AI has long been high on Beijing’s priority list, and it is considered as one of the core drivers for high-quality economic development, according to China’s 2021 to 2025 development guidelines.
Speaking at a Politburo meeting in October 2021, China’s President Xi Jinping pledged to “fight the battle for key core technologies” and achieve a high level of self-reliance.
Xi particularly mentioned the use of a new whole-nation mechanism, which would see Beijing rally every possible resource for its development, similar to what it did to develop its satellite, nuclear weapons and space programme in previous decades.
Since Beijing approved a mega data transfer plan last year to move user data in the east of the country to the western region with its abundance of energy and vacant fields via eight national computing hubs, more than 400 billion yuan (US$56 billion) has been poured into the plan.
Western countries, on the other hand, tend to count more on private entrepreneurs, and some government incentives to support tech investment.
In order to gain the upper hand, Beijing has embarked on an all-out effort to harness the potential of generative AI, with an emphasis on new infrastructure, to propel its computing power and close the technology gap with the US.
China, though, still lags behind in a variety of technical dimensions, as the US has 3.5 times more private investments, half of the world’s language models and the highest repository citation count, according to a study by Stanford University.
“The United States is still ahead in terms of AI conference and repository citations, but those leads are slowly eroding. Still, the majority of the world’s large language and multimodal models (54 per cent in 2022) are produced by American institutions,” said the report.
But China is quickly catching up by producing the most AI journals, conferences, publications and leading industrial robot installations, surpassing the total in all other countries combined.
Despite the escalating tech rivalry, both countries are engaged in the highest volume of AI research collaborations, which have quadrupled since 2010. But the pace of cooperation slowed after 2020, resulting in a modest 2.1 per cent increase by the end of 2021 compared to a year earlier.
Amid Washington’s ongoing tech sanctions, Beijing is seen to have both the resources and determination to scale up its computing capabilities with immense technological infrastructure investments to help narrow the gap.
Computing power is seen to be becoming increasingly crucial in AI, helping it to rapidly process mountainous amounts of information, revolutionising the speed and accuracy of system analysis.
“The rivalry between the United States and China may well be determined by computing power,” said historian Chris Miller, the author of Chip War: The Fight For The World’s Most Critical Technology.
China accounts for 33 per cent of the world’s computing power, which is only 1 percentage point lower than the US, according to the China Academy of Information and Communications Technology (CAICT), an affiliate of the Ministry of Industry and Information Technology.
“China’s focus on increasing its computing power certainly provides an opportunity for China to catch up to the US in AI,” said Nestor Maslej, a research manager at Stanford University’s Institute for Human-Centered Artificial Intelligence.
But although the gap might narrow, China is still far behind the US in terms of AI technology due to its superior investment environment for AI companies, higher quality research and dominant AI model.
Last year, the US outproduced China in producing AI machine learning systems by more than fivefold, creating 255 new significant systems in comparison to China’s 44, according to Maslej.
Boosting the level of computing power is likely to help Beijing take a shortcut to AI development, as the release of fundamental models is viewed as a flawed factor limiting the country’s AI development.
The advancement of fundamental models requires large amounts of data, computing infrastructure and funding, resources that are inherently available to a greater extent to industry actors than those in academia.
“China increasing its country’s computing power will likely give a larger subset of actors, like universities and even industry players, the opportunity to train these increasingly important foundation models,” Maslej added.
According to the CAICT, for every one yuan invested in computing power in China, it will drive three to four yuan of economic output.
Chips are playing an important role in the increase of computing power in China, with the share of graphics processing unit chip computing power in the computing field increasing from 3 per cent in 2016 to 41 per cent in 2020, according to CAICT data.
But Washington’s choke point on China’s chips is similarly besieging the development of its computing industry, as the performance of chips directly determines the efficiency and performance of computer systems and devices.
“China’s domestic AI industry is now generally short of computing chips, and if the US further sanctions China’s chip technology, it will definitely affect computing power development in the short term,” said Li Yangwei, a technical consultant working in the smart computing industry in Shenzhen.
As Chinese chip technology gradually becomes more autonomous, the impediment to the development of domestic computing power caused by chip shortages may be eased, Li said.
“I would estimate the impact of the chip shortage on computing power development to be mitigated in two to three years at the earliest and five years at the slowest,” Li said.
The construction of data centres requires a large number of high-performance computing, storage and server chips, which is undoubtedly the challenge China encounters in dealing with massive amounts of information.
“China’s computing power improvement faces multidimensional challenges, and there is still a huge gap in the transformation from chips to computing power,” CAICT said in its white paper on computing power development released in November.
The government think-tank suggested that China should take advantage of its “mega-market and new national system” to enhance its computing power, while also promoting chips and other developments.
US tech decoupling, along with the fallout from the coronavirus pandemic and its ageing population, is seen to be one of the new factors holding back China’s potential growth.
According to Zhang Xiaojing, head of the Chinese Academy of Social Science’s Institute of Finance and Banking, US tech decoupling will lower potential growth by 0.3 percentage point each year over the next three years and 0.5 percentage points each in 2026 to 2040.
“Those three new factors are going to postpone China’s overtaking the US to the year of 2033 from the baseline estimate of 2031,” Zhang wrote in an article published in April.
Despite the fierce competition between China and the US, many politicians have warned about the challenges brought by AI and have called for cooperation.
Former US secretary of state Henry Kissinger said in an interview with The Economist in April that the fate of humanity depends on whether the US and China can get along, while the rapid progress of AI leaves only five-to-10 years to find a way.
“AI is not a contest of two countries,” Kissinger told a close-door meeting hosted by JPMorgan in Shanghai in late May, adding that AI is a new era of human consciousness that requires collaboration between the US and China to understand its potential and risks.
Despite this, an AI investment frenzy has now unfolded across China.
The Shenzhen government announced in early June that it would set up a 100 billion yuan AI investment fund to vigorously develop the city’s computing power and turn it into a pioneering AI zone in China.
The tech hub also plans to accelerate the use of AI in undergrounds, airports and hospitals, aiming to serve as a computing platform for the Greater Bay Area.
The Greater Bay Area refers to the Chinese government’s scheme to link the cities of Hong Kong, Macau, Guangzhou, Shenzhen, Zhuhai, Foshan, Zhongshan, Dongguan, Huizhou, Jiangmen and Zhaoqing into an integrated economic and business hub.
Shanghai has also spent over 250 billion yuan on new infrastructure over the past three years, with private capital accounting for more than 30 per cent.
The increase in spending has encouraged many China forecasters to remain upbeat about the country’s development path.
And Liang Haoguang, executive director of the China Centre for Modernisation Research under the Chinese Academy of Sciences, is convinced that China’s economic growth will be increasingly driven by technology.
“Judging from a tech innovation perspective, the Chinese economy is likely to catch up with the US size in 2028,” he told a seminar in Beijing last month.
This article was first published on SCMP.
Source: CNA
