Qualcomm firmly adheres to the 5G+AI development strategy and deploys cloud computing

On April 19, Qualcomm held an AI Open Day event with the theme of "Making AI within reach". At the meeting, Qualcomm brought us Qualcomm's strategic plan for AI.

It can be predicted that 2019 will be the first year of 5G commercialization. At the same time, the parallel deployment of 5G+AI will become a future technological trend, which can analyze and use massive data.

Meng Pu, Chairman of Qualcomm China

Qualcomm has achieved more than ten years of basic technology research and development results in the field of AI, and has also promoted the implementation and popularization of AI in different industries. At this conference, Qualcomm also joined hands with nearly 20 AI ecosystem partners to showcase more than 40 AI applications based on Qualcomm's AI Engine, covering a wide range of AI use cases such as photography, audio, games, translation, gesture recognition, AR, and the Internet of Things.

Establishing 5G+AI strategy

Qualcomm China Chairman Meng Pu said that 5G will redefine everything and usher in a new era of invention, and Qualcomm has always believed that the future will be a dual strategic trend of 5G+AI. Because with the advent of 5G networks, there will be billions of objects connected to each other. In the face of this massive amount of data, not only will there be higher requirements for cloud computing capabilities, but also data processing on the terminal side. AI computing capabilities will be the key.

Qualcomm's AI strategy is 5G+AI, which means combining 5G connectivity with AI research and development to help AI transform other industries through platformization. 5G has the characteristics of high capacity, low latency and high reliability, and it can fully support AI computing at the wireless edge. AI on the terminal side will also play an important role in fully realizing the potential of 5G and open up more application scenarios for 5G.

Qualcomm currently supports a complete cloud-to-end AI solution. On the terminal side, Qualcomm relies on the Snapdragon mobile platform to provide AI computing functions for more than 1 billion smartphones on the market.

Take the Snapdragon 855 mobile platform as an example. It integrates the fourth-generation multi-core Qualcomm AI Engine, including the new hardware core designed for AI processing - Hexagon Tensor Accelerator (HTA). In addition to outstanding performance in AI processing and computing power, the Snapdragon 855 is the world's first commercial 5G mobile platform, which can support 5G connectivity when paired with the Snapdragon X50 5G baseband.

In addition to the top-level Snapdragon 8 series mobile platform, Qualcomm also released the new Snapdragon 7 series and 6 series platforms targeting the high-end and mid-range markets this month, namely Snapdragon 665, Snapdragon 730 and Snapdragon 730G.

The Snapdragon 730 and 730G mobile platforms have many of the top features and functions of the previous 8 series, with a significant upgrade, including the fourth-generation multi-core AI Engine, which improves the processing speed of intuitive terminal-side interactions such as shooting, gaming, voice, and security, and the AI ​​computing power is twice that of the previous generation platform Snapdragon 710. In addition to smartphones, Qualcomm has launched product platforms that integrate its AI Engine for mobile computing, XR, IoT, speakers, and automobiles.

Deployment of cloud AI field

Qualcomm has been working deeply on the terminal side, and with the explosive growth of cloud data in the 5G era, Qualcomm has also begun to focus on cloud AI.

According to Qualcomm data, the market size of AI data centers will increase to $17 billion by 2025. With the advent of 5G networks, cloud data will become increasingly massive, and the market size of AI data centers will also see explosive growth.

"There are more than 200 trillion inference operations every day," said Ziad Asghar, vice president of Qualcomm Technology. Qualcomm seized the opportunity of cloud AI and, relying on its design advantages in the mobile field, officially launched the newly built Qualcomm Cloud AI 100 accelerator for cloud applications.

This chip has a complete software system, and its underlying framework supports Tensflow, etc. At the same time, it adopts a 7nm process, and its AI performance is about 50 times higher than that of Snapdragon 855, which can meet the AI ​​computing needs of various terminals.

It can enable distributed intelligence to spread from the cloud to all nodes between terminals. Qualcomm will provide developers with complete tools and framework support. The product will start sampling in the second half of 2019.

Actively promote AI cooperation

Qualcomm believes that the development of AI and even the entire ecosystem requires the advancement of the entire industry. Therefore, Qualcomm has established cooperative relationships with many cloud service providers, terminal manufacturers and AI software developers around the world.

At this open day, Qualcomm, together with 20 domestic and international partners, demonstrated more than 40 AI applications based on Qualcomm AI Engine.

On-site displays included smartphone products equipped with Snapdragon mobile platforms from OPPO, Samsung, vivo, Xiaomi, etc., which used Snapdragon mobile platforms or AI development kits based on Snapdragon platforms to showcase a variety of applications. Among the manufacturers were ArcSoft, Baidu, Elephant Audio, Avatar, Loom.ai, Megvii, Mobius, Morpho, SenseTime, Tencent Honor of Kings, Tencent AI Lab, ThunderSoft, Youdao, etc.

There are four main categories of AI applications, including smart audio and video, entertainment games, life tools and industry applications, such as mobile AI e-sports teams, AI call intelligent noise reduction, AI howling suppression, AI smart super night scene, AI smart beauty photography, AI smart video blur, real-time voice translation, AR full-screen translation, unmanned supermarkets, etc.

On the day of the conference, Qualcomm, vivo, Tencent Honor of Kings and Tencent AI Lab also jointly announced that they are using the fourth-generation Qualcomm AI Engine to cooperate in the "Imagination Project" project to jointly promote and explore new experiences of terminal-side artificial intelligence applications. The first landing project was pioneeringly applied to vivo's iQOO mobile phone, migrating the AI ​​reasoning capabilities of mobile games from the cloud to the terminal side on a large scale for the first time. In the open day experience area, participants formed teams to compete with the AI ​​e-sports team "SUPEX" on the spot and experience the human-machine duel firsthand.

Of course, the development of AI technology has also brought new growth opportunities to the field of Internet of Things. AI applications for the Internet of Things cover homes, industries/enterprises and smart cities, including manufacturing automation and robots, home and enterprise-level smart security, smart displays and speakers, smart agriculture, home control centers and smart appliances, sustainable cities and infrastructure, digital logistics and retail, and more.

At the same time, Qualcomm is promoting innovation in the automotive industry through products and technologies that meet market demand. The company has currently received a total valuation of more than $5.5 billion in product design orders. For products and markets at different levels, Qualcomm is supporting AI on a large scale, providing rich and powerful solutions for digital cockpits, autonomous driving, and smart transportation.

Qualcomm has also done a lot of work in AI basic technology research and overall industry innovation. As early as 2007, Qualcomm launched its first AI research project, and in 2018, Qualcomm established Qualcomm AI Research to further strengthen the integration of cutting-edge artificial intelligence research within the company. In addition, Qualcomm established an AI venture capital fund with a total amount of up to US$100 million last year to invest in start-ups that transform AI technology around the world.