What are the New Federal Regulations on the Export and Flow of American AI Chips?

To maintain global leadership and protect national security, the United States government has implemented a comprehensive framework of regulations governing the export and flow of advanced American artificial intelligence (AI) chips. These rules are designed to prevent foreign adversaries from acquiring the computational power necessary to develop advanced military systems, mass surveillance networks, and next-generation cyber warfare capabilities.

The regulations represent a significant shift in global technology trade. Rather than focusing solely on traditional weapons or munitions, the federal government now classifies high-performance semiconductors and the infrastructure required to run them as critical national security assets. This framework impacts not only the physical shipment of hardware but also the digital access to computing power.

Core Objectives of the Regulations

The federal restrictions are built around several strategic goals intended to secure the domestic technology ecosystem:

• National Security: Preventing foreign military and intelligence apparatuses from leveraging American-designed AI hardware to modernize their defense systems or develop autonomous weaponry.
• Technological Containment: Slowing the advancement of rival AI ecosystems by restricting access to the most efficient and powerful training hardware available on the market.
• Supply Chain Integrity: Closing loopholes that previously allowed restricted nations to acquire advanced chips through shell companies or third-party intermediary countries.

Key Components of the Restrictions

The regulatory framework targets multiple layers of the AI supply chain, from manufacturing to end-user deployment.

• Performance Thresholds: The government has established strict technical limits based on a chip’s processing speed and memory bandwidth. Starting with rules issued in October 2022, any semiconductor exceeding defined computational thresholds — such as a processing performance rate of 4,800 or more tera operations per second (TOPS) — requires a highly scrutinized, and often denied, federal license for export to designated countries of concern. These thresholds have been updated in subsequent rulemakings, including a significant update published by the Bureau of Industry and Security (BIS) in January 2025.
• Cloud Computing and IaaS Rules: To prevent foreign entities from bypassing hardware bans by renting computing power digitally, Infrastructure-as-a-Service (IaaS) providers are subject to proposed Know Your Customer (KYC) requirements. In January 2024, BIS issued a proposed rule that would require U.S. IaaS providers and their foreign resellers to implement and maintain detailed customer identification and verification programs, aimed at preventing foreign actors from using American cloud infrastructure to train large AI models.
• Manufacturing Equipment Controls: The regulations restrict the export of the specialized machinery, chemical materials, and software required to fabricate advanced semiconductors, preventing restricted nations from building their own cutting-edge foundries.
• Extraterritorial Reach: Through the Foreign Direct Product Rule (FDPR), the controls apply not only to chips manufactured within the United States but also to chips produced anywhere in the world that utilize American software, tools, or intellectual property in their design or fabrication.

Impact on the Technology Sector

These export controls have fundamentally altered how semiconductor manufacturers and AI companies operate globally.

• Hardware Redesign: To maintain access to international markets without violating federal law, U.S. chipmakers have engineered specific, lower-performance variants of their flagship AI chips that fall just below the restricted performance thresholds. NVIDIA, for example, developed the A800 and H800 chips as compliant alternatives to its restricted A100 and H100 GPUs for the Chinese market.
• Increased Compliance Burdens: Technology hardware vendors, distributors, and cloud service providers must invest heavily in compliance infrastructure to track the exact end-use and end-user of their products.
• Market Bifurcation: The regulations are accelerating a divide in the global technology landscape. While the U.S. and its allies heavily subsidize domestic chip manufacturing, restricted nations are aggressively funding indigenous semiconductor research to reduce their reliance on American technology.

Summary

The federal regulations on American AI chips function as a broad technology control framework designed to protect U.S. national security and technological leadership. By strictly controlling the export of high-performance physical chips, proposing requirements around access to cloud-based computing power, and limiting the distribution of manufacturing equipment, the government aims to keep the most powerful AI infrastructure out of the hands of foreign adversaries.