Digital Borders: The Biden Administration’s Final AI Rule

In the final stretch of President Joe Biden’s tenure, the administration unveiled the “Regulatory Framework for the Responsible Diffusion of Advanced Artificial Intelligence Technology.” Announced by the Department of Commerce’s Bureau of Industry and Security (BIS), the framework introduces comprehensive new controls on advanced computing chips and AI model weights. These controls include licensing requirements for exports, reexports, and in-country transfers to a broad range of nations, with exceptions for trusted allies and partners to ensure uninterrupted commercial and research activities.

It’s a daring gambit but one that raises urgent questions. Domestically, how will the incoming Trump team respond to this last-minute regulation as they devise their own tech strategy, walking the delicate line between containing China’s technological rise and supporting the business community? On the international stage, whether it stays under Trump or not, this rule will shape the future of global tech partnerships. At its core, the U.S. strategy is unmistakable: not merely to maintain but to expand its dominance in AI computing—the backbone of AI models—while keeping China relegated to the technological sidelines for the foreseeable future.

With the United States controlling over 70 percent of the world’s advanced AI compute power, Washington holds a commanding advantage in the race for AI supremacy. The newly announced export controls are designed to prevent China from narrowing this gap. 

By dividing the global market into three tiers—Tier 1, consisting of U.S.-aligned nations such as the Five Eyes members, key Western European allies, and AI ecosystem leaders like South Korea, Japan, and Taiwan (notably excluding the entirety of the European Union); Tier 2, a “gray” list that includes countries like India, Israel, Saudi Arabia, the UAE, and much of the rest of the world; and Tier 3, an adversarial bloc led by China and Russia—the United States is actively shaping distinct techno-economic spheres of influence. 

What was once a theoretical framework for techno-economic coalitions and alliances, often discussed in policy briefs and think tank hallways in D.C., has now become a tangible reality with profound financial, technological, security, and, most importantly, geopolitical consequences.

The framework updates the Data Center Validated End User (VEU) program, bifurcating it into Universal VEUs (UVEU) and National VEUs (NVEU). UVEUs, designed for U.S. and allied entities, allow global data center expansion while maintaining stringent restrictions, such as retaining at least 75 percent of advanced chips within allied nations. NVEUs, tailored for foreign entities in non-embargoed countries, permit data center development in specified locations under strict monitoring. These measures aim to secure the storage and use of advanced AI technologies, preventing the diversion of chips and models for unauthorized purposes. The framework builds on examples such as the Saudi PIF-Google partnership and the UAE’s G42-Microsoft deal, where the U.S. government either received advance notification of the details (as with PIF) or played an active role in shaping the deal’s structure (as with G42). Advanced AI model weights, particularly those trained with 10^26 computational operations or more, are now subject to direct controls, with license exceptions for U.S.-headquartered companies and allies deploying less sensitive open-weight models. The 10^26 threshold marks AI models with immense power, critical for advanced tasks in defense and research, while also excluding smaller models from unnecessary safety testing.

However, the rule’s implementation isn’t without resistance. U.S. giants like NVIDIA, AMD, and the broader industry are gearing up to push back, fully aware of the devastating impact the restrictions will have on their ability to access global markets. As countries around the world race to build large GPU clusters for training AI models, these companies face an existential choice: comply with the new restrictions and lose access to key international markets, or challenge the rule to protect their financial interests and retain their dominance.

The Trump administration will face the challenge of deciding whether to retain the regulation while amending it to achieve a delicate balance: addressing industry concerns about maintaining market share in global markets while preserving a strong strategic posture in the technological competition with China. This could include expanding the list of Tier 1 countries to incorporate other key partners previously excluded, contingent on obtaining necessary assurances to protect U.S. technology. If the administration opts to discard the Biden-era AI rule, it will still need to devise its own framework to determine which nations can access advanced AI chips, ensuring these technologies do not inadvertently fall into the hands of adversaries such as Russia, China, or North Korea.

Export controls have emerged as the defining tool for shaping techno-economic spheres of influence in an era where technology determines global power. The stakes could not be higher—how these controls are wielded will decide whether the United States solidifies its position as the architect of a new global tech order or risks forfeiting its dominance in what is undoubtedly the defining race of the twenty-first century.

Mohammed Soliman is the Director of the Strategic Technologies and Cyber Security Program at the Middle East Institute, a member of McLarty Associates, and a Non-Resident Senior Fellow at the Foreign Policy Research Institute. Follow him on X: @ThisIsSoliman.

Image: William Potter / Shutterstock.com.