After DeepSeek Episode 1

After Deepseek: China’s AI Talent Revolution

Key Concepts:

• Deepseek: A Chinese AI company whose RO1 model launch significantly impacted the global tech landscape, particularly Nvidia’s market cap.
• Brain Gain: The increasing trend of highly skilled Chinese professionals returning to China from abroad.
• Human Capital: The skills, knowledge, and experience possessed by an individual or population, considered as an economic asset.
• STEM Education: Education in Science, Technology, Engineering, and Mathematics.
• Strategic Acceleration Model: China’s approach to tech development, characterized by government support followed by private sector expansion.
• ResNet: A groundbreaking deep learning architecture developed by Microsoft Research Asia, fundamental to modern AI.
• Thousand Talents Plan: A Chinese government recruitment program aimed at attracting overseas experts (now phased out).

I. The Deepseek Disruption & China’s Rising Tech Reputation

The launch of Deepseek’s RO1 model in January 2025 marked a turning point, causing Nvidia to lose nearly $600 billion in market capitalization. This event signaled a fundamental shift in China’s global tech reputation, challenging the narrative of reliance on Western technology and demonstrating a growing self-sufficiency in AI. Deepseek achieved this by training an AI model at a fraction of the cost incurred by industry giants like OpenAI, Google, and Anthropic. This success is attributed to a combination of factors, primarily the strength of China’s human capital. As Koko Fun, host of the After Deepseek podcast, notes, she has “a very long experience in this area” having studied AI for 20 years.

II. The Role of Chinese Talent: A Domestic Pipeline & Returnees

The podcast explores the critical role of Chinese talent in driving this technological advancement. Moran Ga, author of The New Geography of Innovation, argues that China is “the only competitor against the US,” and that its talent pipeline is a “decisive factor” in its rapid progress, not just in AI, but also in areas like electric vehicles (EVs) and solar energy. China’s university system is heavily focused on STEM fields – engineering, computer science, and mathematics – creating a strong foundation for AI development.

Beyond the domestic pipeline, there’s a growing trend of highly skilled Chinese professionals returning home. Examples cited include Sun Nan, a Harvard PhD and expert in AI semiconductor chips who returned to Shinua University, and Yashin Zhang, who left Microsoft after two decades to lead an AI industry research lab at Shinua. However, Dr. David Swag, a China expert, clarifies that the majority of returnees are master’s degree holders (85%), while only a smaller proportion (though increasing from 1% in 2000 to 17% currently) are high-level researchers.

III. Motivations for Returning: Opportunity, Culture & Patriotism

Zo, founder of AI fashion tracker JUI Tech, exemplifies the returnee phenomenon. After graduating from Peking University, earning a master’s degree from Keio University, and working at Google for three years, he chose to start his company in China. He cites the greater ease of starting a business in China compared to the US, attributing the difficulty in the US to a “culture gap.” While acknowledging the friendliness of Americans, he found it challenging to lead a culture as a Chinese entrepreneur. He also highlights the greater opportunities in applying AI to traditional Chinese industries.

Multiple interviewees suggest that returnees are drawn to China by faster decision-making, a more impactful environment, and a sense of “patriotism” or “civilizational pride” in contributing to China’s technological advancement. They are often offered greater responsibility and larger research teams than they would receive in the US at a similar career stage.

IV. Government Support & the State’s Role in AI Development

Beijing has actively supported its tech sector through policies like the now-defunct Thousand Talents Plan, aimed at recruiting overseas experts. However, the implementation of this plan faced criticism, creating a difficult situation for Chinese scholars in the US and leading to increased scrutiny from the US government. Dr. Swag describes this as a “terrible phenomenon” resulting in a lack of collaborative research.

Moran Ga argues that the idea of a solely “state-driven” AI ecosystem in China is overstated. She points out that many countries provide state support to tech companies, citing SpaceX and the origins of Silicon Valley as examples of government funding. She characterizes China’s approach as a “strategic acceleration model” where the government builds momentum, and the market multiplies it, with the majority of advancements occurring in the private sector (Alibaba, Baidu, Bytedance, Sensetime).

V. The Rise of Hubs Like Hangzhou & the Importance of STEM Graduates

The podcast highlights Hangzhou (Ho) as a burgeoning tech hub, benefiting from Deepseek’s success and attracting investment. Zo notes that the Hangzhou government provides substantial financial incentives and supportive policies to startups. However, Professor Swag cautions that non-financial institutional challenges remain, such as the requirement for Communist Party members on company boards, which may deter some returnees who value freedom.

China’s commitment to STEM education is a key driver of its success. The country graduates one of the largest cohorts of STEM graduates globally – over 12 million in 2025. Leang Wenfong, Deepseek’s founder, exemplifies this, having achieved the highest score in his province on the 2002 National College Entrance Exam and attending Jang University in Hangzhou. This investment in STEM has enabled China to overcome hardware limitations through software and algorithmic efficiency.

VI. Rethinking Valuation & the Future of Innovation

The success of Deepseek, despite its relatively lower valuation, challenges traditional metrics for assessing a company’s innovative potential. The podcast suggests that massive funding rounds and high valuations are not always reliable indicators of future success. Deepseek demonstrates that significant innovation can be achieved with fewer resources, potentially reshaping the rules of the AI development game. As one expert notes, “the world’s perceptions of China tend to lag reality by about 5 years.”

Notable Quotes:

• Moran Ga: “China is the only competitor against the US.”
• Zo: “If you are using AI to help traditional industry to improve, then I think China is the best place to be.”
• Koko Fun: “Wherever you think China is right now, that's where it was 5 years ago.”
• Dr. David Swag: “The outcome is the lack of collaboration, collaborative research.”

Technical Terms:

• RO1 Model: Deepseek’s AI model that triggered the global tech disruption.
• ResNet (Residual Network): A deep learning architecture developed by Microsoft Research Asia, crucial for training ultra-deep neural networks.
• STEM: Science, Technology, Engineering, and Mathematics – the focus of China’s educational investment.
• Skip Connections: The architectural innovation within ResNet that allows for the training of deeper neural networks.

Conclusion:

The After Deepseek podcast paints a compelling picture of China’s AI revolution, driven by a potent combination of domestic talent, returning professionals, strategic government support, and a thriving entrepreneurial ecosystem. Deepseek’s success is not merely a technological achievement but a symbolic shift, signaling China’s emergence as a genuine competitor in the global AI landscape and a move beyond a “copycat” reputation towards original innovation. The podcast emphasizes that China’s strength lies in its ability to leverage its vast human capital and create an environment conducive to rapid development and impactful application of AI technologies.