“Over the next few years, nations face a critical window to prevent permanent divergence in AI development,” warned Professor YAO Xu, Associate Research Fellow at Fudan Development Institute and Distinguished Research Fellow at Shanghai Data Institute. “Failure to act now risks being left behind as US-China dominance solidifies.”Speaking to Fudan University Media Center at the 2025 Shanghai Forum, Yao presented a comprehensive analysis of global AI development trends, regional disparities, and China’s evolving role. He emphasized how emerging dynamics in geopolitics, data governance, and industrial transformation are actively reshaping the current AI revolution and global competitive landscape.
Q: Artificial intelligence is one of your key research areas. In a previous interview, you noted that “global AI innovation today requires competition and collaboration among China, the US, the EU, and the Global South.” How do you assess the disparities in AI development across regions? What do you see as the next major direction for AI?
Yao: Generally, China and the US form the leading tier. The US maintains an undeniable advantage across the entire AI industrial chain and technological ecosystem. Meanwhile, China has progressed rapidly in recent years, producing groundbreaking AI models like DeepSeek while establishing a strong presence through tech giants and startups. According to a recent report by China’s Ministry of Industry and Information Technology, China now boasts over 4,500 AI companies.
Europe, meanwhile, is striving to catch up. At the Paris AI Summit in February2025, EU Commission President Ursula von der Leyen unveiled the “Invest AI” plan, mobilizing €200 billion for AI development. French President Emmanuel Macron pledged €109 billion in domestic AI investments. Also, Europe is nurturing startups like Mistral AI, a rising star in foundational models.
Other regions aspire to advance their AI ecosystems, but progress hinges on four critical pillars: computing power, algorithms, data, and talent. Few nations outside the US, China, and Europe possess the resources to excel in all four. For instance, China graduates over 5 million STEM students annually—a figure surpassing the total population of many countries. The US benefits from over 15 years of robust digital economy growth, anchored by leading tech companies like the “Magnificent Seven”.However, open-source models like DeepSeek, which optimize computing efficiency, offer new possibilities for the Global South. While leading AI hubs will likely continue driving model and application development, the greatest demand and market opportunities may increasingly shift to developing economies.
Q: Beyond the leading tier, which regions—such as the EU or Global South—show potential to narrow the gap or make breakthroughs?
Yao: AI development is subject tothe Matthew effect,in which early advantages compound. The next few years will present a critical window. Countries that can’t meet AI’s substantial industrial and infrastructure requirements risk permanent disadvantage.
Europe holds particular promise due to three factors: emerging startups like Mistral AI, substantial investments, and strong R&D foundations. Its success will hinge on effectively bridging academic research with commercial applications, while carefully navigating geopolitical relationships, particularly with China.
Q: What are China’s advantages in AI innovation and industrial development? What key steps do Chinese AI firms need to take to expand globally, and what challenges remain?
Yao: China’s AI sector is gaining global recognition, with ecosystem players across the value chain—from core model developers to supporting enterprises—initiating overseas projects. However, two critical risks loom: geopolitical volatility and mismatches between products and local demand.
For the first challenge, firms must develop tailored strategies—selecting target markets and entry approaches based on short-term, medium-term, and long-term assessments. For the second, they must tightly align products with vertical solutions tightly aligned with local use cases, as AI applications remain highly scenario-dependent.
Q: You’ve emphasized that “AI’s future hinges not just on policy-technology-industry alignment but also on data quality and accessibility.” How should we interpret this?
Yao: Data will be a decisive factor in global AI competition. China must enhance policy frameworks for data openness and cross-border flows—a shift mirrored by the EU’s recent pivot from strict regulation todevelopment-security balance.This contrasts sharply with the US’s national security-focused data approach, which explicitly targets China’s AI growth.
Moreover, with projections suggesting existing internet data may be exhausted by 2027–2028, building high-quality datasets is urgent. Current efforts prioritize curating such datasets to sustain next-generation model training.
Q: The United Nations’ “Enhancing International Cooperation on AI Capacity-Building” resolution, led by China and co-signed by 143 nations, highlights governance needs. What are the keys to effective AI governance, and what rules should a global framework include?
Yao: Bridging digital divides, ensuring technological equity, and promoting inclusive development must anchor global AI governance. We must prevent artificial barriers to beneficial technology transfer. For Global South nations, this means developing innovative financing solutions through either reformed multilateral systems or new partnership models.
From China’s perspective, building a UN-centered governance framework in collaboration with Global South countries to counter deglobalization trends represents its overarching approach and direction.
Q: What essential skills should young people develop in the AI era?
Yao: In the AI age, young people should focus on three critical competencies. First, Cultivating logical thinking, curiosity and discernment—without these, you risk “learned helplessness”, where AI erodes your analytical habits. Secondly, resist shortcuts and avoid AI-generated superficial outputs, and instead, training yourself to think independently and create original frameworks while using AI as a tool, not a crutch. Thirdly, be grounded in structured knowledge, leverage AI productively and avoid misinformation traps.
(END)
Writer: FENG Zihan
Proofreader: WANG Jingyang
Editor: WANG Mengqi, LI Yijie
