Episode network report (text / Li Ying) In the early morning, the domestic AI chip unicorn - Shen Jian Technology announced that it was the originator of FPGA, the acquisition of Xilinx, the United States, the specific transaction amount is unknown. Huge response, Xilinx exclusive to the micro-network, Xilinx's strategy to evolve from FPGA devices to adaptive computing acceleration platform providers is to accelerate the deployment of FPGA acceleration technology from cloud-to-end applications. The technology-optimized neural network pruning technology runs on Xilinx FPGA devices to achieve breakthrough performance and industry-best energy efficiency. This is the new Xilinx CEO's promotion of machine learning development after the ACAP March conference. Capacity, a major step in advancing strategic development.
Acquisition of waterways?
In fact, the two have had years of friendship.
Founded on March 3, 2016, Shen Jian was founded by Wang Yu, Han Song, Yao Wei, and Shan Yu from the background of Tsinghua Electronics. It focuses on the realization of AI landing with FPGA. Due to its profound AI technology advantages and industrial advantages, The market is popular.
Since its inception, Shenjian Technology has been developing AI solutions based on Xilinx FPGA platform. It has launched products for drones, security, and data centers. It is known for its neural network pruning, deep compression technology, and system-level optimization. Since then, the self-developed AI chip 'Tingtao' and 'Viewing Sea' have been launched.
It should be noted that since Actel, Altera was acquired by Microsemi and Intel, Xilinx has taken the leading position in the FPGA industry. Shenjian uses Xilinx high-performance FPGA platform to integrate original deep learning technology and accumulate The deep AI technology advantage and industrial advantage have been favored in the market.
In the financing process of the past two years, Shenjian Technology completed the Angel Wheel in April 2016. The investor is Gaochun Capital, Jinshajiang Venture Capital. In May 2017, the A round was completed, and the financing amount was tens of millions of US dollars. The party includes Xilinx, MediaTek, Tsinghua Holdings, Fanghe Capital, the original investor Jinshajiang Venture Capital, Gaochun Capital and Investment. In October 2017, Shenjian Technology announced a $40 million A+ round of financing, of which Ant Gold Serve with Samsung Venture Capital, China Merchants Venture Capital and Huachuang Capital.
It can be seen that since 2017, Xilinx has become a major investor in Shenjian Technology together with other well-known investment institutions around the world.
The acquisition seems to be a big joy. Xilinx said that the FPGA is the most suitable product for innovation and entrepreneurship. The deep learning program of Shenjian Technology has been developed on the Xilinx platform. In the past two years, Xilinx has not only witnessed It has also been accompanied by the growth of this company, including last year's investment, introducing end customers to help them grow. Nowadays, acquisitions can be said to be a matter of course, and the two sides are definitely more effective than synergy.
Everything for cloud + end AI?
What kind of chemical reaction will occur after the two sides combine? What changes will the future layout change?
Xilinx FPGAs are mainly used in traditional fields such as communication, industry, aviation, and defense. However, with the emergence of emerging technologies such as AI, cloud computing, 5G, and autopilot, the application range of FPGAs is expanding. FPGA-based AI machines Learning technology can cover almost all mainstream applications including 5G, automotive, industrial, medical, consumer electronics, data centers, etc.
Similar to CPU and GPU, FPGA chip can provide computing power for training and reasoning of deep learning algorithms. Especially in reasoning applications, FPGA has advantages in power consumption and performance, and becomes one of the core technologies of AI application. See, FPGA has now entered the ranks of AI mainstream processors, from end to edge to cloud, FPGA with rapid power performance ratio and adaptive capabilities, will be rapidly iterative development.
In terms of AI cloud computing, FPGAs will have flexibility, lower power consumption, and higher ROI in terms of data center acceleration. Therefore, most data centers will adopt FPGA+CPU+GPU mixed mode. Side, FPGA has the advantages of flexibility, low power consumption, and has the advantage of multiplexing. It can bring security, confidentiality and multi-sensor fusion technology with only one chip. In the process of continuous improvement and continuous optimization of FPGA, FPGA The flexibility advantage will always be welcomed.
At the cloud computing level, Xilinx's partners are not only foreign Amazon's cloud service leader Amazon AWS, but also domestic Ali, Baidu, Tencent, and Huawei. On the AI end side, Xilinx FPGA chips are in the cloud. Automated driving, embedded vision and industrial Internet of Things are already accelerating penetration.
Especially at the level of autonomous driving, both Xilinx and Shenjian Technology have added horsepower. Xilinx is the second largest ADAS semiconductor supplier, between the fiscal year 2013 and the 2017 fiscal year, Xilinx vehicles The average annual growth rate of revenue in the field reached 60%. It is working with several major automakers to develop an autopilot program. On June 26, Shenjian Technology announced the launch of DPHiAuto, an embedded AI computing platform based on FPGA. Joined the autopilot track. It is reported that it has signed cooperation with well-known car manufacturers and first-tier suppliers in Japan, North America, Europe and China.
In the AI field, Xilinx and Shenjian will work together to leverage technology, channel, platform and landing advantages to advance FPGA deployment of machine learning from cloud-to-end applications. Will a startling AI opponent emerge? Will there be similar actions for the target NVIDIA or Intel?
In any case, the series of actions after the new CEO of Xilinx took office showed that Xilinx wants to play the next big game!
AI chip competition into the watershed?
Shen Jian Technology, Cambrian, Horizon as a representative player of three domestic AI chips, is the top companies in this round of AI chip entrepreneurial wave, their every move is eye-catching.
Nowadays, Shenjian Technology has turned around perfectly. For other startups, it is also self-aware. The AI chip has always faced difficulties in landing, and it is difficult to find application scenarios. Now one of the three representative players is the acquisition of Shenjian Technology. The case may indicate that the AI chip market is about to enter a new wave of integration.
The AI chip battle will enter a new watershed, the capital bubble is coming, and many have been on the verge of bankruptcy. In the future, the AI chip track will be more realistic.
Yao Jian, co-founder and CEO of Shenjian Technology, said that today's AI chip design has crossed the stage of independent chip design and application, and entered the stage of software and hardware coordination. Only in this way can the chip performance be better. In order to allow customers to adapt the chip to the application scenario. However, the difficulty for AI chip companies is that the X86 instruction set is a mature instruction set 30 or 40 years ago, and there are countless developers developing processors and applications for it. For AI chips, there won't be too many developers to develop operating systems, processors, and upper-level software environments for chips. For AI chip companies, in addition to algorithms, performance and applications continue to adapt, We must constantly strive to build relevant ecology.
The final chip competition is not performance, but service. 'NVIDIA products are not much better than others from the perspective of cost performance. But why in the server field, customers are willing to endure five or six times more expensive than others, and also use NVIDIA What's the product? This is because in the GPU and vertical industries, NVIDIA can also provide vertical industry services. ' Yao Yao's words are simple and profound.
Therefore, players gathered on the AI chip track, future competition is not just performance, how to combine chips and application scenarios, how to build an ecosystem, how to provide services from chip to board to software application layer, future competition The road is still long.