Xi Jinping is scheduled to make a state visit to the US, meeting Trump on September 24. Ahead of the trip, He Lifeng is also expected to meet Scott Bessent this weekend in New York. An extension of the Busan agreement, the establishment of a Board of Trade to reduce tariffs on non-sensitive goods and discussions over AI risks are on the agenda.
In anticipation of next week’s meeting, we’re looking in this newsletter at how Chinese establishment scholars assess:
What “constructive strategic stability” actually means
Prospects for progress on trade and tech controls
The scope for cooperation on AI safety
Whether Taiwan could derail the rapprochement
1. Overall Framing
The framing of “constructive strategic stability” has remained consistent since our coverage of the consensus view in May.
“Strategic stability” is presented by many scholars as a consequence of “strategic stalemate”, a term that the IR scholar Jin Canrong has previously explained as emerging from Mao’s On Protracted War. It refers to temporary stability achieved through mutual deterrence, and in this case is widely seen as having been attained as the outcome of improvements in China’s relative strength.
The “constructive relationship” part is more aspirational and is how Chinese scholars often distinguish this paradigm from the nuclear-deterrent-based “strategic stability” between the US and the Soviet Union. Zhang Wenzong of CASS articulates the prevailing understanding (shared but framed differently on either end of the hawk spectrum) that strategic stalemate based on economic deterrence is less stable than a relationship also based on “constructive” cooperation. However, he also acknowledges that it is the more challenging part: “moving from negative to positive strategic stability will take time”.
2. Trade and Tech Controls
The domestic economic situation of both countries is seen as the main factor compelling both sides to value stability at the current time. Echoing others, Li Junyang of CASS presents the first year of the Fifteenth Five-Year Plan as a “crucial year for consolidating foundations and bridging past achievements with future development” (read, domestic economic pressures which a return to a trade war would not help). Trump, meanwhile, is seen as motivated in the short term by a desire to deliver benefits to core constituencies before the midterms.
The aftermath of the midterms, however, could change this calculus. Huang Jing, a senior Shanghai-based scholar, argues that a Democratic House majority would increase congressional oversight of Trump’s bilateral diplomacy, while House polarisation could make perceived softness on China a more frequent line of attack.
The baseline expectation for the Xi-Trump meeting is a prolongation of the Busan trade truce agreed last October. In terms of what a more wide-ranging agreement could involve, competition over the AI ecosystem is a central concern. A recent essay from CUHK-Shenzhen scholar Huang Ping, which we translated last week, sets out an ambitious vision of how the US and China could work out a complementary “division of labour” regarding different product segments in the global AI market, even without the US relaxing tech controls at the frontier:
What could emerge at this stage should be a provisional and non-exclusive functional division of labour that could evolve and allow countries to move into more advanced roles, rather than a fixed hierarchy. The United States would continue drawing on its strengths in frontier research, top-tier chips, cloud computing and global capital. China would draw on its strengths in open-source models, low-cost inference, industrial applications, smart devices and deployment at scale.
As Huang explains, it is in China’s interests to avoid a hard, geographical bifurcation of global AI markets. On the US side, Bessent has more recently signalled openness to discussion on “avoiding bifurcation of our two systems” in AI. An agreement in this area, however provisional, could somewhat ease fears of either side using a safety agreement to gain technological and commercial advantages.
3. AI Safety
Chinese state media has tended to present the current concern with AI safety in US tech circles as a hegemonic tool—a pretext for restricting China’s AI development and unfairly singling out open-source models. Alongside an op-ed from Minister of State Security Chen Yixin, these statements have reinforced the common perception that elite Chinese policy circles do not see AI risk in the same way as their US counterparts.
However, it is reasonable to expect that public statements at this time would be intended to present an uncompromising front, in order to avoid undermining China’s negotiating position ahead of talks.
In fact, some of the same AI safety concerns prominent in the US are also circulating in China’s policy ecosystem. Notably, some establishment scholars have argued that open-source models present greater challenges to risk management than closed models in key respects. Writing in an issue of International Review this summer, Zhang Jiadong, a US-focused IR professor at Fudan, is explicit about the accountability challenges created by open models being redeployed on private servers:
The digital dissemination of open-source models […] is almost impossible to control effectively. When an open-source model is downloaded, fine-tuned and used to generate disinformation or attack code, responsibility involves numerous actors. These include the original developer, the model-hosting platform, those fine-tuning and deploying it, and end users. Yet liability disclaimers in open-source licences provide institutional protection precisely for the upstream actors in this network.
Separately, recently writing in Contemporary International Relations, CICIR’s house journal, political scientists Yao Lu and Tian Yanfei single out the biosecurity risks deriving from the open-sourcing of AI and its deployment for cloud-based labs:
The wider availability of AI and the open-sourcing of technology increase the risk that biological tools will be misused. With AI, traditional laboratories are gradually shifting towards cloud-based experimental platforms, as remote operation greatly increases experiments’ reproducibility, accessibility and scalability. If this remote automation capability is exploited maliciously, it directly magnifies biosecurity risks.
The scholars propose multilateral agreements on the supervision of bio labs, institutionalised sharing of information and standards, and setting up ongoing audits of deployed open-source models.
Given that many open model deployments currently involve Chinese models deployed on servers in the US, carrying out ongoing auditing of these models would likely require Sino-US cooperation. Reputational damage to Chinese open models enabling dangerous pathogen research, the possibility of foreign countries blocking those models and risks to China’s own biosecurity could all provide incentives for cooperation here.
4. Taiwan
Taiwan is presented by Chinese scholars as the factor most likely to derail “strategic stability”. Even if cautious about interpreting Trump’s comments on Taiwan after the May summit as representing a fundamental shift in the US’s institutional position on Taiwan, mainland scholars nonetheless see them as a clearer indication that the US under Trump would not intervene in a war triggered by a Taiwanese push for independence. In a piece we translated in June, the heavyweight scholar Shao Yuqun presented the comments as introducing “a measure of clarity into the policy of ‘strategic ambiguity’”.
However, scholars do not seem to expect a formal shift to opposing “Taiwan independence” from the US. The returnee scholar Li Cheng made this point by opposing a rush to a “fourth communiqué” that would formalise a more non-interventionist US policy on Taiwan, due to the backlash this could create within US institutions. Even without formal strategic clarity, the “time is on China’s side” view has gained a stronger position in public scholarly debate in the aftermath of Trump’s May visit. This includes the recent formulation “de facto unification without formal unification” (不统而统) from the head of the Taiwan studies institute at CASS, Zhu Weidong.
Approval of the pending $14 billion US arms sale to Taiwan, paused since May, is seen by scholars on both sides of the Pacific as the key test of whether this can last. However, Chinese scholars still hedge their language on how severely this would jeopardise the overall framework.
READ MORE
