Analysis

Neither side expected much from Xi Jinping’s late September visit to Washington. Expectations were met. In this regard, it was a success. The success, however, was limited. The agreements on trade and tariffs were narrow in scope and easily reversible (see Business Matters below). These measures will not ease underlying tensions. Given fundamental conflicts over China’s continued espionage and cyberwarfare, subsidies of low cost exports to the West, support to Iran, and threats to its neighbors, the likelihood of resolution or even progress on key topics in the near future is small.

As predicted, AI featured prominently in discussions. China’s “Eight Deliverables and Understandings” from the summit notes the establishment of a China-US AI Dialogue (or Super Intelligence Dialogue in the United States) that will next meet this November, along with a bilateral communication channel for AI incidents. The declared purpose of this dialogue is also limited: namely,  “...to exchange views on risks and benefits related to AI.” Mere exchanges of views will ring hollow for many in the United States, with tech companies like OpenAI recently revealing that PRC-based Moonshot AI was extracting masses of data on its AI models by PRC-based Moonshot AI. Anthropic similarly noted in the lead up to the summit that PRC-based actors were attacking Anthropic’s platforms. If cooperation on AI safety is as important to the PRC as purported at the summit, then it would appear that PRC tech companies have yet to get that memo. US AI labs’ alarms about PRC espionage do not seem to have translated into USG action against the PRC, at least not publicly, as the PRC remains undeterred and continues its attacks.

Actual summit accomplishments were thin: pandas will come to US zoos; the US emphasized cooperation on combatting illegal flows of drugs like fentanyl; and the two countries agreed in loose terms (with slightly different phrasing) that Iran should neither have nuclear weapons nor block the Strait of Hormuz. China may have publicly agreed, but it will do nothing to achieve those outcomes. None of the agreements changes the trajectory of US-China relations away from continued political and economic conflict 

Any hopes that the summit might improve bilateral business relations were misplaced. This is especially salient for US tech companies and importers. As detailed below, the summit took place amid a slew of actions that are accelerating decoupling of the PRC and US economies. This strategic decoupling will exert a much greater effect on business than words at an international meeting.

Analysis

The context of the Trump-Xi meeting was one of intense mistrust and a push to mutually decouple economies that remain intertwined. While full decoupling is arguably impossible, decoupling the US’s military-industrial base remains a top priority for the Trump administration. Recent expansions of various sanctions, export control, and restricted entity lists highlight one way in which the current administration has pushed US companies to find alternatives to PRC suppliers and customers. Addition to these lists will continue. US companies need to prepare for not only finding their suppliers being added to these lists, but also their suppliers’ suppliers.  

For companies with Pentagon contracts, the push to decouple is immediate. In legal terms, the 1260H list of Chinese military companies currently ban these US companies from procuring from Chinese companies on this list. Next year, this list is expanding to include tier 2 and tier 3 entities in a supply chain. In legal terms, as of June 2027 the prohibition will expand to cover "for the procurement of goods or services that include goods or services produced or developed by" entities on the Section 1260H List. Non-compliance creates a risk of liability under the False Claims Act (FCA).

This means that companies with Pentagon contracts will need to demonstrate that they have investigated not only their immediate (tier 1) suppliers, but also their suppliers’ networks and beyond. This is a potentially vast expansion for diligence requirements. Failure to adequately investigate one’s supply chains could result in contract termination without notice. For companies unsure about how to proceed, the Pentagon has launched a new service known as Lynx to connect companies with capital opportunities. 

US companies with US government contracts or funding will find that their space for continued business dealings with Chinese entities will narrow. But the broader trend is important even for companies that do not have defense contracts: ties to PRC entities will raise red flags for government regulators, which could expand to red flags for financiers, lenders, other suppliers, and even customers. In short, having PRC entities in a supply chain will likely not only be risky in terms of private sector concerns, but it will also carry increased legal liabilities for companies that do not find non-PRC alternatives. This is also not unique to the United States; other NATO countries are making similar moves to secure military production supply chains. China is doing the same. 

While the Trump-Xi summit pointed to a potential softening of US-China tensions, moves by the Pentagon and others therefore demonstrate a tightening of the conditions for cooperation. US businesses need to plan accordingly.

Analysis

While the meeting between Presidents Trump and Xi produced few major headlines, this does not mean that nothing happened. In fact, this discrepancy points our attention to what is becoming the new normal: a distinction between “ordinary” and “strategic” commercial products.

As an update on our previous Red Report story, the US-China Board of Trade has officially been given its first task, a “30-FOR-30” agreement in which both sides will reduce US$30B worth of tariffs on a range of products. The United States plans to reduce tariffs primarily on Chinese manufactured products, including small appliances, toys, and holiday decorations. China intends to reduce tariffs on US agricultural products, including corn, wheat, meat and dairy. This agreement was accompanied by an announcement that the current tariff truce is being extended until January 10, 2027.

Another outcome is the formalization of a US-China Board of Investment (中美贸易理事会), which is tasked with reducing barriers to foreign investment and identifying new investment opportunities. Companies and investors alike should keep an eye on how this body develops because its efforts to stabilize US-China bilateral ties may result in windows of investment opportunities that will come with the US government’s preapproval, minimizing regulatory risk, though not necessarily other risks, including economic coercion and IP theft.

Equally as important is what was not included in these agreements, namely any kind of meaningful resolution to disputes over advanced technologies and their building blocks. This includes things like AI (SI), semiconductors and chips, aerospace, and critical minerals. For instance, while both sides agreed to establish a US-China AI Dialogue (中美人工智能对话) to address pressing concerns, there was no language specifying desired outcomes. US export controls on semiconductors and bans on Chinese nationals’ use of frontier AI models remain in place, and the Trump administration has even slowed approval of export licenses for China-bound aircraft components in an effort to create more leverage over China’s rare earth monopoly. 

The outcome of all of this movement is the division of trade into ordinary and strategic commerce. The selective normalization of “non-sensitive products,” which appears to include consumer and agricultural products, will provide stability across a wide range of sectors that are currently facing instability due to fluctuating and unpredictable tariff rates. At the same time, “sensitive products,” like advanced technology, seem to be hardening into their own category that remain subject to distinct, tightening, and evolving regulation. The fact that no concrete measures were agreed on at the recent Trump-Xi meeting does not bode well for short-term regulatory or trade stabilization in these sectors.

Analysis

The CCP has long been willing and able to curtail the movements of its citizens in the name of party power and national security. Overseas travel restrictions are common for military and government workers on national security grounds. But the CCP’s recent extension of travel restrictions to not only top AI talent in the private sector, but also their family members, demonstrates how China is expanding its definition of national security logic to cover its technology sector. 

The party is falling back on policy levers with which it is already familiar. Travel restrictions are one tool of several mechanisms through which the party steers talent and innovation. Other levers include a mixture of carrots and sticks. Incentives include state funding, integration with university research labs, and export support. Disincentives include threats of state seizure of company IP or assets, detention or  imprisonment for noncompliance, and the pervasive risk that the state will cast favor on a competitor in a hyper-competitive market. 

The CCP’s logic is that since AI is central to national security, then AI innovators are akin to military technologists. So-called private citizens working in AI development in the PRC are learning that they are private only if the party allows them to be. Working in advanced technology conveys status as a state asset.

Expanding restrictions to family members signals the CCP is intent on maintaining leverage over its leading AI experts. If individuals are able to move their family (and financial assets) out of the PRC, then the CCP has less leverage over individuals working in AI in terms of keeping them in the PRC or coercing them into developing AI models in line with the party’s priorities. Conversely, if individuals can move their family members to the United States or elsewhere, the PRC has a harder time coercing individuals to do what it wants. The point is party control over citizens, IP, and the innovation process. This also implies that the PRC will restrict outbound foreign investment from the PRC–as it did with Manus at the beginning of the year–if it entails the transfer of sensitive technologies abroad. 

For US tech companies, these restrictions will make it increasingly difficult to hire PRC citizens who are not already based outside of mainland China. Part of the draw for many in China’s tech sector has historically been their capacity to use in-demand skills to attend foreign universities (particularly in the United States) and to work in Silicon Valley or elsewhere, while moving their family out of the PRC. That door is now closing, if not closed. Individuals training at tech programs in the PRC will have far more limited options outside of the PRC. The pipeline from the PRC to Silicon Valley is therefore spluttering, with potentially dramatic consequences for the US tech sector.

Analysis

Earlier this summer, Lockheed Martin tried to ship two F-35 components from Australia to the United States, but the shipment was diverted to Hong Kong and thus to PRC authorities. (Lockheed Martin demanded the return of the equipment. The government of China has thus far not responded.) The equipment, a canopy and bomb bay door from Lockheed Martin’s F-35 strike fighter, contained classified, radar absorbing coatings. The ability to analyze the physical coatings for their properties will help the People’s Liberation Army test how to defeat or reduce the effectiveness of the material, and possibly even to reproduce or improve these stealth features for China’s own use. The damage done to US national security is difficult to calculate. Certainly, at least in the short term, Lockheed Martin will suffer some costs, including to its reputation and possibly to its value.

The diversion of secret technology to adversaries is a problem for commercial businesses as well as government and government contractors. In 2018, Huawei illegally and in violation of a contract shipped Akhan Semiconductor’s diamond coated, unbreakable glass to China and physically tested the sample in an attempt to reverse engineer it and acquire the technology without paying (i.e., steal it). It is true that Huawei could have purchased Akhan’s glass openly on the market, but by that time, Akhan would have established first-to-market advantage. What information companies share and with whom are important. So is timing.

Invalid assumptions played a role in both the Lockheed Martin and Akhan cases. In the latter, Akhan incorrectly assumed that Huawei was a legitimate potential buyer and would respect US law and honor the contract, rather than try to analyze the glass for purposes of stealing Akhan’s intellectual property (IP). In the former case, Lockheed Martin assumed that using commercial shipping was a secure method of moving sensitive technology because it had always done so. (It is not clear that previous shipments were secure; a GAO study from 2023 revealed that one F-35 prime contractor (not further identified), could not account for more than one million spare parts over a five-year period.)

We do not know for certain all the details of the F-35 parts’ diversion because the case is still under investigation. Regardless of whether the investigation finds the diversion to be a result of error or theft, it is certainly in the interest of any company with valuable IP (or classified defense articles) to track, account for, and maintain chain of custody over that IP. “We’ve always done it this way,” or “We follow all applicable laws and regulations” are not responses that will satisfy boards, investors, or Congress. Mechanisms by which competitors illicitly acquire IP are many, varied, and continually evolving. Defenders of that property need to adapt their practices as well.