Nowadays, reading the news, following social trends, passing daily life, it has become barely barely impossible to avoid the term “artifical intelligence”. Considering the high ambitions and tremendous investments into this field, the question arises: on how long can the ongoing hype keep global financial markets stable? Despite the global disruptions caused by the Iran war and the blockade of the Hormuz strait, major indices like the S&P500 show remarkable resilience. The US tech sector is still charging ahead: Six of the ‘Magnificent 7‘ (which make up about a third of the S&P 500) once again outperformed earnings expectations in the first quarter of 2026. With surging demand for new power capacity, U.S. incremental spending on AI compute infrastructure is projected to reach $1.4 trillion by 2029—an investment equal to the entire economy of the Netherlands, ranking 18th in the world.
The financial clout of AI in the U.S. can be summarized into a single question: achieving AGI, or artificial general intelligence. It is a multi-trillion-dollar bet that AI will eventually supersede human proficiency across most cognitive domains. Nevertheless, there is another gambler who entered the bet on the prospects of AI, namely China, a trailblazer in AI innovations. On the one hand, it seems that AI is used everywhere in China, from AI-driven diagnostic tools in healthcare to standardized AI curricula in elementary schools, culminating in the overwhelming success by AI assistant Open Claw. All major mobile apps in China run on AI chatbots. A restaurant in Hangzhou has gone viral as it uses AI to scan the costumer’s face before recommending seasonal and health-focused meals. On the other hand, private AI investment in China is 23 times lower than in the United States. Venture capital funding for AI startups declined by nearly 50 percent in the first quarter of 2025, compared to 2024.
This bears critical questions: Has China become ‘saturated’ in cutting-edge AI development, and what are the core features of its strategy?
In its most recent Five-Year-Plan, published in March 2026, China made clear how it prioritizes AI development as a means of national security. The term ‘artificial intelligence’ presents itself as the most used technology-related word througouht the paper, framed explicitly as a cornerstone of national security. Instead of betting on AGI as a theoretical ‘point of no return’ in technology, it follows a much more pragmatic approach, focusing on three core pillars: maintaining political control, addressing the industry’s computational challenge, and doubling down on open-source models to compete with Western AI systems. In sum, China’s AI ecosystem is heading in a fundamental different direction than the United States.
China’s AI Ecosystem: Important players beyond the monolith
The modern era of China’s current AI strategy began in summer 2017, with the publication of the “New Generation Artificial Intelligence Development Plan” (新一代人工智能发展规划, AIDP). It was a declaration both to the world and to its own citizens that AI had been elevated to a matter of core national security interest. The AIDP set a definitive timeline: by 2030, China wants to be the world’s primary AI innovation center. The Chinese government made clear that it sets strategic goals, while local governments and private firms maintain the agency to experiment independently.
Rather than a centrally directed, rigid roadmap, the AIDP has been refered to as an authoritative ‘wish list‘. In this ecosystem, the state sets the goal, and the private sector, incentivized by state investment and a lack of regulation, competes to reach it. Successful local innovations will eventually be adopted as national standards.
The AI+ Initiative, published on August 26, 2025, plays an equally important role (even though it merely collects policy recommendations by the State Council [国务院关于深入实施“人工智能+”行动的意见]) because it puts the finger on complete immersion and economic diffusion. Its targets are aggressive: The plan aims for AI penetration to exceed 70% in key sectors by 2027, 90% by 2030, and by 2035, China will have fully transitioned to a “new intelligent economy and smart society”. The plus in the paper’s name shows how China’s strategy evolves around implementation and building a full-stack AI ecosystem. China attempts to reach this goal by balancing strict political oversight, litigating its dependency on foreign hardware, and achieving competitiveness with open-source models.
The First Pillar: Maintaining Political Control
Balancing AI’s massive productive gains with its potential threats to regime stability is one of the main concerns for China’s leadership. During the 20th collective study session of the 20th Politburo in April 2025, Xi made unusual open statements on the dual nature of AI technology, noting that AI frontier technology can bring “unprecedented development opportunities” while simultaneously presenting “unprecedented risks and challenges”. For the past years, Beijing’s AI governance has followed a predictable logic: Eras of booming markets and rapid AI developments allowed the central government to prioritize political control and regime security; when the economy was weak and AI companies lagged behind, the CCP felt urged to loosen its regulatory framework and incentivize experimentation and pragmatism. The release of DeepSeek and 2025 its enormous success broke the link between economic growth and technological innovation: Chinese AI capabilities looked surprisingly strong, despite a struggling macroeconomic environment. China’s ‘DeepSeek moment’ suggests a new reality: Beijing may no longer need a booming economy to sustain its technological momentum and political control.
There are signs that the government will return to control-focused policies of the previous years and further prioritize wide-scale AI deployment over frontier research. For example, regulatory frameworks such as the “Measures for Labeling of AI-Generated Synthetic Content” require service providers to guarantee user transparency and, first and foremost, centralize data flows to the government. Internationally, China seems to be more intended to foster collaboration with companies from around the world to contribute to AI safety. The Chinese AI Safety and Development Association (CnAISDA) was launched during the AI Action Summit in Paris in February 2025, describing itself as “representing China in dialogue and collaboration with AI security research institutions around the world.” Yet during the recent Trump-Xi meeting in May, AI governance remained the ‘Elephant in the Room‘, and the summit did not produce any notable results.
The Second Pillar: Achieving Compute Sovereignty
Probably the most pressing challenge for the Chinese AI industry is to become self-reliant in chip manufacturing. Advanced semiconductors are the essential hardware that provides the computation power required to process massive AI datasets. Already starting in 2014, the government has initiated a massive Manhattan Project-like program to establish full-stack domestic supply chains, aiming to become the global leader in the semiconductor industry by 2030. However, its was US policy that catalyzed Chinas focus. Successive US trade restrictions, including Trump’s bans on Huawei and ZTE and the Biden administration’s 2022-2023 export controls, showed how dependent China was on foreign supply. By 2026, China has made substantial progress in semiconductor indigenization, though outcomes vary across segments. In the AI chip sector, domestic firms captured nearly 41% of the market in 2025, a significant shift from Nvidia’s pre-2023 dominance of over 95%.
China’s main manufacturing bottleneck lies in advanced-node fabrication capacity. Here, a European tech giant comes into play. As Chinese foundries (like SMIC and Hua Hong) cannot purchase extreme ultraviolet (EUV) lithography machines from Dutch manufacturer AMSL, China is forced to use older deep ultraviolet (DUV) equipment. This makes the production process much more expensive and results in a high number of defective chips, hindering China to ramp up the volume of production. To bypass these bottlenecks, Beijing supports domestic equipment makers like NAURA and SMEE. According to a report by Reuters, a prototype EUV lithography machine has been reported in Shenzhen. The goal is to quintuple production (from 20,000 to 100,000 wafers per year), but this is difficult to achieve while being cut off from international equipment and high-end tools.
The Third Pillar: China’s Open-Source Offensive
The third aspect of China’s AI priorities the very nature of its AI models. Faced with US export controls, China cannot compete solely on ‘brute force‘ scaling. Instead, Alibaba, DeepSeek and others release open-source models, whose essential features—code, training data, documentation, and others—are freely available to download and modify. This is a calculated power play to make open-source China-led architecture the global standard—and probably the most clearcut difference to the US, where titans like OpenAI and Anthropic guard their black box models behind proprietary walls.
The key strength of Chinese AI models lies in their high degree of adaptability, allowing them to be customized and deployed efficiently a wide range of platforms. This has far-reaching consequences. For instance, Airbnb CEO Brian Chesky recently lauded Alibaba’s Qwen application system as fast, capable and inexpensive for its customer service. Even Silicon Valley startups are increasingly leveraging Chinese open-source models as the architecture for their own products. Developers across Southeast Asia and Africa are actively building vital public infrastructure in healthcare diagnostic tools and education systems directly on Chinese AI frameworks.
Perhaps most visible characteristic of China’s strategy is the ’embodiment’ of technology: physical machinery, or, to rephrase it in a simpler way, AI-empowered robotics. At this year’s CCTV Spring Festival Gala, humanoid robotics were shown as highly agile performers, performing in Martial Arts, freestyle parkour, and Kung Fu. This was a clear message about China’s advancements in embodied AI. Under the “AI Plus” initiative, the goal is to weave intelligence directly into the physical economy, particularly in China’s manufacturing and electronics industries. For example, Hangzhou-based robot maker Unitree recently announced the production of over 5000 humanoid robots. These robotics and embodied systems are a key priority of the recent Five-Year-Plan and primary targets for a new $138 billion national venture capital guidance fund. Furthermore, the overarching strategy of “military-civil fusion” ensures that these commercial breakthroughs in robotics and autonomous systems are seamlessly adapted for the defense sector.
And what about Artificial General Intelligence?
The idea of a global ‘race to AGI’ has become widely popular among US policymakers and analysts. Pioneered by futurists like Ray Kurzweil, who famously predicted AGI would arrive by 2029, this approach suggests that AI will trigger an intelligence explosion with unparalleled consequences for our social life, economy, and military.
The most recent FYP and the 2025 AI+ initiative, however, remain surprisingly vague on that matter.1 Instead, China follows a full-stack approach to AI. It aims to localize every critical component of the AI ecosystems, from chips and computing infrastructure to foundation models. According to Kyle Chan of the Brookings Institute, the goal is to turbocharge the deployment of AI as a general-purpose technology by making the models more efficient, adaptive, and deeply integrated into industry and the military. Of course, it´s possible to find a more nuanced picture when zooming in, with abundant discussion in the US of the merits to further push frontier AI development. Among Chinese AI companies, such as DeepSeek and Alibaba, there are also plans to achieve AGI. Beijing’s decision to block the Manus acquisition by Meta also shows it wants to keep existing frontier models at home.
This, however, is not a common feature of Chinese policy documents. To date, there are no clear references to AGI in the public statements of Xi Jinping. Chinese policymakers appear more concerned with the physical integration of AI. Their priority is the practical application of AI into the real-world economy and ensuring a resilient, self-reliant supply chain, while outbalancing the perceived threats to regime stability.
Further Reading:
Meserole, C. (2024). China’s Overlooked AI Strategy. In Foreign Affairs (Foreign Affairs)
Allen, G. C. (2024). China’s overlapping tech industrial policies. In High Capacity (High Capacity)
- Artificial General Intelligence is translated into Chinese as 通用人工智能, Tōngyòng Réngōng Zhìnéng, with the term 通用 referring to its general nature. As argued by Jordan Schneider, references to this, such as in a 2023 Politburo meeting, are more often about “general-purpose AI” compared to “specialized AI” (专用人工智能), rather than AGI as commonly understood in the US. ↩︎
