The world’s major superpowers, the US and China, are competing for AI dominance, trying to put the same puzzle together using different pieces. And U.S. leadership is not as clear-cut as many think.
“AI is expected to become a critical component of economic and military power in the near future,” according to Stanford University’s Artificial Intelligence Index Report 2023.
Shifts in AI development will be concentrated in certain places, changing the geopolitical landscape.
To limit China’s progress, the U.S. has imposed export controls on high-tech chips, limiting its ability to obtain advanced computing chips, develop and maintain supercomputers, and produce advanced semiconductors for military and other purposes. In response, China has imposed its own restrictions on rare minerals.
How do China and the U.S. compare in this race?
“After more than two decades of outsourcing, North America finds itself in a troubling position: it can develop the most advanced electronics but cannot manufacture them,” says a study by IPC, the global electronics manufacturers’ association.
Nor can China, which is the largest importer of semiconductors.
The recipe for AI superiority probably includes important ingredients such as a strong computing infrastructure, a research community and dedication.
The U.S. is ahead in chip development
Semiconductors that provide computing power are needed to train AI models and power other systems.
Several steps are required to create the final chip, and no country can accomplish all of these tasks on its own. The offshoring of design, manufacturing, packaging and final assembly spans the entire electronics ecosystem.
The U.S. clearly dominates the semiconductor design market with 85% of the global market, and 5% of the market is in Asian countries, according to the Department of Defense.
China, the largest importer and market for semiconductors, depends on U.S. chips. China accounts for 31.4% of global final sales, or $180 billion of the $574 billion in 2022, according to a Citi Global Insights report.
Recent U.S. bans on exports of high-tech components are a step backward for China. However, that doesn’t mean China is left without computing resources and can’t develop its own chips.
Beijing, partly reacting to the U.S. bans, is pursuing policies to catch up with global competitors.
Baidu, Tencent and other Chinese tech companies have announced that they have stockpiled enough artificial intelligence chips to mitigate the effects of the U.S. export ban while finding Chinese replacements. Baidu has long been developing its own Kunlun chips for AI computing.
China has even boasted of developing a photonic chip that is “3,000 times faster than one of the most common commercial AI chips, Nvidia’s A100.”
In August, many experts were caught off guard when Huawei released the Mate 60 Pro phone, powered by the Chinese-made Kirin 9000s 7-nanometer processor. While the 7-nanometer process isn’t cutting edge, it’s still highly advanced, much like Intel’s process in the US.
China manufactures the boards and solder
The silicon wafer is still not the end product. Asia dominates in wafer manufacturing, advanced packaging, PCB manufacturing, and final equipment assembly.
Companies such as Unimicron, Ibiden, SEMCO, Nan Ya and Shinko are involved in advanced IC wafer manufacturing. Here, the U.S. global share is 12%.
On November 20, the Biden-Harris administration announced $3 billion for the National Advanced Packing program “to ensure U.S. leadership in advanced packaging,” which is part of the CHIPS and Science Act.
Advanced packaging, part of semiconductor manufacturing, means tightly assembling multiple chips with different functions on a substrate with extremely precise dimensions.
Chinese media reacted with a comparison: the U.S. chip packaging production capacity is only 3% of the world’s, while China’s is estimated at 38%. It takes years to develop new capabilities.
When everything needs to be stitched together in the advanced packaging phase, 97% of this work is done in Asia. Taiwan again leads the way with nine companies, including ASE+SPIL and Powertech Technology.
“North America lags behind Asia in terms of capacity, but technical capabilities are also a concern,” the IPC study said.
China’s dominance in the commodity, multilayer and high-density interconnect board markets is more than 50% – it has significant flexible circuit manufacturing capacity for IC packaging.
China has the minerals
There are other ways China can retaliate by preventing U.S. companies from producing advanced electronics.
China has already imposed export restrictions on gallium and germanium. However, there are dozens of other critical minerals that the U.S. does not produce domestically, and China dominates their production.
In 2022, the U.S. is 100 percent dependent on imports of 15 non-fuel minerals. This list could be expanded to 51 minerals with dependence of more than 50%. China is the leading source of 26 critical commodities.
Artificial intelligence research: China is leading the way
Semiconductors, as difficult as they are to produce, are useless on their own. They have to drive software.
And China’s progress in AI research is staggering. China leads in the total number of publications in AI journals, conferences, and repositories, as shown in Stanford University’s Artificial Intelligence Index Report 2023.
In 2021, Chinese researchers will publish 40% of all global AI publications, surpassing the UK, Europe (15%) and the US (10%) combined.
In 2021, China produced the largest share of AI conference publications in the world at 26.2%.
Although the US still leads in the number of AI conferences and citation repositories, this leadership is gradually shrinking.
Nine of the top ten AI-focused publishing institutes are in China, and MIT ranks tenth.
U.S. big tech is producing more real AI-based products
Most meaningful AI-based systems are now produced by industry rather than academia.
U.S. companies and institutions have developed the majority of the world’s large language and multimodal models – 54% in 2022.
China has developed only three “significant machine learning systems” in 2022, while the U.S. has 16.
Standouts among U.S. models include OpenAI DALL-E 2 and GPT-3, as well as Google PaLM. The only Chinese large-language and multimodal model released in 2022 was the bilingual GLM-130B. In 2023, we’ll be hearing about new AI models every month.
It’s a similar story with private equity. In 2022, U.S. private investors will bring in $47.4 billion, about 3.5 times that of China. The U.S. leads the world in the total number of newly funded AI companies, with 3.4 times as many as China.
The only areas where Chinese investment in AI was greater than U.S. investment were semiconductors, industrial automation and networking, and sales support.
Six U.S. corporate giants dominate the AI market. However, Chinese companies such as Tencent, Alibaba and Huawei are stepping on their heels.
“However, Chinese companies are gaining momentum. In 2012 there was only one company in the top 10, but in 2021 there will be four. Tencent, Alibaba, and Huawei have surged ahead in terms of the number of AI papers they publish and the citations of those publications,” the Stanford University report noted.
Baidu even claimed its Ernie AI chatbot was better than OpenAI’s ChatGPT, though it couldn’t answer the question of what happened on June 4, 1989, or evaluate China’s “zero-covidian” policy.
The gap is narrowing in AI software development as well. In 2022, Chinese developers contributed 11.7 percent of high-impact changes to GitHub’s public AI projects, up from just 1.9 percent in 2011.
U.S. developers will contribute 20.3 percent to AI projects in 2022, up from 40 percent in 2011, statistics from the Organization for Economic Cooperation and Development (OECD) show.
The Chinese support AI more than Americans do
While the political will for AI development seems strong in both countries, the same cannot be said for the general population.
A majority of Chinese residents, 78% of respondents, agreed with the statement that AI-enabled products and services have more advantages than disadvantages, according to the IPSOS 2022 survey. This is the highest among all countries surveyed.
Only 35% of Americans surveyed agreed with the statement, one of the lowest among the countries surveyed.
Analyst: The U.S. has an advantage, but it’s waning
“The U.S. certainly has better capabilities because it already has the technology, equipment and people,” says Irina Zuckerman, a geopolitical analyst, national security lawyer and president of Scarab Rising, a consulting firm.
However, she believes it is only a matter of time before China becomes self-sufficient in high-tech. The main challenge for the U.S. is to stay ahead of the curve.
“The U.S. has not been very good at dealing with restrictions and sanctions in general, whether it applies to China or anyone else. I think there’s a good chance that sooner or later China will catch up with them, unless something changes and the U.S. and other countries become much more vigilant about their own technology,” Zuckerman said.
With the chip war, the challenge for the U.S. is the need to ramp up its own production to be less dependent on companies in Asia. However, the U.S. lacks the trained personnel and skills that have migrated to India or China. Even the attitude towards labor is a concern, as machinery must run non-stop 24 hours a day.
Although U.S. restrictions have slowed China’s progress, Zuckerman believes the country is making breakthroughs faster than expected.
“The ban is not complete. It’s a restriction on supplying some, but not all, chips and related products to China,” Zuckerman says. “The idea is to slow or prevent China from using this knowledge to get ahead of the U.S. in building military hardware and some computer equipment that relies heavily on these technologies.”
Despite the restrictions, chip-making equipment and advanced chips are making their way into China, sometimes even smuggled through third countries.
“Clearly, China has enough AI chips to mass produce these machines, export them for sale outside the country, and remain the dominant country in this area,” Zuckerman says. “In any case, the U.S. and all of its partners are more concerned about military production.”
She believes the restrictions have led to a slight slowdown in China’s military technology development. However, it is impossible to assess the effect because of low transparency.
“Nevertheless, it has not yet fully caught up with the United States,” Zuckerman says. “We know that it’s desperate to achieve superiority because it’s working hard to find a way to import chips or technology and produce it domestically. So obviously this is a very serious problem.”
While some chips could, in theory, be contained, the U.S. competitive advantage could deteriorate over time while China works on its own parts.
“Banning chips won’t affect the fact that there are people capable of creating these technologies,” Zuckerman says.
China also has the resources to hire scientists or even steal technology and knowledge from other countries. It already dominates commodity markets.
What it lacks is advanced knowledge of the current development process.” China has been quite successful in circumventing the restrictions of intelligence agencies and local companies, infiltrating the country with its scientists and stealing that knowledge base,” she says.
Despite the obstacles, Zuckerman doesn’t expect China to overtake the U.S. in technology unless “the U.S. is vigilant about its own industry, which has not been good so far.”
The bureaucratic communist structure makes it difficult for innovation and creativity to flourish.
“China had cultural and economic problems preventing it from becoming the number one power. Otherwise it wouldn’t have had to rely so heavily on exports and theft,” Zuckerman concludes. “He’s good at replication, capacity building and attracting foreign scientists. But when it comes to the domestic knowledge base, training and creating fully innovative industries, they are at a low ebb.”