The Eagle and the Dragon in high-tech battle

So far, the US leads China 4 to 3 in the race for supremacy in the development of artificial intelligence

09:00 | 13 февруари 2023
Обновен: 16:29 | 13 февруари 2023
Снимка: Krisztian Bocsi/Bloomberg
Снимка: Krisztian Bocsi/Bloomberg

By Maryan Karagyozov

At the end of the last year, the US government imposed stifling sanctions not only on exports of US microchips to China, but also on exports of the software, equipment and technology needed for their production. The sanctions were unprecedentedly broad and are tantamount to a declaration of economic war in the high-tech sphere.

The blow is sensitive since each year Beijing buys more than $300 billion worth of foreign-made microchips, making computer chips China's biggest import item - far more than oil. The country imports 85% of the semiconductors needed for electronics manufacturing.

Signs that the US could use its dominant position in the sector have been piling for some time. In May 2019, the administration of then US President Donald Trump imposed export restrictions against the Chinese company Huawei. The policy did not change also under the new administration of Joe Biden, and in May 2020 the ban was supplemented by the introduction of a rule that all manufacturers of microchips developed or manufactured with American software or technology could not be supplied to the Chinese telecommunications giant without a special license. Three months later the US Commerce Department tightened its grip further by blocking all microchip sales to Huawei, and the unilateral sanctions have since expanded further to include dozens of other Chinese companies - including SMIC, the largest microchip maker in China. Meanwhile the Congress passed legislation known by the acronym Chip and Science Act, which requires the federal government to invest $52.7 billion in microchip research, development, and workforce training. In October 2022, the Biden administration dealt an even more serious blow to China's technology sector: Washington restricted all exports of advanced microchips and semiconductor manufacturing tools to China.

China did not stand idly by. On May 14, 2020 President Xi Jinping announced China's microchip strategy. China has already been the biggest supporter of the industry since the government announced in 2016 that it planned to spend $150 billion over 10 years to develop China's semiconductor industry. For this purpose the Celestial has two large national investment funds, as well as another 15 investment funds at the provincial or city level with a total capital of $25 billion. These funds are added to another $150 billion, which, according to the US Congressional Research Service, was invested in China's microchip industry from 2014 to 2020. Government support also includes government grants, low-interest loans and tax breaks, estimated at another $50 billion.

In accordance with the new strategy, Beijing has also set up a number of research centers whose work in 2022 was evaluated and the best 191 out of a total of 350 were selected. Despite the efforts of China to catch up in a number of cutting-edge chip technologies, it is a long , complex and expensive process. For example, it costs between $10 and $20 billion to establish just one factory for the production of integrated circuit wafers.

And the following image emerged. China has increased its share of the world market in microchip assembly, which is both less competitive and less strategic. In manufacturing, China is doing very well in terms of technology already developed, but in engineering, design and high-end technology, the US still has a dominant position and imposes export restrictions to maintain it.

Opinions are divided on whether US-imposed sanctions will be effective enough. Proponents of sanctions believe that the leading position of the US in the sector makes sanctions a powerful tool. However, some experts are more cautious, citing the sector's complexity. In an article aptly titled "Illusions of Control" published in the journal Foreign Affairs, Sarah Dantzman and Emily Kilcrease point out that the US has a leading position in the production of software for automating the electronic design of state-of-the-art chips. Together with the Netherlands and Japan, they control the supply of equipment needed to manufacture semiconductors, but the US is entirely dependent on Taiwan and South Korea for the production of state-of-the-art semiconductors. Washington has therefore reached out to these partners to ensure that they toe the line it has imposed on limiting China's access to cutting-edge technology.

The Taiwanese company TSMC (which controls about half of the world's production) plans to open a giant $12 billion chip factory in Arizona by 2024. It believes that this facility will likely only serve the American market, but the company's other plants will continue to work with Chinese contractors. Samsung is applying a similar spread of risks - the South Korean giant plans to build two chip plants in the coming years - one in Texas for $17 billion and another in central China for $15 billion.

Dantzman and Kilcrease point out that globally the US produces only 11% of the chips. The raw materials needed for the industry are concentrated in several countries, including China. Also, high fixed costs, technical complexity and high demands on workers in the industry encourage specialization and segmentation. To keep up, companies and countries must compete continuously and fiercely, and if they do not evolve, they may be overtaken by competitors. The forced exit of US chip tool companies from the Chinese market creates a natural niche and they may be replaced by other foreign companies that will have a strong incentive to exclude US technology from their own supply chains as a way to avoid the threat of further American control. Also, the revenue that US representatives in the sector could invest in innovation will decrease. A similar process was observed in the commercial space sector in the early years of the 21st century, after the US unilaterally imposed tighter controls on satellite exports to China. US market share and capabilities declined as European companies began offering products that excluded American components, including for the Chinese market.

For now, US restrictions apply to US individuals and entities operating in China, but do not apply to Chinese enterprises operating anywhere in the world. The significant Chinese diaspora in a number of countries in Asia, as well as a Chinese commercial presence around the world, greatly increases the challenges associated with preventing the import of chips into China.

Computers that draw pictures, produce scientific papers, and effortlessly defeat the best chess grandmasters are not science fiction, but fact. All this is possible thanks to the combination of huge data sets and the technological capabilities of machines to learn themselves - or in other words Artificial Intelligence (AI). The enormous possibilities it holds are expected to revolutionize the world in the coming years, just as computers and the Internet have changed it in the past three decades. Therefore, it is not surprising that the battle between the US and China for who will have the leading role in high technology extends to the field of AI.

The reasons why Beijing is seeking a leadership role in AI are complex. First, it is estimated that by 2030, up to 100 million Chinese will face employment change due to AI, making China the country most affected by this transformation. Second, the changes are likely to exacerbate already widening socio-economic inequality not only among individual workers but also between countries. Because of this, Beijing occupies the position of a global representative of developing countries. Third, AI is an important component of the Chinese military's strategy of asymmetric response to American military and political dominance. China's People's Liberation Army is seeking a kind of containment by achieving a decisive advantage in cutting-edge technology that would allow the Celestial to avoid direct military confrontation.

In his article from December 2019 for the US edition of the National Interest, Graham Allison, one of the doyens of US foreign policy research, warns that China is a full-spectrum peer competitor to the US. Among his main arguments are that the advantages of scale are definitely in China's favor - four times the size of the American population, greater opportunities to gather information, conditioned also by the country's culture, as well as determination and goal-setting at the national level.

In 2016, Chinese leader Xi Jinping announced that AI was a technology in which China should have a leading role, and specific goals were set to achieve by 2030. Five companies (Baidu, Alibaba, Tencent, iFlytek and SenseTime) were assigned a leading role. Down the chain, they were quick to implement the solutions, and just 12 months later, investment in Chinese AI startups exceeded that of the US, and in 2018, the Chinese created 2.5 times more patents in the field compared to the US. China is also investing heavily in the necessary hardware. In 2001, China did not have any of the five hundred fastest supercomputers in the world. Today it has 219, and the United States - 116. And while China's supercomputers used to rely on American semiconductors, today they are built with domestically produced processors.

The competition between the Eagle and the Dragon is also about human capital. On the one hand, China needs to catch up to the US's earlier start in the field. Half of the top AI professionals work for American companies, and the US can recruit from around the world, while China is limited almost entirely to its own population. English is also the world language of business and science. On the other hand, other factors also play a role - such as the fact that computer science is very difficult, which makes it less attractive in the Western world. Also in the US, finance, corporate law and other industries are so lucrative that they attract some of the most talented young people, taking away some of those who could pursue breakthroughs in high technology. Hard work and high math ability are characteristic of Asians, so it is not surprising that in recent years, China has annually graduated four times as many computer science students as the US (approximately 1.3 million vs. 300,000) and three times more computer professionals (about 185,000 vs. 65,000).

Regarding AI competition, former Google China CEO and tech investor and guru Kai-Fu Lee offers a useful analysis in his book AI Superpowers, Kai-Fu Lee. According to him, the first area for AI to enter our daily lives is the Internet-Artificial Intelligence. It's about the algorithms that offer us videos, products, etc. on Amazon, YouTube, etc. Both the US and China have digital champions - huge companies that develop these types of algorithms every hour.

The second wave comes with business AI such as decision-making algorithms in the management of financial portfolios and bank loans. In this area, the US has a major advantage, as it has a large volume of databases containing stock, banking, healthcare and other corporate transactions. At the time of writing Li's book (2018), the author gives a 90:10 advantage to the US, predicting over the next five years that their advantage over China will decrease to 70:30. For many Chinese, their mobile phone is their first computer. Many people do not have personal computers or laptops, so they make purchases and payments through their mobile phones, which is developing China's virtual financial services.

The third wave is perceptual AI, which includes voice and face recognition algorithms. There, the advantage is definitely in favor of Beijing. The Chinese Communist Party has granted China's four largest facial recognition firms access to its database of over 1.4 billion photos of citizens. According to American estimations, Chinese companies in the field have one million times more images than their American counterparts.

In terms of voice recognition and text decoding, the Chinese beat the American companies in all languages, including English. The top voice recognition startup in the world is China's iFlytek, with a user base of 700 million – ie. twice the 375 million who speak with Apple's Siri.

The fourth and final wave is autonomous AI. Here too, progress is at a fantastic speed, with machines that can color-select ripe fruit and collect them, drones are getting smaller, cheaper and have more and more functions, and leading companies are testing and producing driverless vehicles. According to Kai-Fu Li, the US is the leader in this field because of leading companies such as Google and Tesla. In other words, at the start of the race, the US has a tangible advantage, but China wants to catch up quickly. The Chinese government is highly motivated in developing AI-friendly policies, facilitating wider adoption of the technology. China is building a highway and a city the size of Chicago specifically designed for AI cars. It appears that here the forces are already almost equal and the fight will be particularly contested.

It is evident that both countries take the issue of their high-tech competition extremely seriously from the fact that in 2018 the US Congress created the National Security Commission on Artificial Intelligence (NSCAI) chaired by former Vice President of Google's Eric Schmidt and Vice Chairman Bob Work - Deputy Secretary of Defense in both the Barack Obama and Donald Trump administrations.

Although the US pioneered the creation of the two major technology platforms, Apple and Android, thanks to strategic targeting and state-business interaction, Chinese firms are catching up quickly. Currently, of the 7 giants in AI - Google, Amazon, Meta, Microsoft, Baidu, Alibaba and Tencent - 4 are American and 3 are Chinese companies.

All of this shows that achieving and maintaining global leadership in areas such as cloud computing or semiconductors requires massive and sustained public and private investment of financial and human capital.

THE BOTTOM LINE How long will sanctions help the US maintain its dominant position in the semiconductor manufacturing sector?