The advancement of technology is the driving force of previous industrial revolutions, and the human society is entering the intelligent society under the advancement of artificial intelligence technology, the so-called "third industrial revolution." As an artificial intelligence chip, one of the core technologies of artificial intelligence, how is its development status and how the future is going, this is the topic that this article hopes to discuss together.
The history and drivers of the development of artificial intelligence
Donella H. In her book The Beauty of the System, Meadows pointed out that “The first thing a person who is good at systematic thinking in the face of problems is to look for data and understand the history of the system and its trend graphs over time.†For the development of the current hot artificial intelligence, it is also necessary to review the predecessors of this world and the driving factors behind it. The author summarizes the following development history of artificial intelligence and its future development trends.
The development and trend of artificial intelligence
Throughout the history of artificial intelligence, the following factors played a key role in the "three times and two falls" of artificial intelligence:
Algorithm (A-Algorithm): One of the big reasons for the rise of this wave of artificial intelligence is the rise of deep learning algorithms, especially in terms of image and speech. Algorithms are the engine of artificial intelligence development. B-Big Data: The current artificial intelligence of this wave uses more statistical analysis methods. Therefore, whoever owns the data may be at the forefront of this wave of artificial intelligence. Data will become one of the core assets and production materials of future enterprises. It is a new energy source for the development of artificial intelligence. C-CompuTIng Power: Algorithms based on big data statistical analysis and deep learning based on parallel computing require powerful computing power, and the decades of development of Moore's Law have made great progress in computing power, effectively supporting this. The development of wave artificial intelligence. S-Scenarios: The long-term development of artificial intelligence must be able to feed back to the industry and bring real economic value. With the development of the Internet of Things and the digital wave of the physical world, it provides many practical applications for the application of artificial intelligence. Application scenarios, which will further accelerate the development of artificial intelligence.
Therefore, the combination of ABCs factors is driving the third rise of this wave of artificial intelligence, and I believe that this wave will last longer and the social and economic impact will be more profound. "Artificial intelligence will eventually look like Electricity has penetrated into all walks of life in society."
Market Analysis of Artificial Intelligence in China
We can map the above ABCs model to the analysis of the Chinese artificial intelligence market. After experiencing the catch-up of the Internet and mobile Internet, China is becoming an important data power, and China is expected to have 20%-25% of global data by 2020. One of the most important factors driving the development of this wave of artificial intelligence is data. In addition, the Chinese government is promoting the intelligentization and transformation of the informationization of China's industry through the "Made in China 2025" and digital China policies, which provides many practical application scenarios for the development of artificial intelligence. Therefore, in this wave of artificial intelligence, China is standing in a very favorable position.
Segmentation of China's artificial intelligence market and trend of AI chips
One of the important sources of innovation is the precise segmentation of the market. The artificial intelligence market in China can be roughly divided into four categories: data center/cloud training and reasoning, edge (fog) computational reasoning (and training), and device-side reasoning. The market opportunity of data center/cloud training and reasoning has been widely accepted by the industry, and it has further exploded with the practical application of artificial intelligence. As the cloud, especially the training part, needs more flexibility to cater to unpredictable application and data growth, it is expected that the computing platform of Intel CPU+AI accelerator card will be further developed, and the computing platform of the inference part is especially clear in the application scenario. In the case of the CPU and the low-power AI acceleration chip, the multi-chip sealed MCP (MulTIple-chip Package) will be an option for the future. In the end-side reasoning part of the device, it is expected that the integrated AI-accelerated IP SoC will eventually be a trend due to better power control, size and cost performance. At the market level, some of the leading companies such as Apple and Huawei are developing similar solutions for their latest smartphone products. Since the market trend is relatively clear but the application model still needs time to innovate and promote, the market segment is still in the growth stage. For edge (fog) calculations, with the development of the Internet of Things, especially for applications with high real-time requirements, such as autonomous driving, intelligent manufacturing, etc., the market is growing into a new market opportunity, but still in the market. The beginning stage. The market segment has relatively complex application scenarios, and its corresponding artificial intelligence computing platform will also present a variety of solutions.
The development and trend of artificial intelligence chips in China
Under the dual promotion of government and market capital, China's artificial intelligence chip industry is attracting a climax, and the participants are mainly from the following aspects:
New Entrepreneurial Companies: Leading figures in such companies generally have a strong background in artificial intelligence, and have a good balance between business awareness and technology research and development, such as Horizon, ThinkForce, and Exploration Technology. IntEngine) and so on. Currently, the number of these participants is growing rapidly. Large market leader/Internet company: led by TAB (Tencent, Alibaba, Baidu) + Huawei. With their own data sets, algorithms and application scenarios, they plan to develop more suitable artificial intelligence chips to optimize their algorithms and services. Due to its strong financial resources, research and development capabilities and data/application scenarios, it is expected that such participants will become an important force in the Chinese and even global artificial intelligence chip market. Old chip companies: These companies are represented by Huawei Haisi and Ruixinwei. They have very good SoC design experience and customers, and are developing SoC chips that integrate artificial intelligence to accelerate IP. Entrepreneurial companies in the background of colleges and universities: As colleges and universities have been adhering to the basic research of artificial intelligence chip design in the past few decades, they have accumulated considerable technology, and are currently combining industrial capital to promote the industrial transformation of their technology, such as cold. Cambricon, DeepHi, and Tsinghua University Microelectronics.
China's local artificial intelligence chip status and trends
Combined with the market model proposed in this paper, based on the relevant public information, the Chinese native artificial intelligence chip is analyzed. As can be seen from the figure, the products of China's current local chip companies are distributed throughout the market segment of artificial intelligence. In the data center / cloud training and reasoning chip part, Baidu, Alibaba, Huawei, Cambricon, BitMain and ThinkForce are represented, but in addition to Bitmain's reasoning chip, other chips are still under development. Considering TAB+Huawei's strong R&D capabilities and global resource allocation, I believe they will eventually gain a certain position in the data center/cloud inference chip. In the reasoning chip part of the device side, the founding chip company is more concentrated, its performance and power consumption can be matched with similar products overseas, but it is expected that the future competition of this market segment will be more intense and eventually lead.
Summary and recommendations
The current artificial intelligence is in the early stages of industrialization, and all countries are on the same starting line. The Chinese government has given high attention to artificial intelligence from top to bottom and completed a series of top-level design at the policy level. The possession of a large amount of data and the management of data sovereignty and the localization of application scenarios will further contribute to the rise of Chinese local chip companies. As foreign-funded enterprises rooted in China, they should also actively participate in the tide of artificial intelligence development in China, and carry out win-win cooperation with local companies in technology, market and talents to jointly promote the development and growth of China's artificial intelligence industry.
Cr12 Soldering lron,high strength and wear resistance.The cut surface of DIN Rail is smooth and free of burrs.With the 50cm lengthgraduated scale,cutting is accurate.The bezelcan be rotated at 90°,making cutting more convenient.Locking mechanism,humanized design, concenient for storage and transportation.
Crimp Tool WKC8 SeriesSuitable for VE ferrulesWKC8 10-4 Square shape after crimpedWKC8 6-6 Hexagon shape after crimped
Cutting plier ToolSuitable for cutting FBS, JEB bridges
Automatic wire stripping and crimping machineProduct introduction
This model is designed for crimping tube terminal, suitable for wires with cross section 0.5-4mm².Thedesktop type machine is space saving,Theproductionof wiring harness has strong mobility, vibration plate feeding, automatic trigger start,stripping Integratedoperation of stripping wire, insert into ferrules andcrimping is convenient and quick.It is suitable forcrimping, stripping variety of single-grain tubeterminals to wires at one time.
Touch man-machine interface numerical control operation
Stripping through the terminal without reverse wire
Horizontal crimping from left and right
Fool-proof design automatically triggers to start
Electrical Guide Aluminium Din Rail,Industrial Din Rai Power Supply,Mounting Din Rail For Terminal Block,Standard 35Mm Aluminum Din Rail
Wonke Electric CO.,Ltd. , https://www.wkdq-electric.com