NewsCompany newsIndustry News

2019

04/15

AI chip "collision" security: why are giants obsessed with self-research?

MetInfo enterprise content manager system | MetInfo CMS

According to the statistics of China Semiconductor Industry Association, from 2013 to 2017, China's chip sales increased by 25.9 billion yuan to 541.1 billion yuan in five years, and doubled in 2017. China's chip market has achieved an average compound annual growth rate of 21.2%. By 2017, the overall growth rate of the industry has reached a new high in recent years, reaching 24.8%. The market has entered a period of rapid development.

Security industry from analog evolution to digital to network HD, and the current AI, without exception, benefited from the progress of chip technology, especially AI security chip is very concerned. To a large extent, the chip affects the overall function, technical indicators, stability, energy consumption, cost of the intelligent security system, and plays a key role in the future development of the security industry. In the past 2018, many chips have sprung up like bamboo shoots after a spring rain. All kinds of computing power and application scenarios have been produced. The number of floating-point operations per second ranges from 1T Flops to 16T Flops, from cloud to terminal, even the end-side also has a variety of modes, such as single chip and SoC integration.

These technological breakthroughs directly bring about the differentiation in the competition of security industry. In the field of video surveillance, for example, the chip can transform the video image content into structured data that clearly expresses the target attributes, and mine the data in depth, which can effectively improve the efficiency of data processing and make ordinary cameras more intelligent and powerful. And its high cost also makes the price of smart cameras remain high: the current market price of mainstream AI cameras is about 4,000 yuan, which is 3-4 times that of ordinary cameras.

In addition, although the development of chips has brought good news to security+AI, Moore's law is particularly evident in the current security chip industry. Although the supply of computing power is sufficient, it is very expensive. It is very inexpensive for many small and medium-sized security enterprises in China. The cheap and even GSP AI computing power is the life-saving straw for AI to achieve landing in the domestic security industry. The cost of chips is the key to the rapid popularization of intelligent security products.

In this context, the combination of chip and scene for product deployment has become one of the core competitiveness of security enterprises. Developing their own chips has become an important part of paving the way for the future development of security enterprises. Huawei Haisi, Junzheng, Fuhanwei, Guokewei and other veteran security chip providers are stepping up their own research and development and manufacturing.

The good news is that today's AI chips are far from meeting market demand. At present, the security cloud AI chip market has become more mature, and the major global technology giants have taken root. The terminal AI chip market needs to be expanded, which is the opportunity for many AI chip start-ups. In the face of the market of structural innovation, everything is worth expecting and imagining.

Concluding remarks: As the first stop of AI landing, security industry has trillion market scale, large user demand and fragmentation, with the natural advantages of chip landing, giants have built "cores" one after another, and the industry is naturally happy to see its success. Of course, the giant coring may have a great impact on the traditional chip manufacturers. As CNBC points out: You have chips, I have chips, everyone has chips. This trend may eventually threaten the traditional relationship between big buyers and big suppliers. But this will not affect the security enterprise's adherence to the chip, nor will it reduce the chip enterprise's interest in the security field.