The "ZTE event" that is in full swing is continuing to ferment. The US Department of Commerce issued an export ban on ZTE, which is not only a chain reaction brought about by the escalation of Sino-US trade wars, but also highlights the high-tech industries in China and the United States. The strength gap. In modern warfare, the game of science and technology will be an important force, and China and the United States have already fought several times. As early as April 2016, Chinese private equity investment company CanyonBridge proposed the acquisition of Lattice. On November 3 of the same year, Lattice accepted the offer and announced that it would be acquired by CanyonBridge for $130 million in all outstanding shares, including its own debt. Lattice is one of the top four FPGS in the world. The other three are Silinx, Altera (acquired by Intel) and Actel (acquired by Microsemi). Headquartered in Portland, Oregon, USA, it produces communications chips for use in automobiles, computers, mobile devices and other devices, as well as military communications. It is one of the few manufacturers that can manufacture programmable logic chips. one. This is a business that is willing to buy and sell, but it is considered by the US Foreign Investment Committee to be a threat to the national security of the United States and cannot be approved. On September 14, 2017, Trump issued an administrative order directly to stop the transaction. The efforts of Chinese-funded institutions to acquire FPGA technology through acquisitions have failed. Another important event was that in February 2015, the National Development and Reform Commission of China announced a fine of RMB 6.008 billion for Qualcomm. In March 2017, the United States imposed a fine of 892 million U.S. dollars on ZTE, which is also close to 6 billion yuan. In April 2018, before the ZTE incident, the Chinese Ministry of Commerce continued to delay the approval of Qualcomm's acquisition of NXP, and it also sought more protection for Chinese companies. The battle for chips between China and the United States is so fierce that it also makes the Chinese people realize the pain of China's chips. After ZTE was sealed, the country triggered a large-scale discussion. At this time, an AI chip acquisition incident caused a large-scale concern. On April 20, Alibaba Group acquired the only independent embedded CPU in mainland China. IP Core - Zhongtian Microsystems Co., Ltd. On April 19th, Alibaba Dharma announced that it is developing a neural network chip, Ali-NPU. At this point, Ali has stepped into the chip field through independent research and development, acquisition and other means. In the past two years, artificial intelligence has begun to rise globally, especially in China and the United States. The rise of artificial intelligence technology is benefited from the improvement of computing power. GPU, NPU and other chips have contributed greatly, which has greatly accelerated the iterative and research and development speed of artificial intelligence products. Whether this war will affect the artificial intelligence industry, especially the medical artificial intelligence industry that we are concerned about. If the Sino-US trade friction is escalated again, will the medical artificial intelligence company be “snapped to the neck†like ZTE, and the arterial network consulted many industry experts on this question. Medical AI industry's chips rely on the United States Since the research of medical artificial intelligence requires calculation of large-scale data, the demand for chips is very large. There are two main scenarios for using the chip to perform data operations in the field of medical artificial intelligence. One is to use the chip to perform data calculation and AI product iteration during the training of the laboratory model. The other is to embed the chip in an image workstation or medical device after the product is developed. At present, there are four main AI acceleration chips: CPU, GPU, FPGA, and ASIC (Google TPU). Among them, GPU is the most widely used general-purpose chip, which is mainly benefited from the promotion, efficiency and price of NVIDIA. The efficiency of the CPU is too low, while ASICs and FPGAs are custom and semi-custom chips. Although the efficiency is high, the low demand leads to low productivity and high price. In the model training phase, companies use more general-purpose chips. In the process, it will be carried out through NVIDIA GPUs, Intel CPUs and FPGAs from other companies. By the application stage, the landing of medical AI products is also highly dependent on the chip. One is a cloud-based solution that puts products in the cloud to provide medical services to customers, which can reduce costs, but the real-time performance is slightly insufficient. Another way is to integrate the AI ​​algorithm and software system into a custom chip and load the custom chip into the medical device to reduce power consumption, ensure system performance, and reduce device size. It can be seen that both the model training and the scene application have high requirements on the chip. At this stage, although China's medical AI companies are no different from European and American countries in terms of software and algorithms, the chips used are almost all dependent on US imports. There are not many companies with AI chip design capabilities in China. The Cambrian is a representative one, but its packaging production process is still lacking. Will the US ban AI chips from exporting to China? At present, almost all chips for AI products rely on imports, whether it is NVIDIA GPU, Intel CPU, or Google's TPU, the exporting countries are all in the United States. After the ZTE incident, will the medical AI company encounter the same thing? Industry experts told the arterial network reporter that the possibility is very low. There are three reasons: First, ZTE violates the laws of the country where the business is located. ZTE was banned because ZTE made false statements to the Industry and Security Bureau in 2016 and 2017. On April 16, 2018, the US Department of Commerce's Bureau of Industry and Security, ZTE, did not promptly deduct bonuses and issue disciplinary letters from certain employees involved in historical export control violations, and on November 30, 2016 and 2017, 7 In the two letters submitted to the US government on the 20th of the month, the decision was made to activate the rejection order of ZTE and ZTE Kangxun. Unlike ZTE, domestic AI companies only use chips for product development and domestic operations, and have not yet involved export sales. In particular, the medical AI industry is still in the research and development stage, and there is no large-scale implementation of commercial applications, which does not pose a threat in itself. Second, China’s chip purchases are huge and it is impossible to completely block it. Source: Gartner January 2018 According to Gartner survey data, among the ten companies that purchased the most chips in 2017, there are three Chinese companies, Lenovo, Huawei, and BBK, ranking 4th to 6th in the world. ZTE is not on the list. In 2017, Lenovo purchased 14.671 billion US dollars, Huawei purchased 14.259 billion US dollars, and BBK purchased 12.103 billion US dollars. In contrast, ZTE’s chip purchases are not the largest, especially in large companies such as Broadcom and Qualcomm. The proportion of purchases is not too high. According to reports from the Wall Street Journal, 65% of Qualcomm’s $22.3 billion in revenue in FY2017 was from China, compared with 57% in FY2016. Of the revenues of Broadcom in 2017, 54% came from the Chinese market. From these data, it is seen that the ban on ZTE will not hurt the muscles of large American companies. On the contrary, China's purchase of US chips is huge. Once the chip export is completely banned, it will have a huge impact on the US semiconductor industry. 3, the communication chip and AI chip overlap is not high The chip that is forbidden to be sold to ZTE in the United States is not highly coincident with the chip used by AI. The main chips used by AI are GPU, CPU, FPGA, etc. Although these are also imported from the United States, the types of chips used by AI and ZTE do not coincide. Therefore, the impact on the AI ​​company is almost zero in terms of the chip ban. FPGA will be the focus of the chip in the AI ​​field Although the United States is not likely to completely ban chips, the core technology is subject to people. In this "ZTE event," the people feel extremely shy and helpless. During this time, we also learned that independent research and development of chips has become an important trend in the next step of medical AI enterprises. In addition to chip design companies, medical AI companies including Imagination Technology and Heterogeneous Heterogene are also involved in chip research. According to the “Comprehensive Interpretation of the Development of Artificial Intelligence Industry in China and the United States†released by Tencent Research Institute in 2017, from the perspective of the number of chip companies at the basic level, there are 33 in the United States and 12 in China. The United States has both technology giants such as Google, Intel, and IBM, as well as chips such as Qualcomm, NVIDIA, AMD, and Xilinx. There are also many well-developed medium-sized companies and active start-ups. Among the chips used by Medical AI, GPUs and CPUs are mainly used for deep learning algorithms of AI. Whether it is chip architecture, patent or ecology, these two fields are firmly controlled by NVIDIA and Intel, and there is no chance of winning. It can be said that the United States is completely monopolized. Chinese AI companies want to make a difference, only in the development of FPGAs, ASICs and brain-like chips, and currently mainly small and medium-sized companies. Image source: Xinzhiyuan, part of the information source Tencent Research Institute In the AI ​​field, the future development of chips is most likely to break through the FPGA chip. The Chinese name of FPGA is called field-effect programmable logic gate array. It is a semi-custom circuit in the field of ASIC. It not only solves the shortcomings of the full-custom circuit, but also overcomes the limited number of gates of the original programmable logic device. Disadvantages. High performance, low power consumption, high flexibility and hardware programming. If the CPU and GPU are "universal" at the architectural level, the FPGA is "universal" at the lower level of the circuit. After programming the FPGA in hardware description language, it can simulate the architecture of any kind of chip, including the architecture of CPU and GPU. Baidu's machine learning hardware system uses FPGA to build AI proprietary chip, which is made into AI proprietary chip version of Baidu brain - FPGA version of Baidu brain, and then gradually applied in the large-scale deployment of Baidu products, including speech recognition, advertising. Click rate prediction model, etc. Yasen Technology CEO Chen Hui introduced to the arterial network, FPGA canceled the concept of memory, a lot of information transfer efficiency is very high, it can be transferred directly from one unit to another, without buffering in main memory, especially suitable for real-time requirements A higher algorithm. Therefore, the integration of FPGA and AI algorithms will become the main research and development direction in the future. At present, many companies such as Shenjian Technology, Shenzhen Ziguang, Shanghai Anlu and Horizon Robot are engaged in FPGA related research. It is worth noting that the industry barriers in the FPGA field are very high, and nearly 9,000 patents have established long intellectual property barriers. Strong as Intel is also sighed, can not buy Altera to get FPGA tickets for $16.7 billion. According to Chen Quan, a technology CEO, if China conducts chip research in the country, it can make high-quality chips, but the industrial barriers of chips are very high. Even if you master the basic theory, you need in chip design and industrial packaging. Long time research. Since its establishment in 1968, Intel has achieved its achievements today after years of development, and China still has many shortcomings in the production of chip industry. Ding Xiaowei, the founder of Voxel Technology, said that the chip gap between China and the United States has been very early, and the industry is constantly discussing. The ZTE incident only intensifies this matter. However, although there is a gap between China and the United States in the chip field, China Medical AI has its own core technology in application and AI technology. China and the United States also each have their own advantages, but the two sides' work processes and methodologies have their own characteristics, and there are differences in defining medical problems and clinical integration. The understanding of the ZTE incident should be rational. The research and development of the chip needs constant trial and error. It is a long-term accumulation. We need to improve the R&D environment and continue to invest in order to retain talents instead of being hot. China's chip development needs BAT and startups to work together The pain of China's chips has made us pay more and more attention to core technologies. The chip industry has high technical barriers, long development cycle, and very high requirements for funds and teams. A few days ago, Ali acquired the message of Zhongtianwei. Many people think that big companies such as BAT and Huawei are rich in capital and attach importance to the development of AI, so they have to assume the responsibility of the AI ​​chip savior. This view has some truth, but the development of AI chips in China requires not only leading companies such as BAT and Huawei, but also the need for startups to work together. On the one hand, BAT has strong R&D strengths and needs, and they actually have a layout. In March 2017, Tencent Cloud announced that it has formed a full-matrix AI infrastructure computing platform for FPGAs, GPUs and 25G NIC cloud servers, and announced a series of technical and ecological deployments, including the introduction of a 4-node FPGA cloud server and a machine 8 The card's GPU cloud server, as well as the 25G network card will be deployed on the FPGA and GPU cloud servers, to provide network infrastructure for subsequent GPU clusters and FPGA clusters. In August 2017, Baidu cooperated with Xilinx to release XPU at the Hot Chips conference in the United States. It is a 256-core, FPGA-based cloud computing acceleration chip. Baidu also released the DuerOS smart chip, however, this chip is integrated by the Violet Zhan Rui RDA5981, using the ARM mbed OS kernel and its secure network protocol stack. BAT's investment will solve some of China's AI chip dilemma in the future. Image source: Xinzhiyuan On the other hand, we can't ignore the strength of startups. BAT is often laid out by acquiring startups. The startup team in the AI ​​chip field will be more professional and understand more deeply, such as companies like Cambrian and Ziguang Guoxin. Cambrian is a well-known AI chip research and development company in China. It is the world's first AI chip company that successfully streams and has mature products. It has two product lines: terminal AI processor IP and cloud high-performance AI chip. The Cambrian 1A processor (Cambricon-1A) released in 2016 is the world's first commercial deep learning dedicated processor. The Cambrian team originated from the Institute of Computing Technology of the Chinese Academy of Sciences, the first national academic institution in China specializing in comprehensive research in calculator science and technology. A group of high-tech enterprises such as Lenovo and Shuguang were born from the institute. It is also an important shareholder of Cambrian Technology and a long-term partner of industry, education and research. In August 2017, Cambrian Technology completed a $100 million Series A financing, which was initiated by SDIC, Alibaba Venture Capital, Lenovo Venture Capital, Guoke Investment, Zhongke Turing, Yuanhe Origin, and Yongyu Investment. After this round of financing, the company ranks among the unicorns. The Arterial Network interviewed a number of medical AI institutions. When choosing a custom AI chip, they first considered the Cambrian products. In the FPGA industry, the “national team†Shenzhen Ziguang Tongchuang also has a deeper layout. The company was founded in 2013 and is a subsidiary of the listed company Ziguang Guoxin Co., Ltd. In November 2017, Ziguang Guoxin increased its capital for Ziguang Guochuang and established Chengdu R&D Center with a total investment of approximately 597 million yuan to promote the development of FPGA and corresponding EDA tools. Ziguang Tongchuang announced on the homepage that it has made China's first multi-gate high-performance proprietary FPGA. Another startup, Shanghai Anlu Technology, has its core team from Lattice, one of the four FPGA giants. At present, Shanghai Anlu Technology has completed the C round of financing. The investors include Huada Semiconductor, Hangzhou Silan Micro and Shanghai Municipal Government's integrated circuit fund “Shanghai Technology Venture Capital Co., Ltd.â€. It can be said that BAT has funds and needs. The startup company has technology, understands the industry, and has experience. The cooperation between the two parties will accelerate the development of the chip industry. Finally, I would like to thank Ding Xiaowei, CEO of Voxel Technology, Ding Xiaocheng, CTO of Wofang Technology, Chen Hui, CEO of Yasen Technology, Song Jie, CEO of Xi's Heterogeneous, Chen Kuan, CEO of Pushu Technology, and Li Yiming, CTO of Shenrui Medical. Dental Storage Cabinet,Dental Cabinet,Mobile Dental Cabinet,Insrument Cabinet For Dental Clinic Foshan Ja Suo Medical Device Co., LTD , https://www.jasuodental.com
China Medical AI's chips are all from the United States. Will they be stuck in the neck like ZTE?