What, exactly, is the current goal of the U.S. government’s export controls on China? Confusion within the tech and manufacturing industry reached a crescendo in mid-December following an interview given by Commerce Secretary Gina Raimondo. In it, Raimondo stressed that the aim of controls on exports of advanced graphics processing units (GPUs) — a key element of any computer — was to “stop China from developing frontier artificial intelligence models.” 

This was the first time a senior U.S. government official had tied the GPU controls to a specific AI application: “Frontier model” is a relatively new coinage referring to advanced large language models (LLMs) of the type behind ChatGPT. Previously, Raimondo and other officials had stressed that export controls on semiconductors were part of a “small yard, high fence” strategy designed only to protect the most advanced chips, and usually cited military rather than AI applications. Yet National Security Advisor Jake Sullivan is now also talking about “freezing” China’s semiconductor industry capabilities — a significant difference from the Biden administration’s older language around export controls that talked only about building higher fences around narrow areas of technology. 

There is… a mounting sense that government policy reflects an inadequate understanding of the intricacies of these complex, globe-spanning, and critical industries. 

Technology companies, and the semiconductor industry in particular, are finding it increasingly difficult to understand the administration’s goals and seemingly fluid justifications for policy changes. There is, too, a mounting sense that government policy reflects an inadequate understanding of the intricacies of these complex, globe-spanning, and critical industries. 

Last summer, the chief executives of three leading U.S. tech companies — Nvidia, Qualcomm, and Intel — visited the White House to share concerns about the moving goalposts on technology controls and the negative impacts on supply chains, existing business models, and innovation. However, the administration did not substantively address these concerns, according to the New York Times’s report of the meeting. 

These firms are also evidently unimpressed with prominent Washington DC think tanks’ understanding of their industries, with their analysis of export controls often viewed as amounting to cheerleading for the administration’s policies. Typifying this grievance is a series of recent papers and op-eds, such as this NYT guest opinion essay from late December, that do not attempt to assess the underlying rationale for the massive set of export controls issued over the past two years, or for putting nearly a thousand Chinese firms on the Commerce Department’s Entity List for a bewildering array of reasons. Rather, much of the analysis, particularly in this recent op-ed, accepts as valid the administration’s overarching framework for these controls as aiming to slow China’s military modernization. 

A closer look, however, tests in part or in toto the integrity of this framework and reveals contradictory justifications at work.  

Indeed, the goalposts of export controls targeting Chinese firms have shifted repeatedly under the Trump and Biden administrations, beginning with the Entity Listing in 2019 of China’s leading technology company, Huawei. Multiple adjustments in demarcating specific technology capabilities have occurred as officials belatedly realized they had failed to account for the complexity of the manufacturing processes and supply chains at play. A review of these changes, presented below in roughly reverse chronological order, reveals a haphazard process which has resulted in considerable damage to U.S. industry.

Company profiles: Alibaba, Tencent, SK Hynix, SMIC, YMTC, ZTE, Xiaomi.

For the American and global semiconductor industry, these ever-shifting restrictive tools are a nightmare to navigate. Industry trade groups have been reluctant to criticize the administration, routinely acknowledging they understand it has inherently “legitimate” national security goals. 

This changed somewhat last summer when rumors of the October 2023 controls began leaking to the media. The Semiconductor Industry Association (SIA), which represents many major manufacturers, toolmakers, and design firms, issued its strongest statement last July, stating: “Repeated steps…to impose overly broad, ambiguous, and at times unilateral restrictions risk diminishing the U.S. semiconductor industry’s competitiveness, disrupting supply chains, causing significant market uncertainty, and prompting continued escalatory retaliation by China.” 

China’s Semiconductor Industry Association (CSIA) two days later published an equally strong statement, emphasizing that with the rumored expansion of controls, “the globalization of the semiconductor industry is severely threatened. These moves generated growing concerns around the world.” SIA followed this with a terse statement on October 17 pointing out that “overly broad, unilateral controls risk harming the U.S. semiconductor ecosystem without advancing national security as they encourage overseas customers to look elsewhere.”  

The use of the export control system in ways it was never designed for — for example, to perpetuate U.S. dominance across dual-use technologies with global supply chains, like semiconductors — appears to have reached a critical point. Industry attempts to engage in substantive dialogue with the administration have not resulted in any rethinking of the sprawling measures outlined above. Instead, recent comments from Secretary Raimondo suggest a doubling down. 

Industry leaders are now in the position of supporting the sentiments of the SIA statement while not wanting to appear overly critical of the administration as the Commerce Department is deciding on funding grants and incentives under the CHIPS Act: The first grants to firms such as Intel, TSMC, Samsung, and Micron will be issued in early 2024.

…the national security gains of the entire export control effort remain difficult to measure, while the costs continue to mount.

Still, industry leaders such as Nvidia CEO Huang have increasingly warned that U.S. controls are likely to spur China to make heavier investments across the semiconductor value chain, further undercutting market access to leading U.S. firms that previously derived considerable revenue from China that they use to further their R&D efforts and maintain global leadership. In June, Nvidia CEO Jensen Huang was blunt in his assessment: U.S. controls have done “enormous damage” to the U.S. tech industry. His views are clearly representative of much of the industry. 

Evidence is mounting that one of the major impacts of the past five years of export controls has been a significant improvement in China’s domestic toolmakers’ capabilities and major new investments in developing advanced lithography systems. China remains able to use existing tools such as DUV lithography to develop advanced semiconductors, as demonstrated last summer by the release of Huawei’s Mate 60 smartphone. ASML CEO Peter Wennink and other industry figures have been highly critical of the Biden administration’s overall approach—Wennink noted last year that “the more you put them under pressure, the more likely it is that they will double up their efforts.”

We are likely to see yet another round of controls, new Entity Listings, and tightening of licensing around specific Chinese technology companies in early 2024, as the goalposts continue to be moved. In early January, for example, media reports indicated that senior U.S. government officials had weighed in directly to pressure the Dutch government to control exports of ASML’s DUV tools to Chinese customers who had already contracted to purchase equipment. Details around these events suggest that both the Dutch government and ASML do not accept U.S. government arguments about why ASML should not export less-than state-of-the-art equipment to Chinese customers — China accounts for nearly half of ASML’s revenue, and like U.S. toolmakers, the Dutch firm argues that undercutting its market in China for unclear national security gains risks undermining its ability to continue to innovate and lead the sector. 

This and other similar issues mean that industry views are hardening around the conclusion stressed by the SIA statement: That the national security gains of the entire export control effort remain difficult to measure, while the costs continue to mount.

Paul Triolo is Senior Vice President for China and Technology Policy Lead at Albright Stonebridge Group. DGA Albright Stonebridge Group works with a variety of companies in many sectors, including healthcare, fashion, agriculture, and technology. The firm advises companies across the technology stack, including working with semiconductor firms, consumer electronics companies, social media companies, firms part of the Al supply chain, and firms in key verticals deploying Al cross their business operations. Mr. Triolo has been writing regularly on technology policy related issues, including export controls, since 2016.