Microchip, the United States tries to destroy the Chinese technology industry

Microchip

Last month, Chinese e-commerce giant Alibaba unveiled a powerful new cloud computing system designed for artificial intelligence (Ai) projects. The system will be used by Alibaba's cloud customers to train algorithms to perform tasks such as chatbot dialogue and video analytics, and has been built using hundreds of microchips from US companies Intel and Nvidia.

The last week the United States announced new export restrictions, which are bound to make projects like this difficult in the future. Measures imposed by the administration of US President Joe Biden prohibit American companies from exporting to China the advanced chips needed to train or operate the most powerful Ai algorithms.

The purpose of the restrictions is to relegate the China's AI industry backward, while the United States and other Western countries progress. The restrictions also block the export of chip manufacturing equipment and design software and prohibit the world's top silicon factories, including Taiwanese Tsmc and South Korean Samsung, from producing advanced microchips for Chinese companies. br>

Purpose and scope

“The United States is telling China: 'Ai technology is the future; we and our allies are going in that direction, but you can't come, "said Gregory Allen, director of the AI ​​governance project at the Center for Strategic & International Studies (CSIS), a Washington think tank.

According to Chris Miller, professor at Tufts University and author of the recent book Chip War: The Fight for the World's Most Critical Technology, the new export blockade is a measure not seen since the Cold War: "The logic is to put a spanner in the works [to China, ed] ", comments Miller.

The decision of the United States is part of the ten-year boom of artificial intelligence, a sector in which innovations they tie in with advances in computing power. New designs often involve machine learning algorithms trained on supercomputers with hundreds or thousands of graphics processing units (GPUs), chips originally designed for video games but ideal for performing math operations. China's AI ambitions, therefore, largely depend on US silicon.

Baidu, China's leading search engine and a key player in the country's AI cloud services and autonomous driving industry, is also heavily dependent on US silicon. makes extensive use of Nvidia chips in its data centers. Last October the company announced one of the world's largest AI models for language generation, built with Nvidia hardware.

ByteDance, the Chinese company that controls TikTok and its Chinese counterpart Douyin, relies on Nvidia hardware to train its recommendation algorithms, according to the company's documentation for its software. Several Chinese companies, including Alibaba and Baidu, are developing silicon chips designed to compete with those of Nvidia and AMD, but they need to be produced outside of China, an option currently not viable. Alibaba and Baidu declined to comment on the new US measures, while sportsgaming.win US received no responses to requests for comment addressed to ByteDance and other Chinese chip companies.

Tense weather

Large Chinese technology companies, such as US companies, have made large Ai models increasingly central in applications such as web search, product recommendation, language translation and analysis, image and video recognition, and autonomous driving. Advances in AI are estimated in the years to come to transform military technology and influence how the United States and China will clash over issues such as Russia's invasion of Ukraine and claims of independence. Taiwan.

"The Biden administration is convinced that the transformative potential of AI in military applications is real - underlines Allen of Csis -. But the United States also knows which chips are used in Ai's Chinese military systems : they are American, something that is considered unacceptable ".

The new export restrictions contribute to the progressive deterioration of relations between the United States and China in recent years, despite decades of technological codependency during which Chinese production is became the base of the US technology industry. In recent years, the US government has sought to take a more active role in strengthening the domestic AI and chip manufacturing industries, in the face of a growing spirit of competition with China.

Early Effects

Shares of several Chinese tech companies, as well as Nvidia and AMD, fell this week due to the effects of investor restrictions. Although the U.S. Commerce Department announced last month to Nvidia and AMD that they should stop Ai's advanced chip exports to China, the rules announced by the Biden administration last week have a much broader scope. The new restrictions on exports come after 18 tough months for Chinese technology companies, following the Chinese government's extensive crackdown on regulating the sector more tightly after years of uncontrolled growth.

Capacity of any Chinese company to keep pace with advances in chip manufacturing is limited by the lack of access to the extreme ultraviolet lithography machinery needed to produce chips with smaller components. The sole producer, the Dutch Asml, has blocked exports to China at the request of the US government.

The long-term scenarios

The US move, however, will not isolate the Chinese artificial intelligence industry immediately. A source working for a Chinese venture capital fund specializing in Ai, who spoke anonymously due to the sensitivity of the topic, claims that some Chinese companies have stocked up on components for GPUs after a few were disclosed in September. information about the new rules. Chinese companies could also train their own Ai models outside the country using equipment installed in other regions.

The chief executive of a Chinese Ai startup, who asked to remain anonymous, said that In the long run, the new restrictions will slow down Chinese companies' AI progress, but predicts that in the short term they will be able to keep up with the United States by extending the life of older hardware and building AI models that can do more in the face. of an unchanged computing power or by collecting more data. "If the goal is to achieve a certain accuracy, the amount of data can be more useful than the computing power - explains the CEO -. For most of Ai's activities, model training does not always require power. enormous ".

According to Douglas Fuller, an associate professor at the Copenhagen Business School and a scholar of the Chinese technology industry, the central point will be how the new US rules will be applied:" In the short term, I believe this measure will succeed in do what it sets out to do: bring China's high-performance computing efforts to its knees. " According to Fuller, however, China will turn its attention to other countries that have chip manufacturing experience and may try to get components into its smuggling territory.

This article originally appeared on sportsgaming.win US .