America Introduces Strict Export Regulations for Semiconductors
================================================================
The Biden administration has implemented a new framework for exporting U.S.-designed computer chips, particularly advanced AI chips like Nvidia's H20 and AMD's MI308. This move is part of the AI Diffusion Rule, introduced by the U.S. Bureau of Industry and Security (BIS) on January 15, 2025.
The AI Diffusion Rule establishes a tiered, risk-based global system that regulates exports, re-exports, and transfers of advanced AI model weights and high-performance computing chips. Countries are categorised into three tiers based on their risk levels and the sensitivity of the technology involved.
Tier 1 includes the U.S. and its key allies, such as Australia, Japan, South Korea, and Taiwan, which face minimal restrictions. Tier 2 comprises 120 nations, including Singapore, Israel, Saudi Arabia, and the UAE, and they will encounter limits on the amount of computing power they can purchase. Tier 3 imposes a complete ban on receiving the technology for Russia, China, and Iran.
One of the most significant impacts of these new rules is the cap on GPU exports at 790 million TPP until 2027, roughly equivalent to 50,000 Nvidia H100 chips. The restrictions are aimed at denying other countries, particularly China, access to advanced technologies, including AI solutions.
Entities with "Verified End-User" status are limited to around 320,000 advanced GPUs over the next two years. However, in April 2025, the U.S. government abruptly extended license requirements on these chips’ exports to China “indefinitely,” causing Nvidia a significant financial charge, resulting in a $5.5 billion quarterly charge.
Following high-level lobbying, the Biden administration agreed to resume licensing for Nvidia’s H20 GPU exports to China, which Nvidia confirmed would resume soon. This partial reversal highlights the tangible business impact of these restrictions.
The rules also introduce mandatory licenses for certain chips and new controls in allied countries like Malaysia, aimed at preventing chip diversion and third-country transshipment used to bypass U.S. export controls.
Tech giants, including Nvidia and Oracle, have expressed concern about these export controls and the volatile trade tensions with China. They argue for predictable, consistent rules amid shifting policy landscapes.
Elon Musk has called the Colossus training cluster, launched by the AI startup xAI using 100,000 Nvidia H100 GPUs, the most powerful AI training system in the world. Meanwhile, some companies like Amazon Web Services or Microsoft are exempt from the new restrictions because they meet specific licensing requirements.
The Biden administration has also introduced new restrictions on importing U.S.-designed computer chips, and the 120-day comment period for the new rules has been opened. Tech giants like Nvidia and Oracle have criticised these measures, stating they threaten to disrupt innovation and economic growth worldwide.
Oracle's Executive Vice President, Ken Glueck, argued that the changes amount to overregulation rather than safeguarding the interests of the U.S., its partners, and allies. Earlier, Chinese military-affiliated institutes used Meta's publicly available AI model to develop a tool potentially suited for military applications. China has also initiated an antitrust investigation into Nvidia.
These measures and responses illustrate the Biden administration’s strategic, security-driven approach to regulating exports of advanced AI-enabled semiconductor technology, met with significant concern and active engagement by major U.S. tech companies like Nvidia.
- The recent AI Diffusion Rule, enacted by the Biden administration, impacts the technology sector significantly, particularly companies specializing in data-and-cloud-computing like Nvidia and Oracle.
- The new framework for exporting US-designed computer chips includes financial implications for technology companies, such as Nvidia, as shown by the $5.5 billion quarterly charge incurred due to the extension of license requirements on chip exports to China.
