Summary of Key Points
This article corrects the one-sided conclusion that “China has surpassed the United States in AI” by analyzing two widely shared charts on AI talent. While China is indeed the largest global hub for AI talent development (47% of top AI researchers graduated from Chinese universities), and it is rapidly catching up in terms of contributions to top academic conferences (with more top AI researchers working in China by 2025 than in the U.S.), these figures do not represent overall strength. The competition between China and the U.S. in AI has shifted from a focus on individual metrics to a battle for comprehensive ecosystems. China’s strengths lie in talent cultivation, engineering capabilities, and the application market, while the U.S. excels in attracting global talent, fostering original innovation, and building a robust ecosystem. The key to future success will be “who can make the most of their talent,” rather than simply comparing the number of talents or papers published.
The “Tricks” Behind the Charts: Don’t Be Misled by the Numbers
The statistical methods used in these two charts conceal important information and cannot be directly equated with overall AI capabilities:
- The first chart (workplaces of first authors at NeurIPS): NeurIPS is the premier conference in the AI community, similar to the Oscars in the film industry. However, it only tracks the current workplaces of paper authors. For example, if many researchers from Chinese universities or companies publish more papers, it indicates an increase in China’s research strength, but this reflects only one aspect of AI capabilities (such as basic research) and not the entire picture (including industrial applications and computational power).
- The second chart (places of undergraduate graduation): Although 47% of top AI researchers graduated from Chinese universities, this does not mean they all work in China. Many talented Chinese students may study at prestigious institutions like Tsinghua or Peking University before pursuing doctoral degrees and working for companies like OpenAI or Google DeepMind in the U.S. China is a cradle of talent, but not all graduates stay domestically.
China and the U.S. Have Different Strengths: One Focuses on Talent Cultivation, the Other on Attraction
The advantages of China and the U.S. are distinct, rather than one being superior to the other:
- China’s strengths:
- It is the largest global hub for AI talent development, producing a large number of AI-related undergraduates each year.
- Strong engineering capabilities, as evidenced in advancements in autonomous driving, AI robotics, and short-video algorithms, which can quickly transform research into practical products (such as TikTok’s recommendation systems).
- A vast application market with a population of 1.4 billion smartphone users provides ample opportunities for testing AI technologies (e.g., AI-driven food delivery and e-commerce recommendations).
- The U.S.’s strengths:
- It is a global magnet for top talent from China, India, and Europe (many core researchers at OpenAI are Chinese).
- Leading in original innovation with breakthrough models like GPT and Claude developed by U.S. companies.
- Access to advanced technology and capital, including cutting-edge chips (e.g., NVIDIA GPUs) and substantial venture funding for AI startups.
Changes in Talent Flow: From “One-Way Outflow” to “Two-Way Movement”
In the past, many Chinese talents went to the U.S. for career development, but the situation is now reversing:
- Talents staying in China: More AI undergraduates are choosing to pursue doctoral degrees or work in domestic companies (e.g., ByteDance and DeepSeek).
- Return of Overseas Talent: Some Chinese scientists working at OpenAI and Google are returning to China due to better opportunities and higher salaries.
- U.S. Policies as a Barrier: Restrictions on high-end chip exports and tighter research visas are deterring some foreign talents from moving to the U.S.
In the future, talent will not flow solely to the U.S.; both countries will have competitive advantages.
The Heart of AI Competition: It’s About a Comprehensive Ecosystem
The competition in AI is no longer about who has the most papers or talents; it’s about having a complete ecosystem. This includes:
- Education: Producing skilled talent.
- Research: Generating innovative results.
- Capital: Funding research and development.
- Industry: Transforming research into market-ready products.
- Computational Power: Providing the necessary computational resources.
- Policy: Supporting innovation and attracting global resources.
What China Needs to Improve:
- Original innovation capabilities, such as developing breakthrough technologies like GPT.
- Access to high-end chips that are not dependent on others.
What the U.S. Should Be Concerned About:
- A potential decline in its ability to attract top talent due to visa policies and other factors.
- The possibility of its application market being smaller than China’s.
Therefore, rather than debating who is ahead, it is more important to focus on how each country can maximize the value of their talents. For example, whether the talents cultivated in China can create world-class AI products domestically, and whether the talents attracted by the U.S. can continue to drive original innovation. These are the real issues behind these two charts.