Unlock Editor’s Digest Lock for Free
FT editor Roula Khalaf will select your favorite stories in this weekly newsletter.
Nvidia is looking to build a research and development center in Shanghai, helping the world’s leading manufacturer of artificial intelligence processors stay competitive in China.
According to two people with knowledge of the issue, Jensan Hwang discussed the plan with Mayor Gong Zen, the mayor of Shanghai, when he met in a Chinese city last month. Nvidia is leasing new office space in Shanghai to accommodate existing employees and potential expansion.
According to people with knowledge of the planning, the R&D center will investigate the specific requirements of Chinese customers and the complex technical requirements needed to meet the curbs in Washington.
However, actual core design and production remain overseas due to legal sensitivity to transferring intellectual property to China. Nvidia said: “We have not sent GPU designs to China to be changed to comply with export controls.”
The Shanghai team will also work on global R&D projects, including sector focus research such as chip design verification, optimization of existing products, and autonomous driving, people with knowledge of the issue said.
Huang also wants to ensure access to top Chinese-based artificial intelligence talents. Nvidia is currently advertising its role based in Shanghai. This includes engineers to “lead the development of next-generation deep learning hardware and software,” and “developing and optimizing competitive ASIC designs on a global scale.”
While the Shanghai government had shown preliminary support for such a plan, Nvidia had lobbyed from the US administration for approval, one of the people said. The Silicon Valley Company has around 2,000 employees in Chinese cities, but mainly has sales and related support features.
As Nvidia seeks to maintain its leadership position in one of its largest overseas markets, it is expanding its research footprint in China, fearing that local competitors led by Huawei can take over by providing a rival AI ecosystem.
China accounted for around 14% of Nvidia’s revenue last year, but Huang estimates it could become a $50 billion market in a few years.
Recommended
“We want to build on the global AI (here) that American standards are being adopted all over the world,” Huang said last week at an event at the Milken Institute. “If we leave the market for good, there’s no doubt someone else will step in. Huawei, for example, is very formidable. They will step in.”
This week, the Trump administration issued a warning to American and foreign companies around the world that the use of AI chips manufactured by Huawei could cause criminal penalties for violating US export controls.
Fan flew to Beijing to meet China’s vice president on April 17 to meet his Lifeeng, the Chinese vice president, and a watering model that had already been redesigned to comply with Biden-era controls, restricting sales to China just days after new export restrictions were imposed on Nvidia’s H20 chips.
Given the limitations of bestselling chips in China, NVIDIA offers its L20 processors (without high bandwidth memory or low computing power) to its Chinese clients, according to people with knowledge of the issue.
The high-tech giant is understood to be hesitant to order as processors cannot compete with rival Chinese products in performance.
“We’re in a troublesome situation where we choose a worse nvidia chip than it runs on (its software system) CUDA. This means reducing operational costs or switching completely to a Chinese chip and get through the pain of the switching system.”
The client led by bytedance is evaluating whether Alibaba and Tencent can oversee geopolitical developments to provide Nvidia with redesigned high-end chips to meet their needs. Nvidia was exploring a variety of options, but there was no final plan due to legal uncertainty, people said.
