Careful Connectivity: Responding to China’s AV Drive
October 15, 2021
The Biden administration has not clearly decided how extensive it wants the U.S.-China economic relationship to be. Its June 2021 study on supply chains identified risks from overdependence on China, but did not lay out a clear solution. During her recent speech at CSIS, U.S. Trade Representative Katherine Tai said that she opposed “decoupling” from China and instead preferred a “recoupling” of the economies that was fairer and more in line with U.S. interests. Commerce Secretary Gina Raimondo has also pushed back against calls for decoupling, yet said that an important goal should be to slow Chinese innovation, which implies less connectivity, particularly in advanced technology sectors. Their comments highlight a critical question the United States and its allies must answer: How can they smartly chart a middle path between outright decoupling and unconditional engagement?
For an initial answer to this question, the autonomous vehicle (AV) sector might offer some guidance. It is an excellent critical case to consider the larger issues because of its multifaceted significance. The AV sector portends a revolution in transportation and lifestyles; it is composed of an array of technologies that draw on extensive research and development (R&D) and require advanced manufacturing abilities; the collection, analysis, and storage of its data present challenges to privacy and national security; and it utilizes some dual-use technologies that could aid U.S. adversaries if they fell into the wrong hands.
Beyond the general risks are recent warnings from successes in other parts of China’s automobile industry. China has far and away the world’s largest electric vehicle (EV) market. In the first eight months of 2021, it sold 1.7 million EVs, accounting for almost 11 percent of all auto sales. By contrast, even with a recent surge in EV sales in the United States, the total market in 2020 was slightly under 300,000 vehicles, accounting for 8.7 percent of total sales. China’s CATL is now the world’s largest EV battery producer, and the country’s auto firms are starting to export to Europe and Asia and invest abroad as well. As documented in a 2020 CSIS report, Chinese state support for the EV sector is mammoth, meaning that China’s state-fueled EV industry could distort every corner of the global industry, threatening competitors in the United States and elsewhere.
It is no huge leap to think that some of the same dynamics could play out in AV, and if so, it might make sense for the United States and others to take preemptive action and start mitigating risks now by severing the Chinese and U.S connections in the sector. If there is any industry where the United States should consider decoupling from China, AV would be a top candidate.
But that would be a huge mistake, as it actually would make the United States worse off—not just its economy, but potentially its national security as well. Instead, an approach of careful connectivity—that extends China’s dependence on the West while mitigating the worst risks for the United States—is the best way to move forward.
The Chinese Challenge
China’s AV sector has been advancing rapidly and doing so in a way that is certainly discriminatory, as it is shaped by China’s desire to favor domestic producers and reduce its dependence on foreign technology. Although the large majority of Chinese industry participants are private firms, the heavy hand of the state is everywhere.
China issued broad plans for intelligent and connected vehicles in 2017 and 2018. Since then it has provided substantial state support via subsidies and government guidance funds for companies to carry out R&D, testing, and manufacturing and provide related services, such as robo-taxis. Although there have been severe restrictions on road testing due to safety concerns, regulations in early 2021 expanded approval for such activity. According to the Ministry of Industry and Information Technology (MIIT), testing is permitted in 27 provinces, including 16 test demonstration zones, covering over 3,500 kilometers of road. So far over 700 entities have received licenses for testing, and they have collectively driven over seven million kilometers. Also in early 2021 China issued a blueprint for technology standards for the sector, setting a goal of having major standards in place by 2025 and a complete set of standards by 2035. And most recently, in August, the Cyber Administration of China issued new data security draft rules.
Having received the positive signals that AV is a high-priority sector, in addition to pure AV firms (such as Pony.ai), others have also jumped in: traditional automakers (SAIC), new upstarts (NIO and Byton), internet firms (Baidu, Alibaba, and Tencent), driving services companies (Didi), telecom hardware makers (Huawei), and even smartphone makers (Xiaomi). The Baidu-led Apollo coalition originally dominated, but a slew of other consortia and individual firms are investing in every layer of the industry, including advanced components (such as semiconductors and LiDAR), entire vehicles, infrastructure, and services.
There is far from a level playing field in China. Foreign companies face a phalanx of restrictions with regard to conducting R&D; testing, collecting and using mapping data; and offering value-added services. They also have to worry about having their intellectual property (IP) stolen and their employees poached by local competitors. Meanwhile, Chinese companies who can obtain licenses are able to test their autonomous vehicles on U.S. roads, where they scoop up data on every mile they cover. Chinese AV companies have attracted substantial U.S. venture capital. Pony.ai and AV trucking firm TuSimple are two of a number of Chinese companies refining their technology in collaboration with U.S. tech firms and investors.
The Risks Are Too Great . . . to Decouple
Given China’s ambitious state-led push, its clear effort to build a free-standing AV sector, and the potential risks from Chinese AVs collecting data in the United States, the decoupling option would seem to make a lot of sense. The United States could outright order U.S. companies, as well as those from elsewhere who use U.S. equipment in their own production, to stop selling their components and AV-enabled cars in China, force Chinese companies off U.S. roads, and require U.S. investors to divest from Chinese firms. One day this option may make sense, but based on where things stand now and the likely trajectory of the industry, doing so would likely have a host of negative consequences for the U.S. auto industry, including the various players in AV, and potentially spur China’s technological independence down the road.
Instead, the United States and its allies should use their place in the AV ecosystem to maintain their dominance over Chinese competitors, while simultaneously reducing the risks that come from being interconnected. Why? Because the United States is far ahead of China because of, not in spite of being connected.
Chinese AV companies have improved gradually over the past five years, but they are no match for their Western counterparts. The best place to compare them head-to-head is how their cars operate on the road. No one has successfully rolled out genuine autonomy where occupants can take their hands off the wheel and doze off into space (what is known as Level 4 automation) or where there is no steering wheel at all (Level 5). But on account of their more advanced algorithms, greater testing and data collection, and other technological progress, U.S. firms—among them Waymo, Cruise, and Tesla—have created far more advanced drive-assistance systems than their Chinese counterparts.
California publishes highly detailed data from AV companies who test on their roads. In 2020 Baidu had 4 vehicles on the road compared to 228 for Cruise and 148 for Waymo. U.S.-based AVs travel many more miles than their typical Chinese competitors without any need for human intervention. In 2020 Waymo testing vehicles required interventions once every 29,900 miles and Cruise every 28,500 miles. They were far ahead of Chinese rivals Pony.ai (10,700 miles) and WeRide (6,500 miles). The Chinese have improved over past years but are still far behind. Chinese AV companies occasionally publish videos of their testing in China, which provide a visual take of their abilities. Arcfox, which has developed an AV system in cooperation with Huawei, recently shared a video in which the driver had to intervene multiple times within a few minutes, in part because Chinese roads are so unpredictable. In China, “edge” cases—unusual circumstances that should occur infrequently—are all too common.
But the rubber really meets the road in the underlying technology, where the United States and its allies are even further ahead. Perhaps the most important component of AV systems is highly advanced artificial intelligence chips that collect and process data about the vehicle’s surrounding environment. U.S. chip firm Nvidia dominates the market for graphic processing unit (GPU) chips that perform these functions. Nvidia is a supplier to just about every AV maker in the world, including in China. There are a few Chinese upstarts in this space, such as Iluvatar CoreX and Biren Technology, but they are still significantly behind Nvidia in terms of basic technology and the breadth of the ecosystem they can support. And more importantly, their chips are built on the foundation of Nvidia’s CUDA architecture, meaning that it would be extremely difficult for them to elevate themselves to the top of the technology hierarchy. CUDA has a similar dominant status as Qualcomm’s CDMA (code division multiple access) technology for mobile telephony, Microsoft’s Windows in desktop computing, and Google’s Android in smartphones.
Technological decoupling would temporarily set back China’s AV dreams, but it would also set the nation on a course to eventually build an alternative architecture, one it could not only sell at home but disseminate among friendly countries elsewhere around the world. While the United States is ahead and China operates in a state of dependence, it makes little sense to push the Chinese to pursue a totally different, independent path.
Discontinuing those commercial relationships would also quickly mean a reduction in sales for Nvidia and other Western AV producers that have a massive customer base in China. It could also result in car makers such as GM, Ford, and Tesla facing greater restrictions on the full range of their business in China, far and away the world’s largest auto market. Reduced sales means less profits and funds for R&D as well as likely cutbacks in employees across their production facilities, including in the United States.
In short, outright decoupling would be a path to a smaller, less dominant U.S. AV sector, the exact opposite outcome U.S. policymakers ought to be pursuing.
A Smarter Approach
A better strategy would combine three elements: promotion, protection, and standard building.
Federal agencies and state governments need to facilitate the continued development of the AV sector, providing support for R&D, manufacturing, infrastructure, and expanding consumer demand (with rebates and other incentives). Making progress on rolling out fifth-generation (5G) cellular technology will also support the AV industry and ancillary services. Much work needs to be done so that 5G networks are ubiquitous, stable, and secure. Although the frequency of accidents is low, it needs to be even lower to ensure the safety of passengers and pedestrians, and the insurance industry and regulators need to build systems that appropriately allocate responsibility and protect consumers. Finally, regulations to ensure protection of the data collected about the cars, their passengers, and their surroundings need to be instituted. The sector will only grow if consumers believe their interests are first and foremost in the minds of companies and regulators.
At the same time, the United States needs to mitigate risks from being part of a global industry in which Chinese firms are advancing and the Chinese government is expanding its control over companies at home and acting aggressively abroad. The United States needs to have a fuller grasp of U.S. industry’s cooperation with Chinese counterparts in China. The best way to do that is through more regular consultations between U.S. industry and government about their overseas investments and supply chain challenges; that would be a better route than extending the mandate of the Committee on Foreign Investment in the United States (CFIUS) to include outward U.S. investment, which would likely end up being overly expansive and eliminate valuable commercial activities that pose little risk to the United States.
The main way to restrict potentially worrisome technology flows is through more vigorous use of export controls. The Commerce Department may need to add certain kinds of advanced semiconductor technology to its Commerce Control List and place more Chinese entities on its Entity List. Doing so may result in restrictions on sharing certain technologies, but an equally important utility would be to approve sales but then have a means to better monitor their sales and the behavior of licensed end users. There is a low likelihood that Chinese-related AV companies are collecting sensitive data during normal testing while on U.S. roads, but that risk should be more fully evaluated, followed by the adoption of rules regarding data collection, storage, and transfer. A substantial portion of Nvidia’s AV-related chips are fabricated by TSMC in Taiwan; it may make sense to prioritize the diversification of their manufacturing given the downward trajectory of PRC-Taiwan relations and the growing possibility of a conflict that could result in major shortages and loss of technological leadership. And lastly, long before China starts selling AVs in the United States, regulators need to consider the effects on U.S. producers and those from other market economies as well as the potential risks to personal data.
The final element in a smart strategy involves expanded efforts by the United States and its allies to set international standards for AV. In addition to the process to promote sustainable mobility for AVs occurring through the United Nations Economic Commission for Europe (UNECE), top priority should be placed on international venues such as the International Organization for Standardization (ISO), the International Telecommunication Union (ITU), 3GPP, SAE International, and the Autonomous Vehicle Computing Consortium. Moreover, Washington should consult with its allies in Europe (through the U.S.-EU Trade and Technology Council) and Asia on the whole range of AV issues, including on standards. Finally, the United States and others should continue to jointly press China for access and fair treatment in China’s domestic standards bodies. Very simply, standards leadership is central to industry leadership.
If the United States adopts this group of policies, its AV sector will stay on the right road and maintain its position ahead of China. And if the United States can successfully pursue this approach in AV, then it arguably can do so in a wide range of other sectors where it seeks to maintain its advantage against growing Chinese competition.
Scott Kennedy is senior adviser and Trustee Chair in Chinese Business and Economics at the Center for Strategic and International Studies. He is finishing a report, Beyond Decoupling: Maintaining America’s High-Tech Advantages over China.
This commentary was made possible by support from Cruise LLC.
Commentary is produced by the Center for Strategic and International Studies (CSIS), a private, tax-exempt institution focusing on international public policy issues. Its research is nonpartisan and nonproprietary. CSIS does not take specific policy positions. Accordingly, all views, positions, and conclusions expressed in this publication should be understood to be solely those of the author(s).
© 2021 by the Center for Strategic and International Studies. All rights reserved.