Cambricon aims to be the first AI chip stock on the Sci-Tech Innovation Board: incubated by the Chinese Academy of Sciences, valued at 17.5 billion after 4 years of establishment, with Alibaba and Lenovo as shareholders
Qian Mingyuyang from Yunaofeisi
Quantum Bit Report | Public Account QbitAI
Cambrian has clearly stated that it is preparing for an IPO.
After four years of establishment, it has won four rounds of financing and its valuation has reached 17.5 billion yuan (2.5 billion US dollars). As China's fastest-growing AI chip unicorn, it will be listed on the Science and Technology Innovation Board. If successful, it will also become the first AI chip stock on the Science and Technology Innovation Board.
The official website of the Beijing Securities Regulatory Commission announced on the evening of February 28 that on December 5, 2019, Cambricon Technologies and CITIC Securities signed an A-share listing guidance agreement, and will be listed on the Science and Technology Innovation Board next.
In June 2018, two years and three months after Cambrian was founded, its founder and CEO Chen Tianshi publicly stated that in the future, the company would consider listing on the domestic A-share market. Moreover, due to its name as a Chinese Academy of Sciences and a domestic product, it had not previously accepted financing from investment institutions with overseas backgrounds.
After the launch of the Science and Technology Innovation Board, Cambrian and other AI technology companies have attracted much attention. When interviewed by the media, Cambrian remained tight-lipped and did not disclose specific information.
Now, documents on the official website of the Beijing Securities Regulatory Commission have revealed that Cambrian’s listing on the Science and Technology Innovation Board has finally been finalized.
Four years after its establishment, Cambrian’s journey to chip manufacturing
Cambrian's journey into chip manufacturing officially began in 2016, and it has now established a chip matrix product covering end, edge, and cloud.
In March 2016, when the company was officially established, it launched its first Cambricon-1A processor.
It is known as the world's first commercial deep learning dedicated processor, and was selected as one of the fifteen "World's Leading Internet Scientific and Technological Achievements" at the Third World Internet Conference. It was also used in Huawei's Kirin 970 chip. In the following years, it was one of Huawei's important suppliers of Kirin.
The NPU used in Huawei mobile phones once made Cambricon famous. But later, as Huawei entered the field of AI chips and began to replace them with self-developed ones, Cambricon's role as a supplier also began to change, and the relationship became delicate.
In October, it launched an AI inference server test chip and launched Phaneron, the world's first AI inference dedicated server based on Cambrian chips, jointly developed with Sugon, equipped with Cambrian dedicated AI chips.
In November 2017, the company launched the 1H8 processor for low-power scene vision applications, and the 1H16 processor with wider versatility and higher performance, which were applied to Huawei's Kirin 980 chip. In the same month, Cambricon began to deploy in the autonomous driving industry and launched the Cambricon-1M processor.
In May 2018, the cloud AI chips MLU100 and MLU200 and the cloud server board equipped with MLU100 were released. MLU100 focuses on reasoning, while MLU200 focuses on training. The former is aimed at small and medium-sized servers, while the latter is aimed at enterprise-level artificial intelligence R&D centers.
At the 2019 World Artificial Intelligence Conference, Cambricon demonstrated for the first time its new generation cloud AI chip, the Siyuan 270, which integrates a series of Cambricon technologies in the field of processor architecture.
Cambrian revealed that its theoretical peak performance for processing non-sparse deep learning models has been improved to four times that of the previous generation MLU100, reaching 128TOPS (INT8); it is also compatible with INT4 and INT16 operations, with theoretical peaks of 256TOPS and 64TOPS respectively; it supports floating-point operations and mixed-precision operations.
In terms of specific architecture, Siyuan 270 adopts Cambrian's self-developed MLUv02 instruction set, which can support highly diverse artificial intelligence applications such as vision, voice, natural language processing, and traditional machine learning, and integrates video and image encoding and decoding hardware units for visual applications.
Cambrian said that in terms of peak performance, the data of Siyuan 270 is very close to NVIDIA T4.
In November 2019, Cambricon released the edge AI series of chips and modules, MLU220, to further cover end, edge and cloud scenarios and create systematic products.
Born in the Chinese Academy of Sciences, a member of the national AI chip team
Cambrian originated from the Chinese Academy of Sciences, and its origins can be traced back to a startup research group of only 7 people in 2009.
In March 2016, Chen Yunji and Chen Tianshi officially founded Beijing Cambrian Technologies, hoping that when the AI explosion comes, they can change the long-term blank and backward situation in China's chip field in terms of "deep learning" "neural network" processor chips.
The team members include not only technical elites from the Chinese Academy of Sciences, but also core participants in China's first general-purpose CPU, Loongson-1.
In terms of specific goals, they hope to increase the computing efficiency of AI chips by 10,000 times and reduce power consumption by 10,000 times, so that the most advanced AI applications such as "AlphaGo" can be installed on mobile phones.
However, due to the different personalities of the two brothers, they have a clear division of labor.
My elder brother Chen Yunji was born in 1983. He entered middle school at the age of 9, was admitted to the prestigious Junior Class of the University of Science and Technology of China at the age of 14, and entered the Institute of Computing Technology of the Chinese Academy of Sciences for a master's and doctoral degree at the age of 19. He became a member of the research and development team of the first domestically produced general-purpose CPU "Loongson" at the time, and became the chief architect of the 8-core Loongson 3 at the age of 25.
His position in the company is more research-oriented, thinking about parts related to technical paths. He rarely uses the Cambrian title and is mostly known as "Researcher at the Institute of Computing Technology, Chinese Academy of Sciences".
The younger brother, Chen Tianshi, was born in 1985. He was also a young genius and followed in some ways the path of his older brother, Chen Yunji. He was admitted to the junior class of USTC at the age of 16 and obtained a doctorate degree from the School of Computer Science at USTC at the age of 25. His supervisors were Chen Guoliang and Yao Xin.
After graduating with a doctorate in 2010, Chen Tianshi became a researcher at the Institute of Computing Technology of the Chinese Academy of Sciences. In 2014, he won the ASPLOS Best Paper Award and the MICRO Best Paper Award, becoming the first Chinese to win the Best Paper Award at these two top international conferences.
He is more extrovert and acts as the founder and CEO of Cambrian.
So far, Cambrian has completed a total of four rounds of financing.
In the initial angel round, the company was valued at US$100 million, with US$10 million invested by funds under the Chinese Academy of Sciences and Lenovo Capital.
In August 2016, half a year after its establishment, it received Pre-A round of financing from Yuanhe Yuandian, iFLYTEK, and Yonghua Investment.
In mid-2017, Cambricon completed its Series A round with a valuation of US$1 billion, led by SDIC Venture Capital, with participation from Alibaba, Lenovo Capital, CAS Capital, China Science & Technology Turing, Yuanhe Yuandian and Yonghua Investment, making Cambricon the world's first unicorn company in the field of AI chips.
In June 2018, Cambricon announced the completion of a round B financing of several hundred million US dollars, with a valuation of US$2.5 billion, or about RMB 17.48 billion. This is more than double the US$1 billion raised a year ago. Investors include the State-owned Capital Venture Capital Fund, China New TusPark, China Investment Capital, and China New Capital.
It was at this time that Chen Tianshi revealed the company's listing plans to the public and started the journey to list on the Science and Technology Innovation Board in December 2019.
Reference link:
http://www.csrc.gov.cn/pub/beijing/bjfdqyxx/bjfdqyjbqk/202002/t20200228_371257.htm
http://www.csrc.gov.cn/pub/beijing/bjfdqyxx/bjfdqyjbqk/202002/P020200228704383779830.doc
-over-
During the epidemic prevention and control period, are the children at home a little bored?
Recommended for children aged 7-12, a fun and interesting thing: learning programming at home. This is an excellent way to improve logical thinking, mathematical thinking, and computational thinking.
The Corgi Children's Introductory Programming Course is offering a limited-time discount for a total of seven lessons, now only 48 yuan, and the tuition fee will be fully refunded after completing the course!
Parents are welcome to scan the QR code to view and sign up:
Learn programming at home | Corgi Programming Dual Teacher Interactive Class
AI Insider | Pay attention to trends and seize opportunities
Internal Reference has been upgraded! Expand your network of high-quality contacts, get the latest AI information & paper tutorials, welcome to join the AI Internal Reference community to learn together~
Quantum Bit QbitAI · Toutiao signed author
Tracking new trends in AI technology and products
If you like it, click "Watching"!