Wired: China wants to build powerful AI chips to challenge Nvidia

Source: Content from Wired, thank you.

According to Wired magazine, in July, China released a new strategic plan with the goal of catching up with the United States in artificial intelligence (AI) technology within three years and becoming a world leader by 2030. In October, the Ministry of Science and Technology of China released a call for scientific research projects online, which revealed many details about the plan. The plan targets Silicon Valley chipmaker Nvidia, a leading supplier of chips for machine self-learning projects.

The Ministry of Science and Technology document lists 13 technology transformation projects that are expected to be funded in the coming months and delivered in 2021. One of the projects is to develop new chips to run artificial neural networks.

To compete with Nvidia, China has set a standard for the project: the chip provided by China will be 20 times more efficient in both performance and energy saving than Nvidia's M40 chip, which is called a neural network "accelerator." Currently, Nvidia's M40 chip has only been developed for 2 years and is not its latest and most powerful chip, but it is still used in AI projects.

China has long planned to compete with Nvidia. In October, the National Development and Reform Commission put forward a requirement to develop high-performance AI chips when soliciting research plans. In August, an investment fund under the China Development Investment Corporation took the lead in investing $100 million (about 664 million yuan) in Beijing AI chip startup Cambricon. In early November, Cambricon announced the launch of two server chips that, if they live up to expectations, will replace Nvidia chips used in some AI projects.

Cambricon will be an important part of the rise of Chinese companies and AI chip startups. In the United States, similar startups including Google are already competing with Nvidia. In October, Beijing Horizon Robotics, founded by former Baidu veterans, raised $100 million (about 664 million yuan) in funding, and DeePhi Technology raised $40 million (about 265 million yuan). Veteran accessory maker Huawei is also working with Cambricon to jointly develop AI chips for mobile phones and other devices.

Chinese technology companies have a legitimate reason to compete with Nvidia, as it provides hardware for AI projects and has already established a large and lucrative market. And as more and more companies invest in AI, Nvidia's stock market value has increased 10 times in the past three years. It has begun to provide chips for robots, drones and self-driving cars, and has signed partnerships with Toyota and Volvo.

China faces many challenges in realizing its AI and hardware dreams. Although China has more computer science graduates and machine learning research papers than the United States, it still lags behind in the advanced expertise required for advanced AI projects.

China has been trying for years to compete with the United States, South Korea and Japan in the chip industry. To replace Intel and other American processors, China has launched chips for world-class supercomputers, but has yet to develop chips widely used in servers and personal computers. For example, Alibaba, China's largest cloud provider, still relies on Intel and Nvidia chips. Paul Triolo, who investigates Chinese technology and related policies at the Eurasia Group, said China has great ambitions, but it is still far behind the United States in designing chips and building factories to mass produce them.

Although China has shown its intention to compete in the AI ​​chip market, its AI chip companies have not yet launched an aggressive attack on Nvidia's core server market. Horizon Robotics, DeePhi Technology and Huawei have all focused on chip research and development to apply AI functions such as image analysis to devices such as cars and cameras.

Chris Rowen, an investor in a Silicon Valley chip design company, said that China's healthy market regulation is one of the driving factors of this trend. Putting an AI chip into a camera can enable it to automatically recognize people, objects or actions. Recently, Google also said that the AI ​​chip it uses can enable cameras to have image recognition capabilities, but this function is mainly used for shooting rather than monitoring. Rowen also said that China's chip start-ups are also paying attention to such huge market potential, such as applying AI functions to home appliances, auto parts and other Chinese-made equipment. And the best way to popularize this technology is to reduce costs.

Despite cross-border competition in AI, there are no clear battle lines between Chinese and American companies. Cambricon is licensing designs from Silicon Valley chip designer Arteris, for example, to move data on a chip to an interconnected “backbone.” Although Intel has its own products in the AI ​​market, it is investing $100 million in Horizon Robotics to develop AI chips for self-driving cars. In September, Nvidia CEO Jensen Huang announced new deals with Chinese internet giants Alibaba, Baidu and Tencent. Triole said that for some time, Chinese AI applications will run on American hardware.

Today is the 1467th issue of content shared by "Semiconductor Industry Observer" for you, welcome to follow.

Follow the WeChat public account Semiconductor Industry Observation , reply to the keyword in the background to get more content

Reply Popular Science , read more popular science articles about the semiconductor industry

Reply to DRAM , read more DRAM articles

Reply Lithography , read more articles related to lithography technology

Reply Samsung , read more articles related to Samsung

Reply Full screen , read more articles related to full screen

Reply Dual camera , read more articles about dual cameras on mobile phones

Reply Millimeter wave , see more articles related to millimeter wave

Reply to IPO to see more articles related to semiconductor company IPOs

Reply Exhibition , see "2017 Latest Semiconductor Exhibition and Conference Calendar"

Reply Submit your article and read "How to become a member of "Semiconductor Industry Observer""

Reply Search and you can easily find other articles that interest you!

Click to read the original text and join the Moore Elite