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/.
Presidents Trump and Xi are due to meet on the sidelines of the APEC Leaders’ Meeting in less than a week’s time. Progress on the key issues since the White House summit meeting on September 24 have been a mixed bag. The issues are best clubbed into four baskets.
In One Sentence
Mark the Essentials
Keeping an Eye On…
Let’s say it briefly at the outset and get it over with. The U.S. midterm election results might have been transformative from a domestic political and policy standpoint, but from a U.S.-China relations standpoint, the results are not expected to lead to much change at all. Perhaps, the prevailing sentiment against data centers will lead to a decline at the margin in terms of China and East Asia-sourced imports of data center-related hardware and other inputs. On the other hand, with the elections done, maybe the urgency to ensure China’s fulfilment of its agricultural goods import targets will also lessen. And following the Jensen Huang scandals and resignation, advanced AI chip exports to Southeast Asian data centers will almost certainly be subject to more stringent conditions to prevent their circumvention. Beijing’s access to H200 chips will also come under the scanner, particularly the identity of procuring Chinese entities and their ties to end users.
On aggregate though, China did not figure in the elections insofar as the electorate was concerned; as such, the expectation is that Democrats will prioritize domestic grievances over China policy. Hating on China solves little on the affordability, cost of living, and corruption fronts. Second, the president retains considerable constitutional latitude over foreign and security policy as well as a number of congressionally delegated trade policy enforcement authorities. President Trump’s veto proof margin could more-or-less allow him to operate on business-as-usual terms on China policy if he so chooses. The past history of second-term presidents during the post-Cold War era points in fact to their final two years in office as being a generally favorable interlude for Beijing.
Finally, hawkish views on China cross-cut the Beltway political establishment. As such, a switch away from the Republicans on Capitol Hill does not materially alter much. And in fact, on matters such as export controls on AI chips and connected devices, the Democrats are the ones liable to cause greater damage – especially in the wake of the Jensen Huang mess, as earlier noted. The probability of passage of the AI OVERWATCH Act that would explicitly require licenses for sale of advanced AI chips to countries of concern, the MATCH Act that would require allies to align export controls with the U.S. or be subjected to increased control), and the Chip Security Act that would require location verification mechanisms on chips to prevent circumvention, or variations thereof, is now more likely in this lame duck presidency or the 120th Congress.
The more intriguing point to reflect on is the thread – or rather the key data point – that connects the Republicans’ mid-term shellacking with past and recent presidential election outcomes. Although theories abound regarding their causes, one metric or data point routinely appears at or near the top – which is, the lack of median real inflation-adjusted household income growth in the U.S., going as far back as 2000. Though Trump ran and prospered politically on an economically populist, socially nativist and anti-globalist platform, what lifted him over Hillary Clinton in 2016 was the fact that U.S. median household had seen zero income growth between 2000 and 2015. Where household incomes stood in 2000, they stood at the same level in 2015, only clearing their peak during the last year of Obama’s second term. Similarly, what undid Kamala Harris was the decline in real wages through the first half of the Biden-Harris term as well as the aggregate inflationary effects on the price level. Now, along comes Trump’s shellacking for the very same reason: the lack of real wage growth over his 22 months in office and its decline, in fact, over the past twelve months (compounded by the inflationary effects on overall prices).
But why return convicted felon and sexual abuser Trump to office in the first place in 2024? That’s because the first three years of the first Trump presidency saw a handsome rise in real median incomes, reaching more than 10% over the Bill Clinton peak of 2000, and connoting that “The Donald” was a steady hand on the tiller. But was this really the case? Did he possess some secret policy sauce? Every corner of the developed world saw handsome macroeconomic and income growth during the 2015-2019 period after having spent years digging out of the deep malaise of the 2008-2009 Global Financial Crisis, suggesting that Trump may merely have been the fluke beneficiary of an economic tide that lifted all boats during the second half of the 2010s. Consumer confidence levels had been improving every year from 2011 on. And so, on this relatively narrow and perhaps shortsighted basis, the American electorate voted Trump back into office in 2024 and now will have to grapple with his disruptive and destructive ways for two more years.
One thing should be clear though: U.S. politics will no doubt continue to be refracted through the lens of contested ideologies spanning MAGA/America First to the Democratic Socialists of America in the years ahead. But those who ignore that tiny data point called median real household income/wage growth while in office do so at their own peril.
Expanded Reading
In One Sentence
Mark the Essentials
Keeping an Eye On…
Jenson Huang’s first reported affair with Megaspeed’s Alice Huang was not a good look for China; his second affair with his company’s vice-president of health care, Kimberly Powell, is turning out to be an equally bad look for the U.S. AI industry. The third reported affair just adds fuel to the fire. We’ll leave it to the FBI to determine whether Megaspeed’s Huang was indeed a Chinese state ‘honeypot’ or merely a ‘rainmaker’ who made advanced GPUs – both barred as well as legally authorized ones – rain on the Chinese AI ecosystem. On the latter point though, and after sifting through Megaspeed’s incestuous linkages with Nvidia OEM compute partners and Southeast Asian data center and related entities, it is hard to avoid the conclusion that Megaspeed was not just a procurement front for Alibaba (which is authorized by the Commerce Department to receive H200 chips), but also a conduit or diversionary front for Alibaba and other China-based entities to acquire Blackwell-equipped servers which prohibited from export to China. Megaspeed may not have been the direct recipient of the stash (Nvidia says it placed Megaspeed on its block list from mid-2025), but it certainly seems to be a key corridor through which the illicit materials passed.
Let’s limit ourselves, though, to identifying the various measures that are likely to be on the anvil in Washington to shut down these illicit channels following the Huangs and Megaspeed outing.
First and foremost are measures that would prevent chip exports as well as their diversion to non-trusted companies globally, and specifically to China. Among the first acts of the Trump administration had been to decline enforcing Biden’s AI Diffusion Rule, which had created a tiered, global licensing framework for AI chips and had grandfathered the tough existing controls on advanced chips destined for China and Chinese firms. Expect Congress to formally step in and legislate the prevailing executive branch prohibitions on advanced chip exports to China as well as to foreign subsidiaries of China-headquartered firms, including those operating under the very noses of U.S. regulators on U.S. soil, such as Aivres Systems. Simultaneously, expect Congress to double-down on chip location verification requirements to prevent circumvention. The OVERWATCH Act should mostly take care of the former; the Chip Security Act, the latter.
Second, and on a trickier note, expect new rulemaking on a Data Center Verified End User program that would impose checks on data center ownership security, baseline security of chips and data, software and network security as well as supply chain security, in order to prevent the diversification of chips. Expect Southeast Asian countries (and perhaps Gulf states) to be swept meticulously into every last detail of these new rules. Whether such rulemaking extends to cloud computing access in these data center destinations, especially in terms of know-your-customer requirements, remains to be seen. Access to compute capacity via Infrastructure-as-a-Service providers has, after all, been shown now to have been a conduit for the diversion of physical chips. And in this regard, the Commerce Department does enjoy the authority to prohibit a wide array of information and communications technology and services-related transactions under the first Trump administration’s overarching May 2019 ICTS Executive Order.
Finally, expect Commerce Secretary Lutnick to hightail it back to NYC sooner rather than later. He has lost the confidence of his peers and soon, probably, of his president.
Expanded Reading
Presidents Trump and Xi are due to meet on the sidelines of the APEC Leaders’ Meeting in less than a week’s time. Progress on the key issues since the White House summit meeting on September 24 have been a mixed bag. The issues are best clubbed into four baskets.
Board of Trade and Board of Investment
The Board of Investment does not seem to have gained any traction since the White House summit in September. Since little has been achieved on this front so far, the Board of Investment can essentially be written off – at least, for the time being.
The Board of Trade, meantime, seems to have gone into something of a minor tailspin.
On agriculture, the U.S. maintains that China is obligated to purchase at least $17 billion worth of U.S. ag. exports in 2026, 2027 and 2028, plus stick by its November 2025 purchase commitment of at least 25 million metric tons of soybeans annually through 2026, 2027, and 2028. From January-September, China had purchased $12 billion worth of agricultural goods with half of the purchases being soybeans. Effectively, China has purchased only a third of its committed non-soybean agricultural goods. China does not contest these targets. But it argues that these purchases fall under the Board of Trade’s 30 billion-for-30 billion tariff reduction framework, and purchase expansion must be contingent on implementation of this agreed-upon reciprocal tariff reduction arrangement.
On rare-earths, the U.S. argues that China is only fulfilling two-thirds of the amounts that it pledged as part of the Busan truce of November 2025, and that in the case of yttrium (which was specifically referenced in the U.S. fact sheet after the May 2026 Beijing summit, and regarding which Japanese PM Takaichi has directly expressed her concerns to Trump), the amount is even less. China rejects this claim but is known to have concerns regarding U.S. stockpiling of Chinese rare earth flows. The cascading result of their disagreement on rare earth flows is that the U.S. side is holding up export licensing for airline parts and components bound for China, which has in turn led to the slow-walking of the purchase orders for 200 Boeing aircraft that Beijing had approved at the May Summit meeting. China’s domestic national champions in the passenger aircraft sector (AVIC & COMAC) are still dependent on foreign joint venture partners for key parts, such as flight control and navigation systems, hydraulic systems and fluid, landing gear, etc. On a more positive note, the U.S. recently green-lighted the transfer of the next batch of GE-Safran JV-produced jet engines for COMAC’s C919 narrowbody jet. These jet engines had been a point of leverage in past Bessent-He talks.
None of the above disagreements on agriculture and rare earths are fatal. The Shenzhen meeting does not constitute a drop-dead deadline. That deadline, rather, is January 10, 2027, when the Busan Arrangement of November 2025 – suspension of implementation of: (a) the U.S.’ 24% reciprocal tariff and China’s equivalent 10% countermeasure, (b) the U.S. 50% Affiliates Export Control Rule and China’s expansive new October 9, 2025, rare earths export control measures, and continuation of business-as-usual flows; (c) reciprocal investigations on maritime, logistics and shipbuilding industries – expires, unless extended. As such, the December 2026 Trump-Xi meeting on the sidelines of the Miami G20 meeting assumes greater significance on the trade and export controls-related front.
AI Dialogue
AI is an area where the two sides have made significant progress – not bilaterally, but in concert, individually. In September, at the White House, the two sides had agreed to set up a dedicated AI Dialogue as well as a bare-bones notification system. Key areas of concern to be discussed within the dialogue framework are: problems stemming from uncontrolled AI agents; non-state actors’ access to AI-enabled dangerous cyber and biological capabilities; finding agreement regarding what frontier models must not do. In Shenzhen, the two sides are expected to work towards drawing-up AI safety and notification protocols as well as exchange best practices on advanced AI model pre-release testing. Relatedly, while China has a standardized incident reporting system (since September 2025), the U.S. has yet to formalize one. On the other hand, while the U.S. has a voluntary but systematic pre-release testing process to identify dangerous capabilities in advanced AI models, China’s pre-release assessments to detect dangerous capabilities still require much work.
The more remarkable progress on AI Safety has occurred at the individual level at both ends. In early October, the White House unveiled a new AI Task Force led by Director of National Intelligence Jay Clayton, which is due to report back in 120 days on a number of key risk areas: cybersecurity and critical infrastructure protection; biosecurity and dual-use applications; and model control and accountability. The task force has a wider remit, including AI buildout as well as consumer protections and AI’s impact on the workforce. And Scott Bessent, the U.S. interlocutor on the bilateral dialogue framework with China, is a member of the task force. Across the Pacific, in mid-October, China’s MOFCOM proposed a draft regulation that would essentially license – and potentially export control – some of the public-facing IP and knowhow of its near-frontier, open-weight models, which would make them less open and could slow down their global adoption.
As such, both sides are moving forward on AI Safety in concert, and pairing their individual rulemaking with a bilateral framework for incident notification and exchange of best practices on pre-release model testing.
Strategic Stability Consultations
Ever since the latter years of the Biden administration, the U.S. side has sought to engage the Chinese in arms-control and non-proliferation consultations, given Beijing’s rapid nuclear arms build-up. During Biden’s last year, the U.S. Nuclear Employment Guidance was readjusted to account for this build-up in the context of multiple nuclear-armed adversaries, and the Trump team has sought to trilateralize these consultations by including Moscow. The Trump team also has concerns regarding anti-satellite weapons being developed by Moscow and Beijing and their potential placement in space (U.S. recently confirmed its own deployment of weapons in space).
China’s answer has been a resolute ‘no’ to trilateral strategic stability consultations and continues to hold the position to this day. On the other hand, Beijing has restarted the bilateral arms-control and non-proliferation track – which it had cut-off in 2024 after an arms sales package to Taiwan – following Trump’s pausing of the latest Taiwan arms sales package. To sweeten Xi-Trump relations, China has also agreed to begin discussions on reciprocal visits to each other’s nuclear facilities, including testing sites, as a confidence-building measure as earlier this year, the U.S. had accused China of conducting low-yield tests. Whether the two sides can formalize their progress on this front in Shenzhen remains to be seen.
As for Putin’s potential presence in Shenzhen, the Chinese side is looking to manage the politics and script outcomes carefully without being drawn into expansive G2, G3, or ‘Yalta-like’ pronouncements. Perhaps, what might be palatable to Beijing is some sort of formulation on AI governance related to nuclear weapons. With China and the U.S. sharing a consensus going back to the Biden years that AI must never be allowed to dictate the use of a nuclear weapon, and with both sides having co-sponsored each other’s resolutions on AI at the UN General Assembly, a trilateralized formulation on AI governance related to nuclear weapons would be a relatively easy lift.
Mil-Mil Consultations
War Secretary Hegseth and PACOM Commander Paparo are expected in Beijing in the run-up to Trump’s meeting with Xi in Shenzhen. The two sides have been in discussions on a MoU on mil-mil crisis management and prevention, holding a working group meeting in early November. Mil-Mil crisis management and prevention is one area where the U.S. and China do not have any agreements or understandings. They have a Military Maritime Consultative Agreement (MMCA) from 1998, followed by a 2014 MoU on Safety and Maritime Encounters and a 2015 Supplement on Air-to-Air Encounters. At this time, it appears China is ready to sign on the MoU but internal divisions prevail on the American side with DoW on one side and State and NSC on the other. On the other hand, Beijing appears to be the one pushing back against formalizing certain protocols related to advance notification of missile tests – its nuclear submarine-launched ICBM test earlier this summer having come in criticism for lack of adherence to standard notification practices. It remains to be seen if Hegseth and Paparo do indeed visit China in the short time ahead, and whether they can iron out their objections in Beijing prior to Trump’s arrival.
Conclusion
For the past eighteen months, trade, investment, and export controls have been at the heart of the U.S.-China political dialogue, and modest accomplishments have been achieved in their regard. Going into the Shenzhen meeting, however, it is the least developed of the various policy tracks. While the AI track appears the most likely to bear fruit in Shenzhen, developments on the mil-mil and strategic stability tracks may also well surprise on the upside.