ICAS Trade ‘n Tech Dispatch (online ISSN 2837-3863, print ISSN 2837-3855) is published about every two weeks throughout the year at 1919 M St NW, Suite 310, Washington, DC 20036.
The online version of ICAS Trade ‘n Tech Dispatch can be found at chinaus-icas.org/icas-trade-technology-program/tnt-dispatch/.
In One Sentence
Mark the Essentials
Keeping an Eye On…
The EU is in the throes of a debate on its future China trade, investment, and industrial strategy, with several proposed tools, draft “derisking” targets, and questionably WTO-noncompliant measures being bandied about. That this debate does not veer into the territory of China Derangement Syndrome necessitates that Brussels objectively assess its economic ties with China and weigh its options carefully going forward.
A key driver of the EU’s debate has been the run-up in the bilateral trade deficit, which is an undeniable fact—the deficit having more than doubled from EUR 165 billion in 2019 to EUR 359 billion in 2025. On closer inspection, however, the run-up in the deficit is almost entirely concentrated in two product groups: new energy products (silicon wafers and PV cells; lithium-ion batteries; electric vehicles) and chemicals. The former can be chalked up to China’s genuine, world-beating competitiveness in the sector (with an assist from both fair and unfair industrial policy initiatives); the latter, which is energy-intensive, to the run-up in fuel-switching costs following Russia’s invasion of Ukraine. Most EU employment losses have been registered in this sector. The vast majority of other Chinese product exports have more or less tracked their unremarkable pre-2019 trend. As such, arguments that undervaluation of the RMB, redirection of exports from the U.S. market, or root-and-branch Chinese subsidies are to blame cannot be sustained. Higher EU tariffs to fight circumvention, or the case for a powerful tariff-rate quota-based safeguards tool, cannot be justified outside of the new energy products and chemicals sectors.
More broadly, the EU’s China debate would be better informed by a return to first principles, which includes gleaning insights from the policy framework that has underpinned China’s own industrial policy rise. Here are ten that merit consideration:
First, in pursuit of the EU’s decarbonization and manufacturing resilience goals, considerations of competitiveness must in principle override those of security. An excessive security focus at the expense of the competitiveness imperative will ensure that neither is achieved.
Second, achieving competitiveness means fundamentally running an open economy from a trade and investment standpoint. Where security considerations must, willy-nilly, trump those of competitiveness—say, in making the case for a proposed “resiliency tool” or certain mandates within the draft provisions of the Industrial Accelerator Act (IAA)—the benefits of such interventionism must exceed the associated security premium to be paid, and that premium should be spelled out clearly.
Third, tariffs and trade defense instruments (safeguard, AD/CVD measures) must fundamentally adhere to multilateral treaty law—although creative, consensual workarounds like price undertaking arrangements that provide breathing room to industry, such as that for EVs, could be proactively scoped out. As a general rule, intermediate goods imports from China must face the least border restrictions.
Fourth, with regard to Chinese inward foreign investment, greenfield investment from private actors in the new energy sector (solar, batteries, EVs) should be welcomed with open arms. The test for Chinese private sector acquisitions of EU-based firms in other emerging and mature industries should be whether the acquirer adds to or detracts from local production and employment, especially with regard to core manufacturing capabilities. As for Chinese state actors, their investments would require more critical review, depending on the sensitivity of the sector and the extent of their financial subsidization. They should be excluded from procurement markets and from making acquisitions in advanced manufacturing sectors where Chinese capabilities reside at the lower end of the value chain—lest these investments become an avenue for siphoning valuable local IPR.
Fifth, again with regard to Chinese inward foreign investment, the overarching goal should be to ultimately localize key segments of the intermediate goods value chain. This requires less—rather than more—interventionism, with Chinese manufacturers first incentivized to offshore aspects of their intermediate goods supply chains and local producers thereafter incentivized to absorb and capture a larger share of such intermediate goods production. China’s intermediate goods sector, world-beating today, followed this market-led pathway.
Sixth, industrial policy should be seen as a complement to competition policy, not a substitute for it. The goal of industrial policy must be the production of sophisticated tradable goods in new and existing industries—including through targeted subsidization and technology transfer incentives—but without the state picking winners or protecting incumbents. Market discipline must weed out the over-subsidized and weak without fear or favor.
Seventh, sticking with industrial policy, the incubation of strategic future industries should also borrow from the Chinese playbook, matching policy tools to product development stages. At the breakthrough stage, multiple technology pathways should be prioritized. At the commercialization stage, proof-of-concept and pilot platforms should be privileged. And at the scaling-up stage, standards-setting, procurement mandates, consumption subsidies, and public goods infrastructure should be deployed to create demand and cut user costs.
Eighth, hard security lines—be it regarding critical infrastructure, software control, sensitive data, or dual-use items—should be narrowly and rigorously drawn. The test should be less about the access and standing of Chinese players within the supply chain and more about their control or leverage over a critical node, core network, or interconnected system—say, via remote updates—as well as their privileged access to sensitive non-public data platforms. Local rather than remote presence over industrial control systems should be encouraged, and Chinese players excluded from a narrow subset of “high assurance” projects.
Ninth, Brussels should be ready to wield its countermeasures tools, of which it has many. There is significant co-dependency running in both directions through the bilateral relationship, with Brussels likely wielding greater influence over a broader set of core technologies than vice versa. The threat of countermeasures must be wielded credibly, however, which requires that they be deployed prudently.
Finally, and most importantly, Brussels needs to sit down and earnestly discuss its priorities and anxieties with Beijing. The conversation must be framed as one between peers—not one featuring a fading grandee and a tainted upstart. From the former framing will flow the working out of satisfactorily productive bargains, even if it leaves a bitter taste of mutual disdain on both ends. At the end of the day, both sides are invested in the success of their economic relationship, unlike the case of U.S.-China ties, where Washington is agnostic about failure.
The EU’s creeping drift toward China-style industrial policy measures has not gone unnoticed. The IAA’s inward foreign investment provisions contain joint venture and technology transfer requirements. The Clean Tech draft plan and the IAA contain headline targets reminiscent of China’s Made in China 2025 (MIC 2025) plan. For the first time, the update to the EU’s Dual-Use Control List includes items related to emerging technologies not listed under the various multilateral regimes. Such flexibility should be encouraged, not scorned—and the underlying strategic intent could provide a firmer basis for more engaged and productive exchanges between Brussels and Beijing.
Expanded Reading
In One Sentence
Mark the Essentials
Keeping an Eye On…
At long last, the Trump administration’s AI policy framework appears to be settling into a more coherent pattern—divided, as it has been, between striking a balance between the technology’s capacity for transformative innovation and its potential for serious harm. Earlier this March, the administration laid out its preferred AI legislative framework, balancing considerations of innovation and security, supporting creators while enabling fair use, and championing First Amendment rights while guarding against harms to children. With the unveiling of Anthropic’s powerful Claude Mythos model in April and its remarkable ability to identify software vulnerabilities, the administration has raised its regulatory game a notch. The National Institute of Standards and Technology (NIST) has expanded its work on testing models before public release, and the General Services Administration (GSA), starting earlier this year, has sought to incorporate a new standard clause in federal IT contracts that would give officials substantial control over the AI systems they use. At the Pentagon, a related “all lawful uses” agreement was struck with eight leading U.S. AI companies to ensure deployment of their capabilities on the Department’s classified networks.
Topping things off just last week were two significant White House releases: a tightly scripted executive order seeking to develop a classified benchmarking process to assess the advanced cyber capabilities of frontier models, and a more extensive National Security Presidential Memorandum (NSPM) on AI seeking to accelerate the use of AI across intelligence and warfighting domains, including through military-civil fusion. The documents strike a balance between innovation and security, and between voluntariness and mandate—and thankfully come in thousands of words shorter than the Biden administration’s own executive order on the topic, which ran to a mammoth 19,700 words. It was also reported last week that Anthropic’s Mythos is already being used by the National Security Agency for offensive cybersecurity operations against China. And the U.S. Commerce Department set out guidance last week on enforcing its advanced chip export controls and licensing requirements vis-à-vis China. The scrapping of Biden’s AI Diffusion Rule in January 2025 without a successor rule had sown considerable confusion in this regard.
The AI race between the U.S. and China is well and truly on.
If the Trump administration has been getting its act together on the AI regulatory front with a view to extending its lead over Beijing, the Chinese side has spent even more time and effort building out its own foreign-facing technology export controls and countermeasures toolkit. Indeed, it has been on a veritable tear over the past 18 months—the Outbound Investment regulation taking effect on June 1 being only its latest such measure.
In December 2024, the Regulations on Export Control of Dual-Use Items entered into force, and the first additions to its “Control List” began to be populated in January 2025. Article 49 of the regulation contains the functional equivalents of the U.S. de minimis content rule, the foreign direct product rule, and the principle of extraterritorial jurisdiction—all rolled into a single article.
In March 2025, the Anti-Foreign Sanctions Law (AFSL) of 2021 was expanded with two dozen new provisions, the most significant of which empowers Chinese entities to file lawsuits in Chinese courts against foreign persons “executing or assisting” in discriminatory measures against them. The AFSL has since made its courtroom debut in a case involving a Swiss marine equipment company and is also being invoked by the Chinese firm Wingtech in its dispute with Dutch auto chipmaker Nexperia. A draft International Civil and Commercial Judicial Assistance Law is also in the works.
In October 2025, a sweeping export control on rare earth materials was introduced that comprehensively regulates the supply chain of Chinese-origin rare earths, from raw materials to mining and manufacturing technologies, production equipment, and downstream products. The global implementation of this measure is currently suspended until November 2026 as part of a U.S.-China tech and trade truce.
In December 2025, an amended Foreign Trade Law was adopted by the National People’s Congress, codifying the implementation of unilaterally determined “corresponding measures” that stand outside the WTO framework—as well as other trade treaty frameworks—when Beijing determines that a given treaty’s dispute settlement mechanism is being frustrated.
In March 2026, the Regulations on Industrial and Supply Chain Security were released, authorizing “corresponding measures” against foreign persons that disrupt normal market transactions, impose discriminatory restrictions, or damage the security of China’s supply chains. Foreign firms that lobby for supply chain sanctions or controls on Chinese firms may also find themselves in its crosshairs.
In April 2026, the National Security Review Mechanism of December 2020 was invoked to block Meta’s acquisition of the Chinese AI firm Manus. Article 13 of the recent Outbound Investment regulation also clarifies prohibitions on “deemed exports”—an issue in the Manus case—by covering transfers of know-how, not only goods, via the dispatch of technical personnel abroad, cross-border training arrangements, or the provision of technical guidance across borders.
In May 2026, the State Council utilized its April 2026 Regulation on Countering Improper Foreign Extraterritorial Jurisdiction, as well as its 2021 “Blocking Measures,” to compel five U.S.-blacklisted Chinese refiners to defy Washington’s sanctions. The regulation comes with a “Malicious Entities List” as well. China’s countermeasures toolkit has clearly gone from bark to bite—and the world would do well to take note.
Expanded Reading
Legislative Developments
Hearings and Statements
Expanded Reading