A New AI Dialogue, An Old Chip War: What Trump And Xi Face In Shenzhen

US President Donald Trump is travelling to China's technology hub of Shenzhen next week for the Asia-Pacific Economic Cooperation (APEC) summit, where he is expected to meet Chinese President Xi Jinping. Trade, tariffs and rare earths will be among the issues confronting the two leaders, but the most consequential discussions could concern access to advanced semiconductors and the future of artificial intelligence (AI).

The meeting comes as Washington seeks to restrict China's access to computing technology with potential military applications, while allowing limited exports of advanced Nvidia processors to approved Chinese customers. At the same time, US lawmakers are pressing for legislation that would extend export controls to foreign entities accessing American computing hardware remotely through cloud services.

The issue has become particularly prominent following a US congressional investigation into reported Chinese intelligence links involving Nvidia founder Jensen Huang. On Thursday, Nvidia announced that Huang had resigned as chief executive, appointing finance chief Colette Kress as interim CEO. The company has not attributed the leadership change to the investigation.

The Shenzhen meeting will therefore take place against a wider contest over trade, industrial supply chains and control of advanced computing. The question for Washington is how to preserve its technological advantage while managing the commercial interests of American semiconductor companies and preventing restricted technology from reaching Chinese military-linked users.

What Is On The Platter At The Trump-Xi Summit?

The 33rd APEC Economic Leaders' Meeting will take place in Shenzhen on November 18–19, with Xi chairing the summit. The accompanying APEC CEO Summit is scheduled for November 17–18 and will bring together business and political representatives to discuss trade, artificial intelligence, connectivity and economic development.

Trump said on Sunday that he would travel to China for the meeting and expected to meet Xi. Their September discussions in Washington had already established an understanding that the two countries would support each other's APEC and G20 events, with both leaders intending to attend.

APEC itself is a 21-member economic forum whose decisions are based on consensus and voluntary, non-binding commitments. The bilateral Trump-Xi meeting on its sidelines will therefore be separate from the wider leaders' deliberations. No detailed final agenda for their meeting has been publicly established.

Trade will be an immediate priority. In September, the two countries announced recommendations under a newly established US-China Board of Trade, identifying US$30 billion worth of non-sensitive goods on each side that could receive more favourable tariff treatment in the future. This is a proposed framework, rather than a completed tariff agreement.

Washington is also pressing Beijing to fulfil earlier commitments on agricultural purchases and critical minerals. On November 6, US Trade Representative Jamieson Greer and Treasury Secretary Scott Bessent held a video call with Chinese Vice-Premier He Lifeng, with Greer emphasising the importance of China's commitments on agricultural purchases, rare earths and other critical minerals.

These issues are interconnected. China's control over parts of the rare-earth supply chain gives it leverage over American manufacturing, while US restrictions on advanced chips constrain China's access to technology considered important for AI and military modernisation.

Why Are Semiconductors At The Centre Of The Rivalry?

Advanced semiconductors are essential to training and operating sophisticated AI models, high-performance computing systems and military applications. Their processing capacity, memory bandwidth and ability to handle large volumes of data make them strategic assets in the technology competition between Washington and Beijing.

The US began tightening semiconductor export controls towards China in October 2022. These restrictions have since expanded to cover advanced computing chips, semiconductor manufacturing equipment and related software and components.

In December 2024, the Bureau of Industry and Security (BIS) introduced further restrictions, adding controls on 24 categories of semiconductor manufacturing equipment, three types of software tools and high-bandwidth memory. It also added 140 Chinese entities to the Entity List, tightening restrictions on companies and organisations considered involved in China's semiconductor and military capabilities.

Nvidia's H100 and H200 processors illustrate the stakes. These high-performance GPUs are designed for demanding computing workloads, including AI model training. The H200 belongs to Nvidia's Hopper generation and offers substantial high-bandwidth memory capacity. Its access to the Chinese market has consequently become a point of disagreement between American national-security officials and the commercial interests of US chipmakers.

Beijing has criticised Washington's restrictions as attempts to contain China's technological development. It has also pursued domestic semiconductor production to reduce dependence on foreign suppliers.

The dispute extends beyond the chips themselves. It concerns who can access the computing infrastructure needed to develop advanced AI systems, where that infrastructure is located and whether export controls can prevent foreign users from accessing it through third parties.

Has Trump Opened The Door To Nvidia's H200 Chips In China?

The Trump administration has adopted a more conditional approach towards some advanced chip exports.

On December 8, 2025, Trump announced that Nvidia's H200 and similar processors could be shipped to approved Chinese customers under controlled conditions. In January 2026, BIS revised its licensing policy, allowing export applications for the H200 and AMD's MI325X to be considered case by case.

The licensing conditions require applicants to demonstrate that exports will not reduce semiconductor capacity available to US customers. Chinese purchasers must also maintain export-compliance procedures, including customer screening, while the processors must undergo independent third-party testing in the United States.

This is not a general authorisation for unrestricted H200 sales to China. Companies still require government approval, and the Chinese government has imposed its own restrictions on purchases.

Nvidia's second-quarter fiscal 2027 filing, covering the period ending July 26, 2026, illustrates the limited commercial impact. The company said it had received US licences for small quantities of H200 processors destined for specific Chinese customers, but had been unable to complete all permitted sales because of restrictions on the Chinese side. Shipments under the licensing programme accounted for less than 1% of its Data Center revenue in the most recent quarter.

Nvidia also said it remained effectively unable to compete broadly in China's data-centre computing market under the combined US and Chinese restrictions. The company warned that continued export controls could strengthen domestic and international competitors by encouraging Chinese customers to develop alternatives.

This creates a dilemma for Washington. Allowing controlled exports preserves some access for American companies to a large market and keeps Chinese customers connected to US technology. Restricting sales more severely may limit China's access to advanced processors, but could also accelerate its efforts to develop domestic alternatives.

What Is The Remote Access Security Act And Why Does It Matter?

The semiconductor dispute is no longer limited to whether a physical chip can be shipped across a border. Chinese entities could potentially access computing capacity through overseas data centres equipped with advanced US-made processors, without importing the chips themselves.

The Remote Access Security Act seeks to address this potential route. Introduced as HR 2683 in April 2025, the legislation would amend the Export Control Reform Act of 2018 to allow the Commerce Department to regulate certain foreign access to US-controlled items through network connections.

Its definition of remote access expressly includes access through the internet and cloud-computing services from a location different from where the hardware is physically situated. The proposed authority would apply where the Commerce Secretary determines that the use of the controlled item could pose a serious risk to US national security or foreign policy.

The bill does not explicitly name China. Instead, it establishes a framework for regulating access by foreign persons to items under US jurisdiction, including computing hardware that could otherwise be made available remotely.

Its legislative status is important. The House passed HR 2683 on January 12, 2026, by 369 votes to 22. It was received in the Senate the following day and referred to the Committee on Banking, Housing, and Urban Affairs. It has not completed the legislative process and is not law.

The legislation has nevertheless returned to the political agenda. On Thursday, the chairmen of the House Foreign Affairs Committee, the Select Committee on China and the House Intelligence Committee urged Congress to pass the Remote Access Security Act alongside other semiconductor-security legislation as part of the National Defense Authorisation Act.

Nvidia has warned in its regulatory filing that the proposed legislation could restrict cloud-service providers from serving companies whose ultimate parent is headquartered in China. The company said this could affect cloud providers, equipment manufacturers and its own business. This is Nvidia's assessment of the potential impact, not a definitive interpretation of how the law would operate if enacted.

The bill's significance lies in the possibility of extending export controls from physical technology transfers to access to computing resources. A Chinese company may not need to own an advanced processor if it can rent sufficient computing capacity from a data centre in another country.

Why Has Nvidia Become A Focus Of Congressional Scrutiny?

The technology debate has acquired a new dimension following reports about Nvidia and an alleged Chinese intelligence connection.

On November 5, the chairmen of three House committees requested a closed-door briefing from then-CEO Jensen Huang on his reported interactions with Megaspeed executive Alice Huang, whom the lawmakers described as a publicly reported Chinese intelligence asset. They also sought information about Nvidia's security procedures against foreign intelligence activity and potential access to sensitive technology.

The request was renewed on November 12, when the lawmakers sent letters to Huang, Nvidia General Counsel Tim Teter and Commerce Under Secretary Jeffrey Kessler. They cited national-security concerns and called for the Remote Access Security Act and other chip-security bills to be advanced.

Nvidia's board had announced an independent investigation on November 8. It said it had no reason to believe that US national security or company secrets had been compromised, while committing to examine the available information.

A day before the latest congressional request, Nvidia announced that Huang had resigned as president, CEO and board member with immediate effect. CFO Colette Kress was appointed interim CEO, while the board began searching for a permanent successor. The company did not attribute the departure to the intelligence allegations or the congressional investigation.

China has rejected the accusations. At a Thursday press conference, Foreign Ministry spokesperson Mao Ning described the allegations as a so-called Chinese spy narrative and accused Washington of using them as a pretext for further technology restrictions. She said China's technological progress depended on independent innovation and open cooperation.

The allegations have therefore added pressure to an already sensitive policy debate, but they have not established that Nvidia's technology or US national-security information was compromised.

What Can Trump And Xi Agree On?

The leaders face competing priorities. Washington wants Beijing to meet its commitments on trade, agricultural purchases and critical minerals, while maintaining restrictions on technologies it considers strategically sensitive. Beijing wants greater access to advanced American technology and opposes export controls that it views as restrictions on its economic and technological development.

The September agreement to establish a bilateral AI Dialogue offers one channel for discussing these issues. The next exchange is expected in November, providing an additional setting for communication even as the two governments continue to disagree over chip exports and security restrictions.

Rare earths will also remain important. China's April 2025 export-control announcement covered selected medium and heavy rare-earth materials, their compounds, alloys and permanent magnets. The measures introduced licensing requirements rather than a blanket prohibition on all rare-earth exports. Their effect on manufacturers and supply chains remains a point of concern for Washington.

A wider agreement on trade or critical minerals could help stabilise the relationship, but the semiconductor dispute is more difficult to resolve. Export controls have become part of the US national-security framework, while China's pursuit of technological self-reliance is a long-term strategic priority.

The Shenzhen meeting will test whether the two sides can maintain the economic understanding reached in September without weakening their respective positions on technology and security. Nvidia's H200 licensing arrangements show that Washington is still prepared to permit limited commercial access to China, while the Remote Access Security Act demonstrates the effort to close potential gaps in the existing controls.

The central question is whether the US can preserve its advantage in advanced computing while managing the commercial and strategic consequences of restricting China's access. For Beijing, the challenge is to secure technology and supply-chain access while reducing its dependence on American chips. The summit may provide a mechanism to manage these competing interests, but the underlying contest over AI computing is unlikely to be resolved by a single bilateral meeting.