vehicle-to-cloud platforms.
According to Meng Feng, in terms of intelligent connection, as early as 2002, General Motors of the United States used the OnStar service in its high-end car series, which used Qualcomm's 3G code division multiple access (CDMA) technology as the connection technology. Today, approximately 350 million vehicles worldwide are equipped with mobile, high-speed connectivity solutions from Qualcomm.
Meng pointed out that in recent years, both traditional and emerging car companies have launched intelligent models that have become more and more popular. Over the past three years, Qualcomm has supported more than 50 Chinese brands in launching more than 160 connected cars, and the speed of adoption of such connected vehicles is impressive. Therefore, Qualcomm looks forward to working together with colleagues in China's automotive industry and semiconductor integrated circuit industry to further accelerate the development of intelligent networking of China's new energy vehicles.
"In terms of AI loads, due to Qualcomm's heterogeneous architecture, which includes CPUs, GPUs, digital signal processors (DSPs), and NPUs, we currently mainly use NPUs to handle AI, especially generative AI needs. However, these chips require not only hardware, but also software capabilities and ecological support. Therefore, we have been working with ecosystem partners to promote the application of AI, especially generative AI, both domestically and globally. It can be seen that with the application of large models in smartphones, AI PCs, and intelligent connected vehicles around us, the specific application of generative AI in various industries is accelerating. Meng Feng said.