A bipartisan coalition in the U.S. Senate introduced new legislation this week aimed at tightening and codifying current restrictions on exporting advanced artificial intelligence semiconductors to U.S. rivals, primarily China, preempting any potential attempts by the executive branch to soften existing controls. The action, spearheaded by Republican Senator Pete Ricketts and Democratic Senator Chris Coons, seeks to cement the technological embargo against adversaries and preserve America’s lead in AI innovation.
The proposed measure, titled the Secure and Feasible Exports (SAFE) CHIPS Act, mandates that the Department of Commerce deny export license applications for U.S.-made AI processors deemed more advanced than current permissible configurations destined for organizations in China, Russia, Iran, or North Korea. This restriction would remain in force for a period of 30 months. Critically, the bill stipulates that future efforts to alter these export control regulations would require a minimum 30-day notice and briefing to Congress before implementation, ensuring legislative oversight.
Preserving Technological Superiority
The effort comes amid heightened advocacy, including closed-door meetings between Nvidia CEO Jensen Huang and Republican senators involved in key export control oversight. Supporters of the bill argue it is paramount to national security, citing concerns that cutting-edge American technology could benefit the military and surveillance capabilities of strategic competitors.
Senator Ricketts emphasized that denying Beijing access to top-tier AI chips is essential for preserving the nation’s technological advantage. Separately, hawkish voices, including Senator Tom Cotton, have accused platforms with Chinese affiliations of unfairly leveraging U.S. technology. The legislation acts as a direct response to concerns that the Trump administration might ease limitations initially imposed in 2022 to restrict China’s access to powerful U.S. data center hardware.
The SAFE CHIPS Act effectively seeks to codify existing curbs, specifically naming high-end specialized processors such as Nvidia’s H200 and forthcoming Blackwell-architecture chips, extending these restrictions to cover competitive offerings from companies like Advanced Micro Devices (AMD) and certain Google-designed accelerators.
The Debate Over Export Strategy
The legislative push arrives as the administration continues to deliberate on whether to authorize sales of constrained variants of powerful chips to Chinese entities. Analysis by Georgetown University’s Center for Security and Emerging Technology previously estimated that the processing power of chips like the H200 vastly exceeds the technical thresholds defined under existing 2023 regulations.
Nvidia, which successfully secured approval earlier this year to export its less powerful H20 chip—designed to sit just below existing regulatory limits—has voiced complex considerations regarding the export strategy. A spokesperson for the company suggested that the broader goal should allow non-military firms globally to adopt an American technology stack, thereby supporting U.S. jobs and security. CEO Jensen Huang has also publicly questioned the market viability of significantly degraded chips, suggesting Chinese customers would likely reject such offerings in favor of developing local alternatives.
Lawmakers on both sides of the aisle contend that the proposed statutory guardrails are necessary. Beyond national security implications, they argue the restrictions ensure U.S. businesses and academic institutions maintain priority access to the most state-of-the-art AI hardware, preventing the leakage of strategic capabilities to geopolitical adversaries and protecting America’s position as the global leader in AI development.
The introduction of the SAFE CHIPS Act highlights the current U.S. policy imperative to reconcile economic interests with national security concerns in the rapidly evolving landscape of artificial intelligence and semiconductor technology.