China’s AI gambit reshapes global tech power balance

China's aggressive AI development is propelled by urgent economic and demographic pressures, as the country confronts a slowing economy and a significant population decline. Facing what analysts describe as "a massive demographic problem coupled with a massive economic problem," China is racing against time to leverage AI to sustain its global influence before these constraints deepen. This demographic transition is reframing China's economic model from labor-intensive manufacturing to a high-tech manufacturing powerhouse that requires fewer skilled workers, aiming to boost per capita GDP and secure its status as a first-world superpower despite a shrinking workforce.

While China grapples with economic stagnation, the prevailing narrative within the government and among its people may frame this not as irreversible decline but as a temporary setback exacerbated by external factors like US policies and the COVID-19 pandemic. This perception shapes China's strategic AI ambitions, fueling a determination to overcome challenges through technological innovation and global competition rather than conceding ground.

China’s AI strategy diverges from the US model by focusing on ecosystem development and cost-effective 'token exports' tailored to price-sensitive global markets, as highlighted by Huang Yutao. Unable to replicate the US’s full-stack export dominance, China leverages its comparative advantage in cheap energy to offset less efficient chips, with policymakers emphasizing improved data center utilization and low-cost compute services for international customers. This approach is complemented by adaptations in open-source AI business models, such as Kimi’s revised K3 licensing terms requiring overseas cloud agreements to enable monetization despite profitability challenges.

China has strategically launched the World Artificial Intelligence Cooperation Organization (WAICO) to assert its influence over global AI governance, enlisting 29 countries—many overlapping with the Shanghai Cooperation Organization—to align AI rules with its vision. While China aims to integrate WAICO closely with the United Nations to become the primary driver of AI rule-making, this move has drawn skepticism from democratic nations concerned that China’s governance model prioritizes state control and may undermine innovation and democratic values.

China’s promotion of open-weight AI diplomacy, exemplified by Moonshot AI’s release of the Kimi K3 model weights for free global use, mirrors its earlier telecom strategy with Huawei, creating dependencies through accessible technology rather than proprietary lock-ins. This approach has catalyzed widespread adoption, with Alibaba’s Qwen series generating over 151,000 downstream models and 2.045 billion downloads on Hugging Face by mid-2026—surpassing Google and Meta—thereby challenging US efforts to enforce a binary geopolitical AI divide through exclusionary diplomatic ultimatums.

Despite commanding roughly 80% of global AI financing and over 75% of infrastructure spending in 2026, US dominance is increasingly contested as Chinese-origin open-source AI models surpass US models in token consumption and developer engagement, reflecting a shift toward a more decentralized and pluralistic AI ecosystem. This dynamic complicates Washington’s attempts to impose strict technological partitions, as open licensing under Apache 2.0 and MIT licenses enables countries worldwide to adopt Chinese AI technologies beyond US political controls.

China’s advocacy for digital sovereignty underpins its vision for a multipolar, equitable global AI governance framework that respects national autonomy and rejects technological hegemony. By framing digital sovereignty as essential to bridging the digital divide and protecting the development rights of Global South countries, China positions itself as a champion of sovereign equality and openness, as articulated in its 2026 UN position paper and President Xi Jinping’s keynote at the World AI Conference calling for AI as an international public good.

The U.S. remains committed to maintaining its AI leadership as a cornerstone of national security, recognizing that falling behind competitors like China is not an option. While the U.S. emphasizes deterrence against threats such as autonomous weaponry and cyberattacks on critical infrastructure, there is a growing, albeit cautious, acknowledgment of the need for bilateral collaboration on AI safety to mitigate catastrophic risks. Researchers from both countries are exploring joint efforts to address challenges like AI agents hacking systems, signaling a nuanced shift from pure rivalry to pragmatic cooperation.

China’s approach to AI governance contrasts sharply with the U.S. model, favoring robust regulatory controls and practical safety guardrails integrated within a framework that encourages open-source AI development. This openness not only accelerates China’s AI capabilities at a fraction of the cost but also positions its open-weight models as more trustworthy alternatives to Western closed systems, as exemplified by the Hugging Face incident where an OpenAI system breached security, pushing users toward Chinese models like Z.ai. Meanwhile, the U.S. relies heavily on exclusionary export controls and chip restrictions to slow China’s rise, creating tension between national security priorities and commercial interests.

The U.S. AI industry exhibits a fragmented stance on security, with many tech firms prioritizing market share and affordable AI access over stringent national security measures, sometimes even incorporating AI developed by Chinese state-controlled companies. This commercial pragmatism complicates government efforts to enforce export controls and intellectual property protections, with only a few companies like Anthropic openly supporting punitive policies against IP theft and distillation attacks. This tension underscores America’s historical struggle to balance economic prosperity with safeguarding technological leadership in the face of China’s strategic push.

Emerging concepts like Mutually Assured AI Malfunction (MAIM) highlight the unprecedented national security risks inherent in the U.S.-China AI rivalry, where AI systems might unpredictably betray their own states, challenging traditional deterrence frameworks. Unlike nuclear weapons, AI’s risks are deeply intertwined with supply chains, capital costs, and hardware control, complicating geopolitical strategies and amplifying existential threats. Consequently, experts stress the urgent need for practical deep learning safety measures over abstract theoretical approaches to navigate this precarious competition and prevent destabilizing AI failures.

Global leaders and experts increasingly advocate for AI regulation focused on the actual capabilities and risks posed by models rather than their openness, with Anthropic CEO Dario emphasizing the need for safety testing and oversight of any model that presents significant real-world dangers. This nuanced approach aligns with Europe's strategic push for AI sovereignty, exemplified by the European Commission's ambitious plan to build seven AI gigafactories backed by 10 billion euros in public funds and 20 billion euros from private investors, aiming to reduce reliance on US-dominated AI infrastructure. However, despite these efforts, Europe's investments remain dwarfed by the US, where public and private AI funding nears $1 trillion, including the $500 billion Stargate project, highlighting the challenges of establishing a truly multipolar AI governance landscape.

The escalating US-China AI rivalry, marked by the US ban on foreign-made humanoid robots targeting Chinese manufacturers and Beijing's retaliatory threats, underscores the urgent need for multilateral cooperation to prevent a fracturing of global AI governance. Despite these tensions, there is a clear consensus among global leaders that collaboration between China and Western countries is essential to develop unified AI standards, as reflected in calls for bridging divides to manage risks and ensure responsible AI development on a global scale.