President Donald Trump told reporters aboard Air Force One on Friday that during his two-day summit in Beijing, he and Chinese President Xi Jinping discussed guardrails for artificial intelligence (AI) and Nvidia’s H200 chips. The summit, which concluded without a signed AI governance framework, left the most-watched pieces of the deal—chip export licenses and rare-earth trade—unchanged.
When asked to specify the nature of the guardrails discussed, Trump described them as “standard guardrails that we talk about all the time,” adding that the two leaders “talked about possibly working together” on them. The vagueness of the term underscores the lack of concrete, mutually agreed-upon definitions between the US and Chinese governments, raising questions about the practical outcomes of the high-level dialogue.
Export Clearance Announced, but No Shipments Yet
Just before Trump’s arrival on Thursday, the US government cleared roughly ten major Chinese technology firms—including Alibaba, Tencent, ByteDance, JD.com, and Lenovo—to purchase up to 75,000 Nvidia H200 chips each under a new export-licensing regime, as CNBC first reported. These chips, while a generation behind Nvidia’s most advanced Blackwell line, remain critical for AI workloads such as large language model training and inference.
Despite the approval, not a single H200 chip has been shipped to the cleared buyers. The licensing framework includes unusually elaborate conditions: China-bound H200 volumes are capped at no more than 50% of Nvidia’s US domestic sales; each shipment must be verified by a US-headquartered third-party laboratory; Chinese buyers must provide certification against military use; and the deal mandates a 25% revenue share that must be routed through US territory. These layers of oversight have turned the approval into a procedural clearance rather than a physical delivery mechanism.
The practical effect of the regime is a paper clearance that has not yet translated into hardware flows. For the ten approved companies, the wait continues, while Nvidia’s ability to service a significant portion of global AI demand hangs in the balance.
AI Guardrails Elude Signed Agreement
In a Time magazine account of the meeting, AI was described as “the elephant in the room” rather than the centerpiece of the summit. The public-facing conversation focused heavily on trade imbalances and the H200 chip question, while deeper bilateral frameworks dealing with autonomous weapons, model misuse, and dual-use AI technology were discussed only in outline. Senior officials briefing on background suggested the two governments are exploring a recurring dialogue track on AI risk, but no schedule, working group, or signed text emerged from this round.
The lack of a binding agreement leaves the AI safety and governance landscape fragmented. The US has led international efforts to develop voluntary AI safety standards, including the Bletchley Park declaration and subsequent summits, while China has pushed its own approach centered on state control and ethical guidelines. The two systems diverge fundamentally on issues of transparency, enforcement, and civilian oversight, making a unified framework difficult to achieve.
Reactions in Washington and Beijing
Senate Democratic leader Chuck Schumer sharply criticized the H200 clearance, stating that “giving China access to this premier US technology is dangerous and threatens our lead in the AI race.” Schumer’s remarks underscore deep bipartisan unease in Congress about technology transfer to China, especially in the semiconductor and AI sectors. The administration, however, has defended the decision as a calibrated move that keeps Nvidia revenue—and US jobs—in the United States while preventing the most advanced chips from reaching Chinese hands.
Nvidia’s CEO Jensen Huang argued publicly last week that the H200 sits one generation behind the Blackwell line, which remains under strict export controls, and that selling regulated Chinese demand to Nvidia maintains the company’s domestic manufacturing base and R&D investment. The company’s financial outlook is heavily dependent on its ability to sell to both domestic and global customers, and the China market—even with the 25% revenue pass-through—represents a meaningful portion of its mid-term revenue forecasts.
In Beijing, the Chinese government has not publicly commented on the summit’s AI outcomes, but state media outlets have framed the talks as a necessary dialog to stabilize tech relations. Chinese rare-earth exports remain about 50% below pre-restriction levels, a retaliatory measure imposed last year in response to earlier US tariffs. Rare earths and chips now sit in the same negotiating folder on both governments’ read-outs, with no movement on either front during the Beijing summit.
Corporate and Geopolitical Stakes
The corporate stakes are visible in the broader AI capex cycle. Hyperscalers have committed more than $650 billion to AI infrastructure across 2026, based on combined Q1 2025 earnings reports from Microsoft, Alphabet, Amazon, Meta, and Apple. Nvidia sits at the center of the supply side of that spending, with its data center chips powering the majority of large-scale AI deployments. A China revenue line, even at controlled volumes and with a 25% pass-through, would materially improve Nvidia’s medium-term growth outlook.
Microsoft and OpenAI’s joint trajectory represents the visible US half of that picture; Huawei’s Ascend chips are the half the export regime is implicitly trying to slow down. Huawei has been developing its own AI chips, such as the Ascend 910, to reduce China’s reliance on US semiconductor imports. The US export restrictions are intended to blunt Huawei’s progress by denying it access to advanced chipmaking tools and designs. However, China has responded with increased investment in domestic semiconductor production and alternative architectures, potentially accelerating a decoupling that neither side fully desires.
Historical Context of US-China Tech Tensions
The summit reflects a long arc of US-China technology competition that predates the current administration. The Trump administration initially imposed tariffs and export controls on Chinese technology firms, citing national security concerns over intellectual property theft and forced technology transfer. The Biden administration largely continued, and in some areas tightened, these measures, particularly around semiconductor and AI technologies. The result has been a layered system of export controls, entity lists, and licensing requirements that have reshaped global supply chains.
China, for its part, has retaliated by restricting exports of rare-earth elements, which are essential for manufacturing high-strength magnets and critical components in electronics, defense systems, and clean energy technologies. The US and other Western nations have sought to diversify their rare-earth sources, but progress has been slow due to the high cost and technical complexity of building new processing facilities.
The Beijing summit was an opportunity for both sides to reset the tone of the bilateral technology relationship, but the lack of any signed agreement suggests that substantive differences remain large. The guardrails discussed were described as “standard” by Trump, but in practice, no common standard exists between the two countries. Future dialogues will need to bridge gaps in definitions, verification mechanisms, and trust if any meaningful framework is to emerge.