The Future of Artificial Intelligence Depends on Asia
The coronavirus outbreak has driven a lot of attention to China and the greater northeast Asia region, and sparked discussions on its potential impact on global economies. In late February, global tech giants Apple, Google, and Microsoft announced plans to move manufacturing of new products from China to Southeast Asia.
This was not the only event that shook the technology market in the past year. In May 2019, the U.S.-China trade war peaked with U.S. President Donald Trump banning American companies from buying technologies or equipment from Huawei, the company being considered a threat to the national interests and security of the United States. The U.S. government also imposed tariffs on many Chinese goods including raw materials, excluding, however, rare earth metals the dependency rate on which goes as high as 80 percent.
Meanwhile Asia witnessed another drama unveiling between South Korea and Japan. In July 2019, Japan placed restrictions on exporting to Korea three elements crucial for production of semiconductors: hydrogen fluoride, fluorinated polyimide and photoresists. The official pretext for the decision lay in a long-standing dispute between Korea and Japan on the use of Korean labor by Japanese during the Second World War. The issue, still very sensitive in Korea, also caused other controversies like a boycott of the Japanese clothing brand Uniqlo over a promotional video, which unintentionally brought back wartime memories to some Koreans.
These three events — all evoked by different reasons and lying in different dimensions of political and economic relations — impact, among other important things, one single process: the production chain of semiconductors and microchips. Every crucial link of that process, which supplies the whole world with microchips for almost all smart electronic equipment, is located in the northeast Asian region: starting with raw material extraction and ending with the manufacturing of high value-added microchips to be installed in consumer devices.
The photoresists and hydrogen fluoride — the elements touched by the South Korea-Japan trade dispute — are made of Chinese raw materials, more precisely rare earth metals.
As of now, China holds a leading position both in deposits and in production of rare earth metals. When Deng Xiaoping visited the Inner Mongolian rare earth production area of Baotou in 1987, he made a shrewd prediction: “The Middle East has its oil, China has rare earths.” Despite the efforts by some countries, led by the United States in 2017, to increase domestic production of these materials, in 2019 China was still responsible for about 60 percent of global production, outperforming by about five to six times its competitors, such as United States, Australia and Myanmar. Thus, China remains the major supplier of these valuable materials for all the countries in the world, including, naturally, Japan which uses it for production of semiconductor chemicals. Back in 2010, a territorial dispute led to a cut of Chinese rare materials exports to Japan, but eventually it was resolved and the supplies were restored in full volume.
In turn, Japan uses China’s rare earths to produce the three chemicals mentioned above that are subsequently shipped to South Korea, Taiwan and back to China, the world’s largest manufacturers of semiconductors and microchips. According to the government reports, Japan produces about 90 percent of fluorinated polyimide, about 70 percent of hydrogen fluoride and 90 percent of photoresists. This makes Japan almost a full monopolist in this type of production, making it very difficult for its consumers to find substitutes or build up enough capabilities to avoid Japan entirely. Commenting on the current tensions between Japan and South Korea experts note that Japan developed the processing technology for decades and it will take several years to replicate it.
From Japan, the value chain moves to the world largest semiconductor and chipmakers – South Korean Samsung and Sk Hynix, which are both designers and manufacturers, and China and Taiwan, which host semiconductor foundries for global producers, such as NVidia or Qualcomm. Historically, semiconductor production is divided into two main stages, meaning that R&D and design is happening in one company — a fabless manufacturer — while physical assembling in another — a foundry. For instance, Apple, NVIDIA and Qualcomm rely on Taiwan Semiconductor Manufacturing Company to manufacture the graphics chips it designs. At the end of the day, it means that no matter where technology intensive processes are being performed the product is being physically fabricated in Asian.
Why is this so important? Semiconductors and their availability on the global market are tremendously important for the development of the cutting-edge technologies, such as artificial intelligence (AI), Blockchain and the Internet of Things (IoT). While artificial intelligence is commonly perceived as a complex software, the physical device which incorporates it is equally important. The new generation of hardware required by AI developers possesses several key characteristics including greater computational power and energy efficiency. According to a recent Deloitte report, AI chips will enjoy explosive growth. The AI chip market is expected to account for over 12 percent of the total AI market by 2022, with a compound annual growth rate of 54 percent.
Today, startups and SMEs contribute to global technology development almost as much as big corporations, which is equally true for AI developers. A recent CB Insights report noted 635 AI acquisitions since 2010, topped by Apple with 20 acquisitions. It is important to note that while corporations mainly use their own hardware, like Google which produced its own Tensor Processing Unit (TPU) not available in retail, startups rely on what is available on the market. Startups, therefore, remain very vulnerable to market shocks. Global access to AI hardware, microchips and semiconductors is crucial for global AI development.
However, while many countries adopted AI strategies in the recent years, not many pay sufficient attention to the development of their own AI hardware capabilities or at least ensuring access of the local engineers and IT specialist to the computational devices. In fact, among the big powers, China is the only one to acknowledge the need to research and develop intelligent computing chips and systems. Since the national governments are making their first steps in supporting AI developers it is clear that the strategies and tactics are to be adjusted according to the market needs and requirements and most likely will include more practical measures to ensure access to AI hardware.
The current situation is shaped by decades of technology developments, but seem quite fragile and easily trembled by myriad external factors. Some countries, pushed by these factors, have already started to take some steps to loosen their dependency on the supplies of this or that particular good, like the United States, which is planning new domestic rare earth mines, or South Korea seeking to diversify imports of chemicals necessary for the semiconductors production. These plans however will take a lot of time, while other did not even start.
All this means that the future of AI and other modern technologies on both the national and global scale will not only depend on the ability of IT specialist to develop new solutions but also on the stability in the northeast Asian region, at least in the short term.
Source: The Diplomat
