America has a limited window of opportunity in its rivalry with China. To exploit this, expanded export controls can be used to slow Chinese innovation. Time is therefore a strategic pillar of technology containment. It enables the sequenced implementation of actions, such as imposing export controls and pressuring allies. In turn, Washington could prioritize rebuilding its domestic industries and forging a Washington-centric, tech-economic coalition that is more resilient than the efficiency-focused approach of the 1990s and 2000s. The United States initiated its tech containment by targeting China’s tech champions, including Huawei and ZTE, forcing allies to remove Huawei from 5G networks. Subsequently, it moved from targeting Chinese companies to targeting the broader tech landscape in China by expanding export controls to hinder China’s AI development and military modernization, while simultaneously offering subsidies like the CHIPS and Science Act—and applying geopolitical pressure on Korea and Taiwan—to rebuild domestic chip capacity.
By slowing China’s military modernization and bolstering Taiwan’s defenses, America aims to delay a potential Chinese invasion, making things more difficult for Beijing. Building a domestic industry will enhance resilience against disruptions in the East Asian supply chain. Ultimately, then, time is the essence of America’s tech containment strategy. While it may not completely prevent China from developing its domestic tech industries, slowing China’s progress and gaining time for the United States is a crucial objective. By implementing a comprehensive strategy that includes industrial policy reforms, coordinated export controls with allies and partners, military and nuclear modernization, and a focused effort to reshape the geopolitical, geoeconomic, and technological landscape in the Indo-Pacific, the United States can use this valuable time to slow down China’s innovation. This temporary setback could be crucial for building up Taiwan’s military as tensions and the likelihood of a Chinese invasion increase. This is not to say that this approach is without risk. Washington should be mindful that excessive reliance on these controls could diminish US leverage over time, harm American businesses with a global technological presence, and strain relations with key allies.
Our central message is this: The American two-pronged effort to contain China’s technological advancement and capacity for military modernization, while building up its own domestic semiconductor industry through initiatives like the CHIPS Act, is a balancing act between the time-sensitive nature of Chinese domestic development and the ability of the United States to leverage its network of close allies—including Japan, the Netherlands, and South Korea—to coordinate on the full extent of US-led export controls. This balancing act is born of the wisdom of both techniques to maintain American technological superiority, but too easily becomes disproportionately centered on the expanding unilateral use of export controls at the expense of coordination efforts with US allies and partners. We are chiefly interested here in re-balancing the American strategy of tech containment by revising its approach to export controls. To this end, we recommend a two-pillar approach: a proactive, aggressive export control regime that aims to solidify a Washington-centric tech coalition coupled with a technology dialogue to coordinate new controls with partners and allies.
The Small Yard, High Fence Approach in Perspective
The October 2022 release of an expansive slate of export controls, designed to cut off Chinese firms’ access to high-end AI chips involving American chip design, manufacturing equipment, components, and workers, was a major volley in an ongoing expansion of the US export control regime. These included a new restriction on the participation of “US persons” (American citizens and companies) in the direct or indirect support for Chinese companies’ advanced chip manufacturing efforts, a “novel” approach that implicates the important servicing and repair of equipment used to produce advanced node semiconductors.
The same month, national security advisor Jake Sullivan described the United States as implementing a “small yard, high fence” approach to export controls in which the broader US-China trade relationship is permitted, whereas a portion of advanced technologies deemed critical to US strategic interests are subjected to stringent restrictions. Indeed, Secretary of Commerce Gina Raimondo reaffirmed this approach in June 2024, noting that the fence (the stringency of the restrictions) could never get too high, but the yard (the number of technologies subject to restrictions) conceivably could become too large. Sullivan echoed confidence in the ability of US-led export controls to short-circuit Chinese technical progress in August 2024, just before a trip to Beijing.
Indeed, American export controls are continuously evolving, consistent with this approach. For example, the United States revoked export licenses in May, which previously allowed Intel and Qualcomm to sell semiconductors to Huawei. That same month, it was reported that the Commerce Department was considering implementing a new control on the export of proprietary (i.e., closed-source) AI models to complement export controls on semiconductors, likely targeting adversaries including China, Russia, North Korea, and Iran—sometimes the four countries are lumped together as the Axis of Upheaval.
At the same time, the United States is continuing a post-hoc effort to align its allies’ chipmaking equipment export controls to China, including the Netherlands (home to ASML) and Japan (home to Nikon Corp and Tokyo Electron), as well as South Korea. This story is mixed, however. In particular, while aligned with the United States and Seoul in the mutual pursuit of greater supply chain sustainability in support of domestic semiconductor industries, Tokyo is reluctant to impose the fuller extent of American-led semiconductor export controls on China. This is because its domestic semiconductor industry develops on the back of state subsidies and because Tokyo is concerned about Beijing’s leverage over critical minerals.
More recent developments in July showed the American export control regime in a precarious form of overdrive, with the Biden administration floating the idea of employing the Foreign Direct Product Rule to impose controls on foreign-made products with minimal American technology. The goal was to coax Japan and the Netherlands into limiting the servicing and repair of chipmaking equipment that is restricted though currently operating within China—with considerable pushback by US industry actors and reluctance from allies. That same month, it was reported that the United States would exempt allies, including Japan, the Netherlands, and South Korea, from its forthcoming rule change, indicating the United States had temporarily eased up on the pressure to align. In August, the Dutch government reportedly decided to align itself with restrictions the United States has already put on companies, including Applied Materials, to withhold the renewal of ASML’s licenses to service and provide spare parts for the company’s deep ultraviolet lithography machines in China.
Washington’s efforts are consistent in this respect, with growing Congressional concerns about American technology potentially ending up in China and Russia. Notably, the House Foreign Affairs Committee passed the Enhancing National Frameworks for Overseas Critical Exports Act by a 43-3 vote in May. The legislation aims to amend the Export Control Reform Act of 2018 in two main ways: first, by clearly empowering the US President to restrict the export of AI systems (as opposed to merely their enabling hardware), and second, by restricting Americans from working with foreigners to develop AI systems deemed a risk to national security. The legislation extends a theme in Executive Branch AI policymaking in which some AI models—current or future—are believed to potentially lower the barrier of entry for malicious actors to design, build, acquire, or use chemical, biological, radiological, or nuclear weapons of mass destruction (see, section C.I.aa), explicitly noted in President Biden’s October 2023 Executive Order on Safe, Secure, and Trustworthy AI. While the legislation is far from a done deal, there is evidently a groundswell of Congressional support for further expanding the American export control regime directly to AI systems and American workers involved in their production with foreigners.
The Nature of Export Controls
The past nearly two years of high-profile export control implementation by the Biden administration spawned an intense debate on the proper direction of the American export control regime vis-a-vis China and advanced computing technologies. Export controls, often implemented without sufficient technological dialogue, pose significant risks and potential negative consequences for American geopolitical and economic interests. These controls, aimed at limiting the transfer of sensitive technologies to rival nations like China and Russia, can backfire in several critical ways. Overreliance on these controls could undermine US leverage in the long term, harm American businesses with global technology footprints, and strain relations with key allies.
First, export controls can inadvertently make China a more attractive partner for nations looking to build their indigenous tech industry. The concern here is both technological and geopolitical. By imposing stringent export controls, the United States may push these countries closer to China, as they seek alternative sources for technology cooperation. This shift could bolster China’s influence in various technologically ambitious regions such as the Gulf or Southeast Asia. Instead of isolating China, these controls might facilitate its deeper integration into the global economy, particularly in regions critical to US interests.
Second, export controls risk undermining US leverage over supply chains in the long term. The implementation of broad and unilateral controls can incentivize the adaptation of global supply chains to circumvent US restrictions. This adaptation might include the development of new supply routes, alternative technologies, and partnerships that exclude the United States. In contrast, targeted controls focusing on specific military applications, human rights abuses, and non-commercial uses could be more effective. However, this approach carries a risk, as it assumes that China is merely reactive and not actively pursuing its own decoupling from Washington.
Third, when the United States imposes unilateral export controls without immediate alignment with allies like Japan and the Netherlands, it creates a window for Chinese firms to bulk-order manufacturing equipment and stockpile essential technologies. This period of unilateral action not only disadvantages US companies but also provides Chinese firms with the time needed to mitigate the impact of future restrictions. The emphasis should be on retaining export controls for highly targeted equipment, such as extreme ultraviolet lithography machines, which are crucial for advanced semiconductor manufacturing.
Finally, the potential for undermining relations with key allies cannot be overlooked. Further requests for export controls, particularly those involving the Foreign Direct Product Rule, can strain diplomatic and trade relations. Allies might view these requests as overreach, leading to resistance or reluctance to cooperate fully. Maintaining strong alliances is crucial for the effectiveness of any export control regime, as coordinated efforts are more likely to yield the desired outcomes.
Balancing Time and Allies
Buying time is the core of tech containment of China. As the United States develops its technology strategy, time is a critical factor. The gamble for Washington is to establish a cohesive set of regulations and policies that serve to create a buffer zone, allowing the United States and its allies to outpace China and Russia in technological innovation, large-scale deployment, and global commercialization. This strategy aims to slow down China’s technological progress. Yet, US policymakers recognize that as they execute tech containment of China via export controls, they must simultaneously strengthen the United States and its allies’ positions in the global tech landscape. To achieve this, the United States has been investing heavily in domestic chip production, with initiatives like the CHIPS Act aiming to revitalize the US semiconductor industry. Additionally, the United States has been working to establish a unified export control system with key allies like Japan and the Netherlands. These countries, which control critical technologies for advanced chip manufacturing, have agreed to implement coordinated, though more limited, restrictions on exports to China.
The United States is thus engaged in an effort to retain its technological edge by leveraging its share of power over the technical foundations of AI and its cross-border flows. This, in effect, requires the United States to balance both time and allies in the continuous evolution and execution of its export control regime on advanced computing technologies and complementary investments in and adjustments to industrial policy (a la CHIPS Act). The downsides of the export control regime we noted above are undeniable, but the potential benefits outweigh the costs given the ultimate aims of US policymakers when it comes to great power competition.
Our driving concern about the US tech strategy thus far, highlighted in the foregoing section, is that prioritizing the use of export controls to contain Chinese technical advancement in areas like basic artificial intelligence research and development can too easily become disproportionate: Not only can the unilateral use of such controls alienate allies while incentivizing China’s development of alternative, US-independent supply chains, but they can shield the need for greater action at home to build on and complement legislation like the CHIPS Act.
We provide two recommendations to alleviate this concern, both involving the approach to export controls.
Re-Balancing the US Strategy of Tech Containment
First, the United States should construct and maintain a proactive, aggressive export control regime that targets the most advanced and sensitive aspects of technology supply chains.
The designation of certain technologies—chief among them AI, advanced semiconductors, and quantum information science technologies—as strategic technologies, thereby making them worthy of the enviable focus of US national security, is born of the correct assumption that the United States possesses a competitive edge in their basic research and development that is worth preserving. Our concern, to borrow Daniel Drezner’s recent framing on the expanding notion of “national security,” is that the export control regime as currently conceived risks being too reactive.
US leaders risk stretching the utility of the export control regime vis-a-vis China and advanced technologies by proceeding from the correct belief in American technical dominance to the more dubious practice of expanding these controls to compensate for a more comprehensive technology strategy. Yet, given the combination of Chinese leaders’ intent and willingness to deploy resources to escape the grip of US-led restrictions over the long term, the effective—albeit unintended—result of the current approach is to play a game of cat-and-mouse that allows policymakers and lawmakers to recruit more restrictions in place of complementary efforts that sufficiently build up America’s long-term domestic tech capabilities. These range from domestic semiconductor production to the commercial deployment of products enabled by advanced technologies at-scale.
Export controls that target the most sensitive aspects of the relevant supply chains—such as the sale and servicing of the extreme ultraviolet lithography machines used by ASML—have the benefit of being both manageable and without an ever-expanding list of accompanying risks that must also be subject to restrictions. Moreover, the aforementioned US persons restrictions released in October 2022 are a wise recruitment of America’s geopolitical power over the foreign talent acquisition and development of advanced technologies. Center for Security and Emerging Technology fellows Emily Weinstein and Kevin Wolf describe the Bureau of Industry and Security as putting “the act of US persons supporting the development or production of advanced node semiconductors and related equipment on the same legal and policy footing as a US person’s support for the development or production of nuclear weapons, chemical-biological weapons, or missiles.” Our own position is that Washington should use every single toolkit to achieve its national security concerns—even if it is an expansion of the concept of “national security”—in the most sensitive technological domains.
The flaw of the all-out, post-October 2022 approach is to mistake the near-term American and allied dominance in state-of-the-art chip manufacturing with long-term success. As scholars Hannah Dohmen, Jacob Feldgoise, and Charles Kupchan argue, not only does the development of military technologies (e.g., tanks, missile systems, drones) prioritize “reliable, well-tested chips, which are typically fabricated on older equipment,” economic Chinese ends are being served by linking multiple, less powerful chips to form “chiplets” for training and running AI models by developing “smaller and less sophisticated AI models” that do not require as much computational power. This latter mode of AI development is part of a broader trend towards practical applications of AI models and a pragmatism about commercial deployment—which in part results from the conditions imposed externally by US-led export controls, but also by internal political conditions and specific needs of the Chinese economy.
The same reasoning would apply to restrictions on the export of AI models themselves, as opposed to merely their enabling hardware, should these come to pass. The designation of current state-of-the-art AI models in highly specific subfields of AI (which, on the surface, often amount to those that have the greatest commercial enthusiasm—an important difference) as matters of national security could prove short-sighted: The basic research and development involved within and across subfields in AI is likely to be a long-term endeavor. The “yard” that includes the models that are considered state-of-the-art today should not be so large as to hamstring US ability to engage in AI diplomacy with partners and allies in a consistent manner, nor to exclude complementary measures taken by the US federal government to shore up domestic AI Research and Development (a matter supported by the US Senate’s bipartisan AI Working Group’s Roadmap).
Second, the United States should lead the establishment of a technology dialogue with partners and allies to coordinate on new export controls and other restrictions relevant to advanced technologies.
Two of our noted concerns with the employment of unilateral US export controls on advanced technologies involve the unintended effect of making Beijing a more attractive partner for state actors in the Gulf and Southeast Asia with technological ambitions of their own designs as well as alienation of close allies through perceived American overreach (e.g., via the Foreign Direct Product Rule). Both concerns can be ameliorated through the US-led establishment of a technology dialogue with both allies and partners.
A technology dialogue would lay the foundation for a coordination mechanism through which the compliance and licensing of critical technologies can be achieved with US partners and allies without excessive strain, cajoling, coercion, and alienation therein. Such a dialogue is decidedly within America’s own interest: It acts to preserve its technical dominance by maintaining an export control regime on the most advanced technologies. It does this without putting the safeguarding of its technical secrets in undue tension with the geopolitical necessities of retaining allied coordination in relevant areas of the supply chain and remaining the most attractive partner to actors’ intent on pursuing the build-up of their domestic tech ecosystems. In this way, a technology dialogue serves the end of balancing both time and allies.
Image: Adobe Stock
This post was originally published on here
