China is closing in on US technology lead despite constraints, AI researchers say
By Laurie Chen
BEIJING, Jan 10 (Reuters) – China can narrow its technological gap with the United States through increased risk-taking and innovation, although a lack of advanced chip-making tools is hobbling the industry, the country’s leading artificial intelligence researchers said on Saturday.
China’s so-called “AI tiger” startups MiniMax and Zhipu AI made a strong debut on the Hong Kong Stock Exchange this week; This reflects growing confidence in the sector as Beijing accelerates AI and chip listings to support domestic alternatives to advanced US technology.
Yao Shunyu, former senior researcher of ChatGPT maker OpenAI A Chinese firm is likely to become the world’s leading AI company in the next three to five years, but a lack of advanced chip-making machinery is the main technical hurdle, said the chief AI scientist at tech giant Tencent in December.
“We currently have a significant advantage in electricity and infrastructure. The main bottlenecks are manufacturing capacity and the software ecosystem, including lithography machines,” Yao said at an artificial intelligence conference in Beijing. he said.
China has completed a working prototype of an extreme ultraviolet lithography machine that has the potential to produce cutting-edge semiconductor chips that could rival those of the West, Reuters reported last month. However, the machine has not yet produced working chips and may not produce them until 2030, people with knowledge of the matter told Reuters.
BEWARE OF THE INVESTMENT GAP
At the conference in Beijing on Saturday, Yao and other Chinese industry leaders also acknowledged that the United States has an advantage in computing power because of its massive investments in infrastructure.
“The US computing infrastructure is probably one to two times larger than ours. But I see that they are investing heavily in next-generation research, whether it’s OpenAI or other platforms,” said Lin Junyang, technical lead on Alibaba’s flagship Qwen large language model.
“We, on the other hand, are relatively cash-strapped; delivery alone probably consumes most of our computing infrastructure,” Lin said during a panel discussion at the AGI-Next Frontier Summit organized by the Beijing Fundamental Models Laboratory of Tsinghua University.
China’s limited resources encourage its researchers to be innovative, especially through algorithm-hardware co-design, which allows AI firms to run large models on smaller, cheaper hardware, Lin said.
Tang Jie, founder of Zhipu AI, which raised HK$4.35 billion in its IPO, also highlighted the willingness of young Chinese AI entrepreneurs to embrace high-risk ventures (a trait traditionally associated with Silicon Valley) as a positive development.
“I think if we can improve that environment by allowing more time for these risk-taking, smart individuals to participate in innovative efforts, that’s something our government and our country can help improve,” Tang said.
(Reporting by Laurie Chen; Editing by Emelia Sithole-Matarise)
