Highly advanced artificial intelligence (AI) systems are becoming increasingly integrated into critical aspects of society, leading governments across the globe to weigh the associated risks and opportunities. IST has been actively engaged in studying AI developments since 2017, beginning with the integration of AI into international security. In 2025, our research has expanded to include developing technical and policy-oriented risk reduction strategies, building a framework for a multi-agency AI chip export controls enforcement program, exploring how to account for national security considerations in AI antitrust cases, and more. This month, Lillian Ilsley-Greene sat down with IST’s Jennifer Tang, Gabrielle Tran, and Fatima Faisal Khan to learn more about these initiatives and how they’re tackling the ever-changing AI landscape.
Q&A: AI’s Global Implications
In August, U.S.-based chipmakers Nvidia and Advanced Micro Devices (AMD) reached an agreement with the U.S. government that allows the companies to continue selling certain AI chips in China, on the condition that they pay 15% of these revenues to the U.S. government. What are the implications of this arrangement for U.S. competitiveness and U.S.-China competition?
Jennifer Tang: “This arrangement marks a notable shift in U.S. export control strategy. Earlier restrictions aimed to block China’s access to advanced processors entirely, both to slow military AI development and to preserve U.S. technological leadership. The new approach allows continued sales, but with a significant revenue share to Washington—effectively trading strict denial for a managed flow of chips coupled with economic capture. It reflects recognition that a full cutoff is commercially costly and strategically difficult to sustain.
For China, the deal offers breathing room. Even performance-limited chips, such as Nvidia’s H20, remain powerful enough to drive large-scale AI training, and Chinese firms have shown skill in optimizing around hardware constraints. Continued access—albeit a step behind the frontier—may ease some of the intended pressure and buy time for domestic alternatives to further mature and scale.
For the United States, the risks are twofold. First, framing export controls as negotiable could weaken their credibility with allies and competitors alike. Second, some market analysts predict that this dynamic may lead to thinner margins on sales to China and could reduce U.S. firms’ reinvestment in next-generation R&D. The central question now is how to balance near-term commercial realities with long-term national security priorities. Our AI Chip Export Control Initiative is examining how a more coordinated multi-agency and allied approach can reinforce controls, limit leakage, and protect U.S. technological leadership while avoiding unintended consequences.”
Prior to the U.S. v Google ruling this spring, Google told the DOJ that requiring the company to sell the Chrome browser and limit AI investments could harm U.S. national security. Now, the courts are days away from a final ruling. What does Google’s appeal to national security in its antitrust defense mean for AI labs?
Gabrielle Tran: “However one weighs Google’s invocation of national security, it is important to remember that antitrust law is fundamentally about market competition and consumer welfare—not national security. Courts and regulators have consistently treated security concerns as secondary unless they directly bear on competition. In areas such as foreign investment reviews and mergers, other bodies like the Committee on Foreign Investment in the United States (CFIUS) are better equipped to assess national security implications.
That said, the Google case highlights a trend worth noting: as AI research and development with national security implications is largely led by the private sector, companies facing antitrust scrutiny are increasingly likely to raise security arguments in their defense. This is precisely the space IST’s AI Antitrust and National Security effort is examining. How should policymakers and courts respond when such claims are raised? How might antitrust remedies imposed on AI labs intersect with security risks? And how can these risks be addressed while respecting the precedent and principles of competition law? Our project does not seek to change antitrust doctrine or argue for one approach over another. Instead, it explores how remedies can be designed to preserve competition while responsibly accounting for national security considerations. In this way, IST aims to place Google’s arguments—and similar claims likely to arise in future AI cases—into their proper context.”
In March, the People’s Republic of China (PRC) announced the creation of a national venture capital guidance fund, a public-private partnership to drive investment towards “hard” technology, sectors such as semiconductors and AI applications. If Beijing is guiding venture capital on a national scale, what should the United States do to maintain a competitive edge?
Fatima Faisal Khan: “The United States’ strength has always been private venture capital that moves quickly, takes risks, and drives innovation. The challenge is that many of the technologies most important for national security – such as semiconductors and advanced manufacturing – require long timelines and large amounts of capital. These deep-tech fields are the foundation that even AI depends on, yet they often struggle to attract investors who prefer faster returns.
This is where policy can help. The U.S. government can make it easier and safer for investors to back longer-term technologies, particularly those with dual-uses, by sharing some of the risk. For example, it could set up programs that match private investment in areas like advanced packaging or chip design tools, or offer temporary tax benefits for funds that specialize in deep tech. It can also expand access to federal test facilities so start-ups can prove their products will work at scale, which makes investors more confident. The recent CHIPS Act funding for advanced packaging is a good step in this direction, because it sends a clear signal that the government is backing the parts of the industry that are hardest to finance. Just as important is keeping capital at home. Guardrails such as the Treasury Department’s new outbound investment rule also play a role here by limiting U.S. money from flowing into sensitive Chinese sectors.
This is exactly what our deep-tech investment project is studying: how state-guided capital in the PRC intersects with U.S. chokepoints, and how smart incentives at home can ensure American innovation keeps its edge. The effort builds on IST’s broader efforts centered on geopolitical and technological competition between the two countries. Last year, IST fellows Pavneet Singh and Michael Brown released a report on the role of venture capital in U.S. industrial strategy, and our forthcoming report will continue that conversation.”
