NVIDIA is seeking to establish a R&D center in Shanghai that will help the world's leading AI processor manufacturers stay competitive in China, with sales falling due to tightening U.S. export control measures.
CEO Jensen Huang discussed the plan with Shanghai Mayor Gong Zheng during a meeting in a Chinese city last month, according to two people familiar with the matter. NVIDIA is renting a new office space in Shanghai to accommodate existing employees as well as potential expansion.
According to those familiar with the program, the R&D center will look at the specific needs of Chinese customers and meet the complex technical requirements needed to contain Washington.
However, due to the legal sensitivity of transferring intellectual property rights to China, actual core design and production will remain overseas. "We did not send any GPU designs to China to modify to comply with export controls," NVIDIA said.
The Shanghai team will also engage in global R&D projects, including verification of chip design, optimization of existing products and industry focus research such as autonomous driving, people familiar with the matter said.
Huang also hopes to secure access to top AI talents based in China. Nvidia is currently in Shanghai advertising roles, which include engineers to help “guid the development of next-generation deep learning hardware and software” and “develop and optimize competitive ASIC designs around the world.”
One of the people said the Shanghai government expressed initial support for such a plan, while the NVIDIA was lobbying for the U.S. government approval. Silicon Valley companies have approximately 2,000 employees in Chinese cities, mainly sales and related support functions.
NVIDIA is expanding its research footprint in China as it tries to maintain its leadership in one of its largest overseas markets, where fears that local competitors led by Huawei can take over by providing competitors’ AI ecosystem.
Although China accounted for about 14% of NVIDIA revenue last year, Huang estimated that this could be a $50 billion market in several years.
"We want to establish American AI (US) standards around the world," Huang said last week at the Milken Institute event. "If we leave the market altogether, there is no doubt that others will step in. For example, Huawei is very strong... they will step in."
The Trump administration warned the U.S. and foreign companies around the world this week that Huawei-made AI chips could result in criminal penalties for violating U.S. export controls.
Huang flew to Beijing to meet with him at LIFEENEN's deputy prime minister on April 17, a watered model that has been redesigned to comply with Biden-ersa Controls restrictions on sales to China.
Given the limitations on China's best-selling chips, NVIDIA offers Chinese customers an alternative to L20 processors (low-end chips with no high bandwidth memory and less computing power).
Understand the tech giants hesitate because processors cannot compete with competitors' Chinese products for performance.
"We are in an awkward situation where we either choose a poorer NVIDIA chip, which (its software system) CUDA, which means lower operating costs or switch completely to Chinese chips and live a painful life by switching systems," said an executive at a leading Chinese technology company in China.
Orc-led customers, Alibaba and Tencent, are monitoring geopolitical developments to assess whether NVIDIA can offer redesigned high-end chips to meet their needs. People say that while NVIDIA is exploring various options, the final plan has not been completed due to legal uncertainty.