In July, the Trump administration announced its highly-anticipated AI Action Plan, which sets out a guide for achieving U.S. AI dominance by propelling innovation and adoption, easing regulation, and promoting the country’s tech stack abroad. Just three days later, China released its own vision, the Action Plan on Global Governance of Artificial Intelligence. In stark contrast to Washington’s competitive framing, Beijing emphasized collaboration, inclusivity, and international dialogue on AI standards.
The two divergent action plans highlight the widening rift between Washington and Beijing in their race for AI supremacy. This rivalry raises urgent questions for global stability, including how reciprocal tariffs are reshaping the industry, what risks arise from concentrated AI chip supply chains, and why Taiwan remains precariously at the center of the contest.
What Do Each of the Action Plans Entail?
Released days apart, the two action plans establish their own strategies for guiding AI development and adoption. The U.S. plan seeks to “achieve global dominance” and accelerate AI innovation by supporting private sector-led initiatives, building out AI infrastructure and meeting growing energy demands, and driving international adoption of U.S. AI systems while preventing “adversaries” from monopolizing U.S. AI innovation and investment.
The U.S. plan makes reference to diplomacy; however, the substance of its contents is far from diplomatic. The overt references to China as an adversary, and a point of comparison for progress, reflect the broader tensions between these two nations.
China’s action plan takes a broader approach by presenting thirteen overarching recommendations on supporting global AI governance, including promoting digital infrastructure construction and establishing forums for international dialogue. While the U.S. plan is overtly nationalistic, its Chinese counterpart is positioned as a diplomatic, globally-focused agenda centered on the goals and principles of supporting the Global South, respecting sovereignty, and advancing safety and control, fairness and inclusiveness, and open cooperation. The two plans nevertheless share several similarities and a number of distinct differences across the common themes of infrastructure, energy, regulation, and AI security.
Infrastructure
What the two action plans have in common is a throughline of supporting AI innovation and adoption and advocating for increased infrastructure to support the growth of the AI industry. Both plans clearly articulate the need to accelerate and scale up infrastructure development with differing approaches on how this can be achieved. Action Four of China’s plan emphasizes the importance of constructing “next-generation networks” and establishing a “unified computing power standard system” to support the Global South and promote AI development in an inclusive and universally-beneficial manner. In comparison, Pillar II of the U.S. plan outlines the need for building infrastructure and revitalizing domestic semiconductor manufacturing to power the country’s own AI future.
Since the release of America’s action plan, V Gallant Limited, a subsidiary of VCI Global Limited, has secured a $22 million sovereign AI infrastructure contract with Codetext, a software engineering and AI deployment company. The contract includes the delivery of a full-stack sovereign AI system that is intended to help Codetext ensure they can operate independently of external cloud providers.
Energy
While the United States and China both call for the acceleration of constructing infrastructure required to power the AI industry, the significant environmental impact of this infrastructure has been a divisive topic. The water and power needs of data centers are astronomically high, representing 1-1.5 percent of electricity use globally and consuming up to 6 percent of district water supplies, in the U.S. regions they are located, during peak periods. China, which has a far greater energy capacity than the United States, has attempted to avoid disrupting electricity grids and water supplies by relying on hydropower and nuclear technologies to securely and inexpensively support the energy needs of large data centers.
The U.S. AI action plan points to China’s superior energy capacity to justify similar growth while dismissing environmental concerns. The Plan, which praises the “momentum” built by the Trump administration’s reforms to the National Environmental Policy Act (NEPA), recommends further removal of regulatory red tape to spur infrastructure development at the speed required for the United States to “win the race.”
Comparatively, China’s plan emphasizes “sustainable” AI development that explores more resource-efficient models and chips and establishes joint AI energy and water efficiency standards. The notion of AI “sustainability” has been criticized as a paradox, as calls for phasing out fossil fuels and “computing within limits” directly oppose the speed and scale currently being pursued. The paradoxical tension between AI sustainability and the AI race may inhibit any momentum towards sustainability goals.
Regulation
At the beginning of his term, U.S. President Donald Trump publicly revoked his predecessor’s executive order on AI. The U.S. AI action plan advances this deregulatory approach by proposing the removal of various regulations deemed onerous and inhibiting the private sector’s efforts to further develop the American AI industry. This includes prohibiting AI-related Federal funding for states with “burdensome” AI regulations.
Many other countries are also adopting a deregulatory mindset, echoing unfounded claims that regulations will hinder AI innovation. Regulations have a demonstrated history of guiding safe innovation. For example, Uber’s success is often attributed to the company’s ability to adapt its business model with regulations to legitimize itself against local competition. The move away from regulation is a sharp pivot from the AI safety pathways many countries have been carving out over the last few years. The United Kingdom, for example, rebranded its AI Safety Institute to the AI Security Institute following the Paris AI Action Summit earlier this year.
China’s approach, by contrast, has been to promote a “common understanding of standards and norms” among standard-setting bodies of states and to advance the governance of AI safety by establishing a “widely recognized safety governance framework.” While one comment refers to fostering “innovation-friendly” policy, the plan’s overall tone is not in favor of deregulation.
AI Security
The U.S. plan warns of the potential for adversaries — namely China — to exploit its AI industry. It specifically alleges Chinese interference in AI regulations and “codes of conduct” proposed by international bodies, such as the United Nations, that run counter to “American values.” China’s plan notably avoids mentioning the United States, or any other country. It also does not reference sovereign security within the scope of its aims.
The U.S. plan also highlights the importance of securing AI manufacturing supply chains to bolster the country’s position in the global AI race. As this supply chain has been an increasing point of tension between Washington and Beijing, the bold and unflinching acknowledgement of that monopoly — accompanied by the call to further restrict that pipeline with strengthened export controls — adds fuel to the fire.
The Trump administration placed specific restrictions on exporting the most advanced AI chips to China. These chips, designed by the company Nvidia in the United States and manufactured in Taiwan, offer greater speed and efficiency than traditional central processing units (CPUs), the primary functional component of a computer.
In a recent turn of events, this stance was surprisingly softened, allowing Nvidia to resume sales of its H20 chips designed specifically for the Chinese market. This is in direct contrast with their action plan which pushes for strengthening export controls to “plug loopholes” in the semiconductor manufacturing pipeline. But China has claimed the chips pose a security risk with alleged “tracking and positioning” and “remote shutdown” features, and discouraged companies from using them.
What Comes Next
The U.S. and Chinese action plans set out competing roadmaps that will shape the global AI environment for decades to come. Their contrasting approaches to governance cannot be separated from the economic and geopolitical battlefield on which they play out. Tariffs, export controls, and supply chain dependencies reveal how Washington’s pursuit of “dominance” and Beijing’s calls for “cooperation” are both entangled with material leverage over AI chips and compute power.
First, intensifying fragmentation. With Washington doubling down on dominance and Beijing positioning itself as the champion of inclusivity, international AI governance risks splintering into rival blocs. Countries in the Global South may be courted by both sides, but they will face growing pressure to align with one over the other.
Second, hardware will remain a chokepoint. The U.S. has paired sweeping tariffs with restrictions on exporting advanced AI chips to China, citing national security risks. This highly concentrated pipeline of AI chips designed largely by U.S. firms like Nvidia and manufactured in Taiwan, remains a bottleneck for Beijing. The experience of DeepSeek, a Chinese AI company that delayed releasing its latest model after struggling to replace Nvidia chips with Huawei’s, illustrates how U.S. control still constrains China in the short term, even as Beijing invests heavily in alternatives.
Third, a global AI governance vacuum looms. While China emphasizes cooperation and the U.S. warns of adversaries, neither plan outlines a credible mechanism for bridging differences. Without trusted multilateral frameworks, the risk grows that AI governance will follow the path of internet governance — fragmented, contested, and prone to great-power rivalry.
Finally, Taiwan remains the most precarious fault line. As home to TSMC, which manufactures 90 percent of the world’s most advanced chips, the island is both an economic linchpin and a geopolitical flashpoint. Beijing has escalated military pressure, while Washington’s stance grows increasingly ambiguous. Taiwan’s position at the center of AI supply chains ensures the global race for AI leadership is inseparable from one of the world’s most volatile territorial disputes.
For allies, partners, and neutral states, the question is not simply which plan to endorse, but how to preserve room for genuine multilateral cooperation while navigating an intensifying contest between two divergent strategies for global AI leadership.
