AI chip scene in 2022
As the technology matures, the application scenarios of AI chips will not only be in the cloud and big data centers, but will also gradually move to the edge with computing power, and will be deployed in fields such as smart homes, smart manufacturing, and smart finance. The variety of products is increasing day by day, deployed in smart terminals such as smart phones, security cameras, and self-driving cars, and the variety of smart products is also increasing day by day. In the future, AI computing will be everywhere.
Cloud: Currently still the center of AI, higher-performance computing chips are needed to meet market demand
Currently, most AI training and inference workloads take place in public and private clouds, and the cloud is still the center of AI. Driven by the need for privacy, network security, and low latency, the cloud has seen the emergence of AI training and inference workloads on gateways, devices, and sensors. Higher-performance computing chips and new AI learning architectures will be the solution to these problems. the crux of the problem.
The Internet is an industry with strong demand for cloud computing power. In addition to traditional chip companies, chip design companies and other participants, Internet companies have entered the AI chip industry one after another, investing in or self-developed cloud AI chips.
Edge side: Data sinks to the edge. With the industry landing market, there will be large increments
. The development of 5G and the Internet of Things and the intelligent transformation and upgrading of various industries have brought explosive data growth. Massive amounts of data will be accumulated at the edge, and data analysis and processing based on the edge will greatly improve efficiency and reduce costs.
As a large amount of data sinks to the edge, edge computing will have a greater development. IDC predicts that in the future, more than 50% of the data will need to be stored, analyzed and calculated on the edge side. higher requirement. As an important basic hardware for realizing computing power, chips will also have more development. ABI Research predicts that in 2025, the edge AI chip market will surpass cloud AI chips.
Driven by artificial intelligence algorithms, edge AI cores can not only independently perform logical analysis and calculations, but also self-optimize and adjust strategies dynamically and in real time. Typical applications are black light factories.
Terminal equipment: The types of terminal products are gradually diversified, and the increase in shipments has led to a large demand for chips.
According to the data of 100 million euros, the growth rate of China's autonomous driving industry will reach 24% in 2022; the growth rate of smart camera product shipments will exceed 15%; The shipments of smart products such as mobile phones, tablets, and VR/AR glasses have also experienced rapid growth, generating a large demand for smart chips.
The types of smart terminal products are also gradually diversified. Consumer hardware such as smart audio, service/commercial robots, industrial products such as industrial/NC equipment, and communication products are becoming more and more abundant. Different product types also put forward more requirements on chip performance and cost.
Reference link
https://mp.weixin.qq.com/s/KgL37Kg_liTUlFU67S7jeQ