President Donald Trump has lifted export restrictions on Nvidia’s H200 AI chips to “approved customers” in China.
The move comes at a time when the demand for high-performance processors has skyrocketed in anticipation of an AI revolution.
The ban of some chip models produced by American manufacturer Nvidia begun with the administration of Joe Biden and the Trump administration tightened the noose in a political game that aims to control, at least in part, who shines the most in the AI revolution competition.
The previous administration initiated stringent export controls to limit China’s access to the most advanced AI semiconductors. These chips are the essential computational blocks for training state-of-the-art Large Language Models (LLMs) and power advanced military applications among other futuristic uses.
- The Original Ban: This policy was designed to prevent China from acquiring chips like Nvidia’s A100 and H100, which are crucial for frontier AI development.
- The “Compliant” Chip: In response, Nvidia developed less-powerful, “export-compliant” chips like the H20 specifically for the Chinese market.
- The April Freeze and August Thaw: The Trump administration initially banned even the H20 in April 2025, only to reverse course in August, allowing its sale on the The H200 is estimated to be nearly six times more powerful than the H20, drastically narrowing the performance gap between what China can legally buy and what is currently restricted.
💰 The Current State: The 25% Fee and the Smuggling Sting
President Trump used his social media platform, Truth Social to make the announcement, stating that the US would receive a 25 percent cut of the revenues from these sales.
Analyts claim that the lift of this ban is a balance on economics and national interest, on one hand : maintaining Nvidia’s global market share, funding US technology, and ensuring the next generation of Chinese AI runs on US-made hardware.
- Political Win for Nvidia: The move is a significant victory for Nvidia CEO Jensen Huang, who has reportedly lobbied the administration heavily to ease restrictions, arguing that market access supports American jobs and R&D.
- The Revenue Share Debate: The 25% revenue share, collected as an “import tax” on chips being re-exported from the US (via Taiwan) to China, has drawn sharp criticism, with some experts questioning its legality under constitutional prohibitions on taxes on exports.
- The Smuggling Irony: In a spectacular display of geopolitical irony, the announcement came on the same day the Department of Justice announced the arrest of two Chinese nationals for their involvement in a sophisticated smuggling operation.
Even with the US green light, Chinese regulators are reportedly considering limiting access to the H200, possibly requiring buyers to prove domestic alternatives cannot meet their needs. Beijing’s overarching goal remains semiconductor independence, and this sudden policy reversal might reinforce their push to boost domestic champions like Huawei, whose Ascend chips are rapidly improving.
Critics, including some lawmakers, worry that this decision hands China a massive tool to accelerate its own AI development, potentially reducing the US’s technological lead. The H200, while not Nvidia’s latest (Blackwell remains restricted), is powerful enough to train highly capable models and support cloud computing services that could compete globally.
The deal establishes a new and highly transactional precedent: the US is willing to trade restricted access to advanced technology for substantial revenue. This shift from a pure national security posture to a revenue-sharing model fundamentally shifts the calculus of export controls. It’s a bold gamble that prioritizes near-term economic benefit and US-led ecosystem dominance over total containment, risking the long-term strategic advantage in the AI arms race.
This high-stakes maneuver means the AI race is no longer a simple sprint but a complex high-stake negotiation, with the H200 now the most valuable chip on the table.