U.S. regulators are drafting sweeping export controls that would require licenses for nearly all global shipments of advanced AI chips, potentially slowing worldwide artificial intelligence infrastructure development and giving Washington unprecedented authority over foreign data center construction.
The proposed regulations would expand existing country-based restrictions into a worldwide licensing system, allowing the Trump administration to approve or deny large-scale AI infrastructure buildouts outside American borders. Under the new framework, shipments involving up to 1,000 Nvidia GB300 GPUs would pass through simplified review with possible exemptions, while larger installations would face pre-authorization requirements before export licenses could be issued.
Nvidia shares fell 1.12% to $181.00 on Thursday following reports of the potential policy changes, with competitor AMD dropping 2.3% as investors reacted to uncertainty surrounding stricter oversight of global AI chip exports. The decline came despite billionaire entrepreneur Leo Koguan revealing he purchased one million Nvidia shares worth approximately $180 million earlier this week.
The tiered licensing system classifies exports based on computing scale, with mid-scale deployments requiring "preclearance before seeking export licenses" from the U.S. Department of Commerce.
Clusters powered by 200,000 GB300 GPUs operated by a single company within one country, similar to those currently deployed by AWS, Microsoft, Oracle, OpenAI, or xAI, would trigger direct intergovernmental arrangements requiring national-security assurances and commitments to invest in American AI infrastructure.
Companies like AMD and Nvidia currently cannot ship high-performance AI processors to select countries including China, Iran, North Korea, and Russia. The new rules would extend these restrictions globally while introducing compliance obligations that may include operational transparency, disclosure of business activities, and potential on-site inspections by U.S. authorities.
For cloud providers relying on Nvidia's GPUs to train large-scale models, the regulations could disrupt global AI supply chains and limit expansion capabilities for tech giants like Google and Amazon. Building clusters comparable to those operated by major U.S. cloud providers outside American territory would face extreme complications under the proposed framework.
The policy represents a significant tightening compared to previous restrictions from both administrations, Trump's earlier limitations on AI chip exports and Biden's highly-criticized AI Diffusion Rule, by establishing direct government oversight over commercial technology deployment at scale.
Nvidia reported quarterly revenue of roughly $68 billion earlier this year, representing 73% growth driven largely by continued demand for artificial intelligence infrastructure in its data center segment. Despite regulatory challenges and geopolitical constraints following Broadcom's earnings report highlighting competition in custom AI chips, Wall Street firms including Baird have maintained outperform ratings on Nvidia with price targets reaching $300.
