An Interesting Case of Nvidia in China. The Chip Paradox of the Year?

Donald Trump’s decision to allow the export of Nvidia H200 chips to China, subject to a 25% tariff, has sparked a strong reaction on both sides of the Pacific. While formally it represents a partial opening to the world’s largest semiconductor market, Chinese authorities have decided to hold shipments at the border. For financial markets, this is a clear signal that the technological rivalry between the U.S. and China has entered a phase of heightened uncertainty, where administrative decisions increasingly have a direct impact on global supply chains and technology company valuations.

From Nvidia’s perspective, the stakes are contracts worth tens of billions of dollars. Chinese technology firms, responsible for roughly 30% of global demand for AI infrastructure, reportedly placed orders exceeding 2 million H200 chips, while the company’s available supply is currently estimated at around 1 million units. This means that declared demand significantly exceeds current production capacity. The new H200 chip offers up to six times the computing power of the H20, which was previously designed as a version compliant with export restrictions. Despite this technological advantage, Chinese customs authorities were instructed that the chips cannot be imported.

From a market perspective, the dispute over the H200 highlights a growing paradox in Nvidia’s valuation. The company benefits from unprecedented global demand for AI computing power, yet an increasing portion of its story depends on political decisions rather than purely operational fundamentals. Any signal of a potential easing or tightening of restrictions on China immediately affects sentiment toward the tech giant’s stock and the broader basket of AI-related companies. As a result, the regulatory risk premium rises, and traditional valuation models increasingly need to incorporate geopolitical scenarios, not just revenue and margin trajectories.

The dispute has also quickly moved into the realm of U.S. domestic politics. Pressure is building in Congress to increase oversight of exports of advanced AI technologies to China, reflected in the ongoing work on the AI Overwatch Act. Supporters of the regulation warn that selling chips of this class could strengthen China’s technological capabilities in strategic areas of artificial intelligence. Opponents, however, caution that excessive legislative intervention could weaken the competitiveness of U.S. semiconductor manufacturers and limit their operational flexibility globally.

For the broader technology market, the H200 conflict may serve as another signal to selectively reduce exposure to China and shift capital toward jurisdictions seen as more regulatory stable. Institutional investors have been factoring in the risk of gradual decoupling in the semiconductor and AI infrastructure sectors for several years, and the Nvidia case shows that even seemingly compromise solutions, such as tariffs or volume limits, can in practice lead to situations where the product does not reach the end customer for reasons unrelated to actual demand.

In this context, Nvidia CEO Jensen Huang’s January visit to Shanghai takes on significance beyond business symbolism. The presence of the company’s leader, whose products today account for a substantial portion of global computing power used in AI development, underscores the nature of the new technological confrontation. Its essence is no longer solely about tariffs and sanctions, but about control over algorithms, access to data, and the direction of technological development, which increasingly determines the global balance of power.