Pages

3 August 2023

Blocking China’s Access to AI Chips Matters to U.S. National Security

Gregory C. Allen

In November 2017, U.S. Air Force lieutenant general Jack Shanahan gave a keynote speech on Project Maven, a pathfinding effort to accelerate artificial intelligence (AI) adoption at the Department of Defense (DOD). In the speech, he said that Maven’s early results were so exciting that the DOD “should not buy another weapons system without AI.”

Where did General Shanahan deliver this speech? At the 2017 Nvidia GPU Technology conference in Washington, D.C.

As the DOD’s AI pathfinder, one of the first challenges that Maven encountered was that the DOD’s traditional computing infrastructure was completely unsuited for processing modern AI algorithms, which require astonishing amounts of specialized computing power. Project Maven had to build its own AI-ready infrastructure from scratch.

That meant buying lots and lots of specialized GPUs, also known as AI chips.

Project Maven’s budget eventually rose to more than $250 million per year, and a significant share of that budget went toward building out and maintaining the specialized AI computing infrastructure.

Since Nvidia was and still is the global leader in producing AI chips, Project Maven was a terrific customer for Nvidia almost immediately after its founding. It is not surprising that Nvidia invited General Shanahan to keynote their conference and explain the potential for AI to revolutionize future military technology and warfare.

Experts in China have also paid close attention to the strategic importance of AI chips. A 2018 report by Tsinghua University in Beijing put it in stark terms:

“Whether it is the realization of algorithms, the acquisition and a massive database, or the computing capability, the secret behind the rapid development of the AI industry lies in the one and only physical basis, that is, the chips. Therefore, it is no exaggeration to say, ‘No chip, no AI’ given the irreplaceable role of AI chip as the cornerstone for AI development and its strategic significance.”

Like their counterparts in the United States, China’s leaders, including Xi Jinping, general secretary of the Chinese Communist Party, believe that leadership in AI is foundational to the future of economic and military power. China’s 2017 national AI strategy describes AI as “a new focus of international competition.”

The Chinese military embarked on its AI modernization campaign at roughly the same time as the United States in 2017, and they encountered the same problem that Project Maven uncovered: a desperate need for AI chips.

But China’s military faced an additional problem: Chinese companies did not make AI chips, so, the Chinese government embarked upon a two-prong strategy. First, buy the AI chips from the United States (for now). Second, develop Chinese alternatives to the U.S. chips (as soon as possible).

The reality of Chinese military purchases of AI chips is not up for debate. It was openly published in unclassified Chinese military procurement contracts. Researchers from Georgetown University’s Center for Security and Emerging Technology reviewed 21,088 such contracts from April to November 2020 and found that all of the purchases of AI chips specified U.S. products. Not a single contract specified purchasing AI chips from Chinese companies. The same massive parallelization architecture that makes GPUs so attractive for AI applications also applies to other kinds of complex computational work, such as high-fidelity aerospace modeling and simulation for hypersonic missiles and nuclear weapons. A recent Wall Street Journal review of Chinese military procurement records since 2020 found that China’s top nuclear weapons research institute is also a repeat buyer of U.S. GPU chips. The U.S. intelligence community undoubtedly is aware of further—classified—Chinese purchases and plans.

However, the goal of producing Chinese AI chip competitors to U.S. products was already firmly established in the mind of China’s leadership—two years before Project Maven even began. In September 2015, China’s State Council published Made in China 2025, the cornerstone Chinese industrial policy. The major goal for semiconductors in Made in China 2025 was “replacement of imports with Chinese-made products basically achieved in key industries.”

China’s explicit policy goal—the one underlying its 2014 announcement of more than $100 billion in semiconductor industry subsidies—was to replace U.S. and allied chips and chipmaking technologies with Chinese alternatives. This was China’s plan not only before President Biden took office, but two years before former president Donald Trump was inaugurated.

Of course, China doubled down on the plan in the wake of the Trump-era trade war and sanctions on Chinese tech firms such as ZTE and Huawei. By the time President Biden took office in January 2021, it was a Chinese national security priority.

This is the vital context through which to view the Biden administration’s October 7 export controls, which restricted U.S. exports of advanced AI chips and the advanced equipment used to make them. China’s leaders had already decided to do anything and everything to eliminate their dependence on U.S. technology. There was no realistic chance of persuading China to walk away from that policy.

Thus, the October 7 export controls are focused on ensuring that both prongs of China’s strategy fail.

To block China from buying advanced U.S. chips, the export controls apply on a country-wide basis, not just for known Chinese military organizations. That is a direct response to the fact that China’s strategy of military-civil fusion has worked to deepen and obscure the linkages between Chinese commercial companies and the Chinese military. Even though U.S. companies have reliably complied with pre-2022 “no military end user” export control restrictions, resellers networks within China have reliably been able to get the chips to the military. Blocking sales to China as a whole was overdue.

To block China from making Chinese alternatives to U.S. AI chips, the export controls continue to allow sales of chips below certain technological performance thresholds and also restrict the sale of advanced chip-making equipment. Chinese AI chip design firms such as Biren and Cambricon were nipping at the heels of U.S. companies such as Nvidia prior to the October 7 export controls. They were perhaps only a year behind Nvidia in terms of chip design quality. But now, those Chinese chip design companies are cut off from accessing the advanced American semiconductor manufacturing equipment (including the equipment powering chipmaking factories in Taiwan).

Chinese semiconductor equipment firms are more than a decade and a half behind the U.S. state of the art. And China is not just cut off from advanced U.S. equipment. In March 2023, the Dutch and Japanese governments announced that they were also adopting new export controls on advanced semiconductor manufacturing equipment. Combined, the United States, Japan, and the Netherlands are responsible for nearly 90 percent of global sales of such equipment.

This reflects the simple fact that more and more countries realize that China means what it wrote in Made in China 2025: China’s strategy for technological trade is to build a future that does not include non-Chinese companies. In 2019, China fired a top official for criticizing Made in China 2025.

In building that future, the October 7 export controls ensure that Chinese firms face an extraordinarily steep mountain to climb. In low-performance technologies, Chinese firms continue to face foreign competition. And to build high-performance technologies, China effectively must replicate the entire semiconductor manufacturing equipment value chain before Chinese firms can make their first chip. Each one of these types of machines—of which there are nearly a dozen—is nearly as complicated as the James Webb Space Telescope. China could spend a trillion dollars and still fail.

Some critics who are not familiar with the particulars argue that the Biden administration’s policy is one of “decoupling.” This is simply wrong. The policy targets advanced versions of niche products within the semiconductor industry. In reality, China’s policy—the one that it has been spending hundreds of billions of dollars on since 2015—is much closer to a decoupling strategy than the Biden administration’s policy.

The AI sector has seen breakthrough after breakthrough since General Shanahan’s 2017 speech. In May 2023, a group of AI industry and academic leaders issued a statement warning that the risks of advanced AI should be viewed in the same way as pandemics and nuclear war. None of those risks will be any easier to manage if China achieves its vision of becoming an AI-enabled authoritarian superpower.

In blocking China’s access to advanced AI chips, the Biden administration is advancing U.S. national security.

After all, no chip, no AI.

Gregory C. Allen is director of the Wadhwani Center for AI and Advanced Technologies and senior fellow with the Strategic Technologies Program at the Center for Strategic and International Studies in Washington, D.C.

No comments:

Post a Comment