Trump Weighs AI Chip Export Policy Shift for Nvidia and AMD

By Joseph, Contributor
August 12, 2025 – 3:30 PM CST, Milwaukee, WI

Former U.S. President Donald Trump is reportedly considering a major shift in American technology export policy that could allow semiconductor giants Nvidia and AMD to sell advanced artificial intelligence chips to China. The proposal, which is still in the discussion stage, would represent a significant reversal of recent efforts to limit Beijing’s access to high-performance computing hardware. At the center of the potential deal is a plan to permit sales of restricted processors in exchange for a 15% share of revenue being paid directly to the U.S. government.

The chips in question are among the most advanced AI accelerators currently produced by U.S. companies. Nvidia’s Blackwell-architecture processors, including the H20 model specifically designed to comply with current export restrictions, are considered state-of-the-art in machine learning and data center performance. AMD’s MI308 accelerator, designed for similar high-performance AI workloads, is also part of the policy discussion. Under current rules, both products face significant limitations on their sale to Chinese entities.

Export controls targeting these products were first introduced in 2022, with the aim of curbing China’s ability to develop and train advanced AI models that could have military or surveillance applications. In the years since, those restrictions have been expanded, limiting not just the raw computational power of exported chips, but also their interconnect bandwidth — a key factor in training large-scale neural networks. U.S. officials have repeatedly cited national security concerns as the primary reason for the controls.

The proposed shift would not entirely remove restrictions, but it would allow sales of high-end chips under a regulated framework. The revenue-sharing provision is intended to ensure that U.S. taxpayers directly benefit from the transactions, while also providing funding for domestic research and development in AI and semiconductor manufacturing. Supporters of the plan argue that this approach could strike a balance between protecting security interests and maintaining the competitiveness of American companies in one of the largest AI markets in the world.

Nvidia has previously warned investors that export controls could have a substantial impact on its financial performance, particularly in the data center division, which accounts for the majority of its revenue. China represents a major growth market for AI hardware, driven by the expansion of cloud computing services, autonomous vehicle development, and AI-powered research. By opening the door to more advanced sales, the company could recapture a portion of the business it has lost due to current restrictions.

AMD faces similar challenges. While the company has diversified its customer base, the Chinese market remains a significant opportunity for growth, especially as AI adoption accelerates in both the public and private sectors. Access to high-end AI accelerators like the MI308 could strengthen its position in a competitive industry that is increasingly dominated by Nvidia.

Critics of the proposed policy warn that even limited access to advanced AI hardware could enable Chinese companies to close the technological gap with the United States. They point out that AI model training capabilities are a strategic asset, with potential applications ranging from autonomous weapons to advanced cyber operations. For these opponents, the risks outweigh the potential economic benefits, and they argue that any relaxation of export controls could be exploited in ways that threaten U.S. interests.

Some lawmakers have already signaled resistance to the idea. Members of Congress from both parties have called for a hard line on technology transfers to China, citing concerns over intellectual property theft and the Chinese government’s integration of civilian technology into military projects. In their view, maintaining strict controls is essential to safeguarding the United States’ long-term strategic advantage.

Others take a more pragmatic stance, noting that if U.S. companies cannot sell advanced chips to China, competitors from other countries may fill the gap. While the most sophisticated processors are still primarily produced by U.S.-based firms, other nations are investing heavily in AI hardware manufacturing, potentially eroding America’s market dominance over time. From this perspective, a regulated sales framework could be a better alternative than leaving the field open to foreign rivals.

The revenue-sharing mechanism has also sparked debate. Some economists have questioned whether the 15% figure is high enough to offset potential security risks, while others worry that introducing a direct payment structure could create perverse incentives for future policy decisions. There is also uncertainty over how such funds would be allocated and whether they would indeed be invested in strengthening U.S. AI capabilities.

Nvidia and AMD have not made public statements endorsing or opposing the plan, but industry analysts suggest that both companies would welcome any policy that expands their access to the Chinese market. Share prices for major semiconductor firms have historically responded to news about export controls, reflecting investor sensitivity to policy developments in this area.

The Chinese government has yet to respond to reports of the proposal, but state-affiliated media have previously criticized U.S. export restrictions as discriminatory and harmful to global innovation. Any official reaction will likely depend on the final terms of the policy, should it move forward.

International trade experts note that the proposal comes at a time when U.S.-China relations remain strained, with disputes over tariffs, intellectual property, and geopolitical issues continuing to dominate the bilateral agenda. The question of advanced technology exports has become one of the most sensitive and politically charged topics in that relationship.

The decision, if enacted, would represent one of the most consequential shifts in U.S. technology policy in recent years. It would also test the ability of Washington to balance national security considerations with economic imperatives in an era where AI is increasingly viewed as a critical driver of global power.

Policy discussions are still ongoing, and no formal timeline for a decision has been announced. Insiders suggest that any change would likely require input from multiple federal agencies, including the Department of Commerce, the Department of Defense, and the intelligence community, before it could be implemented.

For now, the proposal remains a subject of speculation and debate. Whether it ultimately becomes policy will depend not only on economic calculations but also on the willingness of political leaders to accept the potential risks of expanding China’s access to advanced AI hardware. The outcome will have far-reaching implications for the semiconductor industry, U.S.-China relations, and the future of global AI development.