#13 China Scholar Insights: China-U.S. AI Competition
AI has emerged as a pivotal factor in 21st-century global geopolitics, with the primary great-power competition centered on the U.S. and China.
Welcome to the 13th edition of China Scholar Insights!
China Scholar Insights is a feature aiming to provide you with the latest analysis on issues that Chinese scholars and strategic communities are focusing on. We carefully select commentary and highlight key insights. Questions and feedback can be directed to sch0625@gmail.com
I’m SUN Chenghao, a fellow with the Center for International Security and Strategy (CISS) at Tsinghua University. ChinAffairsplus is a newsletter that shares Chinese academic articles focused on topics such as China’s foreign policy, China-U.S. relations, China-Europe relations, and more. This newsletter was co-founded by my research assistant, ZHANG Xueyu, and me.
China Scholar Insights on China-U.S. AI Competition
Background
Since the China-U.S. Intergovernmental Dialogue on Artificial Intelligence in May 2024, the AI domain has undergone significant transformation. The past year witnessed the emergence of pivotal advancements, such as DeepSeek, which incited global discourse. Concurrently, May 2025 saw the implementation of two key U.S. AI policy adjustments: a relaxation of chip export regulations, followed by the imposition of stricter limitations on Huawei’s Ascend chips. These contradictory signals, emerging amidst a period of reduced trade tensions, have amplified uncertainty, underscoring AI’s dual role as a potential collaborative frontier and a focal point of strategic competition.
Summary
AI has emerged as a pivotal factor in 21st-century global geopolitics, with the primary great-power competition centered on the US and China. In this AI rivalry, the US is perceived as employing export controls and standard-setting to constrain China’s technological advancement, thereby reinforcing its global dominance and diminishing China’s influence. Restrictions on semiconductors have already drawn criticism from US tech firms. US containment strategies are evolving, expanding from military applications to civilian sectors, with a growing focus on foundational technologies. This external pressure has spurred significant innovation within China’s AI sector, accelerating its progress toward international benchmarks. Scholars highlight developments such as Huawei’s chip advancements, as well as China’s advantages in national coordination and demographic scale.
Insights
XIAO Qian: Can China and U.S. Rebuild Trust Around AI？
The Urgency of Cooperation
AI’s rapid development has become a new engine for economic growth and a key area of global competition. As major powers in AI, both China and the U.S. face growing pressure to create effective governance rules to manage emerging risks. The launch of DeepSeek this year has boosted China’s confidence in overcoming barriers imposed by embargoes on U.S. tech. Meanwhile, the direction of AI policy under “Trump 2.0” remains unpredictable. Still, cooperation between the two countries holds strategic value—akin to U.S.-Soviet nuclear arms control efforts in the 1970s.
The Trust Deficit Between China and the U.S.
In AI governance, “trust” is a frequently cited concept—trust in machines, in human-machine collaboration, in users, in institutions, and in nations. Yet today, trust between humans themselves is eroding, and the basic norms of the international order are under strain. A deep trust deficit also exists between China and the U.S. As China has risen in the 21st century, U.S. policymakers have increasingly viewed it with suspicion and fear. China, in turn, sees U.S. technological restrictions as an attempt to deny its right to develop. Against the backdrop of great-power rivalry and geopolitical tension, uncertainty about each other’s intentions has become the core driver of mistrust. With Trump’s return, predictability has also become a scarce resource in U.S.-China relations.
U.S.-China Rivalry Under the Shadow of Technological Risk
Despite rapid advances in recent years, AI technology still faces unresolved challenges such as uncertainty, algorithmic opacity, and lack of explainability. While many countries are enacting domestic laws to regulate AI, there remains no unified or binding international legal framework governing AI development. If China and the U.S. continue to build separate AI governance systems, the world risks falling into a form of “high-tech feudalism”. In 2024, there were some signs of cooperation—such as the U.S.-China intergovernmental dialogue on AI and the leaders’ meeting in Lima in November. However, the U.S. has also introduced a series of AI-related export controls targeting China, reflecting the persistent difficulty for both countries in escaping zero-sum dynamics and building genuine mutual trust.
The China and U.S. Must Move Beyond the Security Dilemma
To cooperate on AI governance, China and the United States must break free from the “security dilemma.” In the current climate, rebuilding basic trust is the only viable path forward—requiring both sides to take concrete, incremental steps.
First, both countries need to reassess how they perceive and define each other. Even if competition is inevitable, they should avoid treating their relationship as a zero-sum game. Second, clear national security red lines must be drawn to define non-negotiable, sensitive areas—while avoiding the temptation to expand these lines excessively. Third, on the basis of those red lines, dialogue and cooperation should begin in low-sensitivity areas—for instance, by sharing policy frameworks, best practices, and capacity-building efforts in AI governance, fostering mutual understanding and predictability. Fourth, the two countries should continue working within the UN framework to promote global AI governance and explore the establishment of authoritative institutions to regulate behavior and protect stakeholders. Fifth, as AI powerhouses, China and the U.S. have a shared responsibility to support global AI capacity-building and contribute to achieving the UN Sustainable Development Goals.
Wang Ning: Suppressing China—U.S. Chip Policy Unlikely to Succeed
The U.S.’ New Export Restrictions on Domestic Semiconductors and Suppression of Huawei’s Ascend Chips
Just as tensions in U.S.-China trade relations had slightly eased, the United States introduced two new policies concerning artificial intelligence. First, it announced the removal of previous export restrictions on semiconductors used for AI development, replacing them with non-binding guidelines. Second, the U.S. Department of Commerce’s Bureau of Industry and Security (BIS) issued detailed provisions indicating that companies using Huawei’s Ascend chips—even outside China—could potentially violate U.S. export controls and be penalized.
“One Loosening, One Tightening” Policy ———Balancing the Benefit of Domestic Firms with the Need to Control China’s AI Development
The U.S. shift in its AI chip policy toward China reflects a complex strategic adjustment. Initially, the U.S. imposed broad export restrictions on high-end chips to stifle China’s AI development. However, the loosening of these restrictions comes as a response to growing pressure from U.S. tech companies and the recognition that maintaining such barriers could erode U.S. market share abroad. The relaxation aims to preserve U.S. tech firms' access to international markets while still maintaining a competitive advantage in AI. At the same time, the U.S. has introduced tightening measures targeting Chinese AI chips and extended these controls to third countries, signaling a continued effort to suppress China’s AI capabilities, albeit shifting focus from blocking technology acquisition to restricting its application.
US Strategy Shifts from a Pure Technological Blockade into a Multidimensional Competition with China as It Attempts to Retain Its Leading Position in the AI Industry
This change reflects a broader U.S. strategy of moving from a technological blockade to a multidimensional competition. Under President Biden, frameworks like the "AI Diffusion Export Control Initiative" were designed to isolate China from advanced technology ecosystems. While this approach hurt Chinese access to top-tier chips, it also harmed U.S. commercial interests. The recent policy shift seeks to balance the U.S.'s technological dominance with corporate interests, but it faces challenges due to China’s rapidly advancing AI development, supported by a strong industrial base and extensive consumer market.
The U.S. Lacks the Manufacturing Power to Contain China’s AI Development
Despite U.S. policy efforts, China’s competitive edge in AI is growing. With more countries integrating AI into their national strategies, a more diversified global AI landscape is emerging. China’s focus on practical applications and internal innovation, supported by low-cost models like DeepSeek, will likely lead to breakthroughs in the near future. Ultimately, AI competitiveness will depend on real-world applications and iterative development. Additionally, China's comprehensive industrial system positions it well to withstand external pressures. In this globalized landscape, China is expected to continue advancing technological autarky while fostering international cooperation, such as through the Belt and Road Initiative, to promote shared technological growth.
YAO Shujie: Great Power Competition and the AI Wave: China’s Strategic Choices and Prospects
AI Evolution and Strategic Importance
AI has rapidly evolved through four phases: Perception AI (2015-2020), Generative AI (2020-2025), emerging Agentic AI, and future Physical AI. NVIDIA’s Jensen Huang has emphasized that this 10-year transformation fundamentally differs from previous industrial revolutions. Unlike mechanical power transitions, the fourth industrial revolution centers on AI's cognitive capabilities. AI is the decisive factor in 21st-century great power competition and the fight for global geopolitical influence.
Development Targets and Challenges
China’s “Three-Step” AI strategy aims to grow the industry from 150 billion yuan (2020) to 400 billion yuan (2025) and 1 trillion yuan by 2030, positioning China as the world's leading AI innovation hub. Key challenges include chip dependencies, algorithmic bias, and regulatory gaps. Meanwhile, China has reduced urban-rural income inequality from 3.1:1 to 2.34:1 through digital infrastructure investment, contrasting with widening US wealth gaps where the top 0.1% holds 13.8% of total wealth versus the bottom 50%'s 2.5% share.
China’s Catch-Up and Breakthrough in the Global AI Competition
The U.S. leads AI development, with top companies valued at $18.7 trillion versus China’s $2.82 trillion. However, China leverages its ability to coordinate its socialist market economy, a population 4 times the size of the U.S., over 250 million higher education graduates, over $400 billion in annual R&D investment, and dominance in emerging sectors like electric vehicles (60% global market share) to build comprehensive innovation capabilities despite missing earlier industrial revolutions.
Evolution and Future Trends of Great Power Competition in AI
The U.S.-China AI rivalry dominates global competition as both nations possess the unified economic scale needed for technological leadership. Europe has advanced capabilities (like ASML) but lacks cohesive super-economy status, while India trails significantly in GDP and AI competitiveness. Since 2018, the U.S. has systematically constrained China through tech blockades, export controls, and financial pressure on semiconductors and advanced manufacturing. China counters by accelerating domestic development and expanding international cooperation through Belt and Road Initiative and RCEP partnerships.
The Strategic Direction of US A.I. and China’s Countermeasures
Mark Kennedy from the US Wahba Institute proposes dual AI strategies reflecting ‘America First’ thinking. Offensive measures include attracting STEM talent and computing support and strengthening research funding. Defensive strategies encompass securing supply chains, imposing export controls, and influencing international AI standards to maintain technological leadership. Under these circumstances, China must develop systematic countermeasures emphasizing balanced security and openness, including empowering defense modernization through military AI applications and strengthening military-civilian integration to boost competitiveness.
LI Yan, ZHAI Yiming, et al.: U.S. Withdraws AI Diffusion Rule, Signals Shift Toward Stricter China-Focused Controls
On May 13, 2025, the Bureau of Industry and Security (BIS) rescinded the Biden-Era Artificial Intelligence Diffusion Rule, simultaneously announcing three new guidelines to reinforce export controls on AI chip shipments to China. The Trump Administration and major tech entities like NVIDIA, had criticized the original rule for imposing heavy compliance burdens and harming U.S. international AI cooperation. The move reflects growing domestic debate over AI policy and marks a clear shift toward stricter China-focused regulations under Trump 2.0.
New Guidelines Reveal Clearer U.S. Strategy to Contain China’s AI Rise
Issuing guidance that using Huawei Ascend chips anywhere in the world violates U.S. export controls.
This action aims to impede China's indigenous AI advancements and follows U.S. export restrictions on NVIDIA's advanced chips. It must be viewed in the light of Huawei's Ascend chips approaching NVIDIA's performance levels. To maintain its dominance in AI chips, the U.S. seeks to compel global entities to align with its strategic objectives, potentially paving the way for future sanctions against Ascend chip users.
Issuing guidance warning the public about the potential consequences of allowing U.S. AI chips to be used for training and about the interference of Chinese AI models.
Unlike the first guideline that bans Huawei Ascend chips, this is not a rule but a warning. It demonstrates that the U.S. government wants tighter controls but lacks a clear strategy. Even under Biden, there were concerns that China could bypass chip bans through cloud services or overseas model training. With the AI Diffusion Rule revoked and no unified approach in place, this warning signals the direction of future policy.
Issuing guidance to U.S. companies on how to protect supply chains against diversion tactics.
Diversion tactics clearly refer to China’s circumvention strategies, including smuggling and third-country rerouting to obtain U.S. AI chips. Reuters reported on May 5th that House Democrats intend to introduce legislation mandating regulatory oversight of AI chip deployment and the interdiction of unauthorized chip operations, potentially in concert with U.S. firms. This is with the aim to halt all chip exports to China.
U.S. AI Strategy Aims to Dominate Global Markets and Undercut China’s Influence
The global promotion of U.S. AI technologies emphasizes not only technological advancements but also market penetration and the establishment of international standards. Breakthroughs by firms like DeepSeek and Huawei have raised concerns over U.S. aims, prompting moves to block chip exports and prevent technology transfer to China. The broader goal is to construct a global AI ecosystem that excludes China and secures long-term U.S. dominance.
XIE Bo and WANG Ying: U.S. Export Controls on China in the Context of the U.S.-China Technological Competition—Evolution, Motivations and China’s Responses
The Background and Motivations of U.S. Export Controls on China
The U.S. strategically employs export controls to suppress China’s high-technology advancement and sustain its global technological hegemony amid intensifying bilateral competition. This dynamic creates a self-reinforcing cycle: technological rivalry drives increasingly stringent controls, which in turn escalate strategic competition while actively undermining mutual trade cooperation. Crucially, motivations extend beyond pure technological containment. Geopolitically, the U.S. leverages these controls to construct targeted technology coalitions aimed at isolating China and consolidating U.S. technological primacy to safeguard its geopolitical interests. Domestically, heightened bipartisan consensus exploits "China threat" narratives to divert attention from economic pressures while promoting techno-nationalist sentiment and reindustrialization, using export controls as a political tool.
The Evolution Trends of U.S. Export Controls on China
U.S. export control mechanisms continue to evolve across three dimensions. In scope, controls have expanded beyond military applications to encompass broad categories of civilian technological goods, justified through deliberately elastic interpretations of "national security" that enable discriminatory restrictions. Regarding focus, targeting has progressively narrowed toward foundational technologies—notably semiconductors, advanced computing chips, and associated manufacturing equipment—deemed critical to China’s industrial modernization. In methodology, controls are systematically weaponized through geopolitical channels: coercing allies into multilateral control regimes, diplomatically obstructing third-country technology transfers to China, and politically motivated designations of Chinese entities to the export control Entity List.
China’s Potential Responses to U.S. Export Controls
China should establish a multi-faceted counter-strategy anchored in systemic resilience and strategic autonomy. First, prioritize breakthroughs in bottleneck technologies by fully leveraging the "new national system" to concentrate resources. Second, mandate enterprises to strengthen risk prevention mechanisms and enhance supply chain diversification for greater resilience. Third, accelerate the development of a specialized export control system aligned with China’s strategic requirements. Concurrently, pursue calibrated cooperation with the U.S. in non-sensitive domains while rigorously managing core technological competition.
Conclusion
In the context of the U.S.-China AI competition, a strategy reflecting America First thinking has been proposed. This involves maintaining technological superiority through measures such as attracting STEM talent and increasing investment in computing and funding, while simultaneously implementing defensive strategies like supply chain security, export controls, and standard-setting. In response, China should construct a counter-strategy centered on systemic resilience and strategic autonomy, focusing on breakthroughs in bottleneck technologies and promoting supply chain diversification. Beyond competition, both parties should rebuild trust and engage in dialogue and cooperation in less sensitive areas, such as AI governance, while respecting national security red lines.
Writers and Editors for Today’s Newsletter:
Writers：WEN Yiran, WANG Jingyi, Ngoc(Jade) Bui, PEI Yutong, and XIAN Ruilian
Editors：SUN Chenghao, ZHANG Xueyu, and Hannah Shirley