Will the U.S. ease technology export restrictions applicable to China?

TL;DR

I assign a 52.6% probability of YES, conditional on the September 24 meeting occurring. The main YES route is a new commitment to extend the BIS Affiliates Rule suspension beyond November 9, 2026, which the resolution treats as qualifying and the current rule says otherwise expires then (Federal Register). The probability is only slightly above even: China is asking for that exact delay and both sides previously said they would work toward extending the linked package, but Commerce Secretary Howard Lutnick has said he does not expect export-control easing at the summit and final terms remain unsettled (MOFCOM; Reuters; Lutnick report).

Context

The January 2026 policy allowing case-by-case review of Nvidia H200, AMD MI325X, and similar China-bound chips is already in the baseline. Repeating that policy or approving an ordinary license under it would not qualify (BIS).

The key live measure is the Affiliates Rule. It would automatically apply restricted-party controls to entities at least 50% owned by covered listed parties. The United States suspended it from November 10, 2025 through November 9, 2026 as part of the prior Trump–Xi bargain, with the restrictions scheduled to return on November 10 absent further action (Federal Register; White House).

Evidence

The historical backbone is seven prior in-person Trump–Xi bilateral meetings from April 2017 through May 2026. Two produced an identifiable, meeting-linked United States technology-export relaxation within roughly the same announcement window, for a raw base rate of 2/7, or 28.6%.

That base rate is weak because the sample is small and most meetings lacked a named export-control measure approaching expiration. The present case is more favorable to YES. In May 2026, China’s Commerce Ministry explicitly described the reciprocal arrangement as covering both the United States 50% Affiliates Rule and China’s related export controls, then said the two economic teams would work to extend the arrangement (MOFCOM, May 20, 2026). This was a promise to negotiate, not a definite extension, so it raises the forecast without entering the baseline as an already-granted concession.

The freshest reporting is also unusually specific. Reuters reported on September 17 that Beijing was pressing Washington to delay further the rule that would block thousands of Chinese firms from receiving advanced United States technology, while Washington sought better access to Chinese rare earths and critical minerals (Reuters). NPR reported that China had earlier offered a rare-earth moratorium through January 2029, while the United States sought only a six-month extension to be announced after Xi’s September visit (NPR). The NPR report does not say outright that Washington agreed to extend the Affiliates Rule in return, but I read the requested timing and the existing reciprocal structure as evidence that a matched, shorter rollover is under active negotiation.

Outside assessments point the same way but are correlated. Brownstein called a one-year extension of the broader 2025 truce the most likely summit deliverable, though it also expected limited relief for advanced semiconductors (Brownstein). Paul Triolo was more explicit, predicting that extending the arrangement holding both the Affiliates Rule and China’s October 2025 controls in abeyance would be the primary result (ChinaFile). These are informed judgments, not separate independent data points, and “truce extension” does not always mean every component will be renewed.

The strongest evidence for NO is Lutnick’s September 2 answer that he did not believe export-control easing would be addressed at the summit. His explanation focused on China’s failure to take up the H200 offer, which makes the statement more decisive against another chip opening than against maintaining the Affiliates Rule pause. Still, he runs the department responsible for BIS, so the broad wording deserves real weight (Asia Today).

Chinese compliance is another obstacle. NPR found unfinished interagency preparations seven days before the meeting, continuing shortages for some specialized magnets, and a senior United States official calling China’s performance “very disappointing” (NPR). The September 17 Chinese Commerce Ministry briefing said only that the teams remained in close contact on relevant issues; it disclosed no export-control agreement (MOFCOM briefing). That silence is not proof of failure, but it confirms that no public deal text was available at the cutoff.

The announcement mechanics favor YES once a political agreement exists. The October 30, 2025 meeting was followed by a specific White House fact sheet on November 1, and BIS implemented the stay afterward. The current question likewise needs only a definite official commitment by September 26, not a completed Federal Register rule (White House).

My event tree is:

$$P_{\text{main}} = 0.72 \times 0.76 \times 0.93 = 0.508896$$

The terms are a 72% chance of a definite extension of a meaningful part of the 2025 package, a 76% conditional chance that it expressly includes continued suspension of the Affiliates Rule, and a 93% conditional chance that the announcement is timely, specific, and not already locked into the pre-meeting baseline. If that route fails, I assign a 3.5% conditional chance to another qualifying measure, such as a new product-specific review policy, named-entity relief, or exceptional authorization:

$$P(\text{YES}) = 0.508896 + (1-0.508896)\times 0.035 = 0.52608464.$$

What's non-obvious

The obvious read is that meaningful chip relief is unlikely, so the answer should be NO. That misses the resolution rule. A temporary extension of an existing suspension counts as a new easing, even if officials market it as continuity rather than a concession. The question can therefore resolve YES without any new access to frontier AI chips.

The other subtle point is that “extend the trade truce” is too vague for forecasting this question. Some summit packages could extend tariffs and Chinese mineral measures while omitting the United States Affiliates Rule. I therefore give more weight to evidence naming that rule—the Reuters report and the May MOFCOM description—than to general predictions of a successful or stabilizing summit.

Uncertainties

The decisive text is private. The largest gap is whether the September 19–20 preparatory talks produce a matched export-control extension or a package limited to tariffs, purchases, and Chinese mineral licenses. There is also no auditable dataset showing the number, value, processing time, or denial rate of Chinese rare-earth licenses for United States firms, so claims about compliance remain qualitative and contested. Finally, a broadly worded announcement that merely says the “truce continues” could fail the question’s specificity test even if both governments privately intend to preserve the Affiliates Rule suspension. A reasonable uncertainty range around the point estimate is roughly 35% to 70%.