Xi Jinping’s paranoid approach to AGI, debt crisis, & Politburo politics — Victor Shih

Victor Shih analyzes China’s political and economic system, highlighting Xi Jinping’s centralized, micromanaging leadership style and cautious approach to AI development to maintain Party control. He also discusses China’s high local debt, structural inefficiencies, and the elite’s technocratic background, emphasizing that true power remains concentrated with Xi and his inner circle.

In the interview with Victor Shih, he provides an in-depth analysis of China’s political and economic structure, emphasizing its unique blend of decentralization and central control. While China appears decentralized in terms of government spending—about 85% at the provincial and local levels—this decentralization is largely superficial. Historically, local governments had significant fiscal autonomy, especially before 1994, when the central government centralized tax collection, notably through value-added tax, and began controlling revenue streams. Although localities gained some land revenue autonomy during the real estate boom, recent restrictions have increased their dependence on the central government, making the system highly top-down in practice.

Shih discusses the composition and backgrounds of China’s elite leadership, noting that many Politburo members possess advanced technical degrees, including PhDs in engineering, economics, or Marxist theory. While these credentials suggest a technocratic class, their governance skills vary. Some leaders, like Li Keqiang, have a solid understanding of economics, but others with technical backgrounds may lack broader policy or political acumen. Xi Jinping himself is portrayed as a politically astute leader who micromanages policy through numerous meetings and leading small groups, ensuring his directives are implemented and maintaining tight control over decision-making processes.

The conversation highlights Xi Jinping’s leadership style, comparing it to Stalin’s micromanagement, with Xi personally involved in daily policy discussions and decision-making. His trusted lieutenant, Ding Xuexiang, plays a crucial role in managing key policy areas like cybersecurity and AI, and his relationship with Xi is characterized by deep trust, partly built during their time working together in Shanghai. Xi’s approach involves consolidating power through a series of coalitions and purges, similar to historical authoritarian tactics, and he maintains control by ensuring no one else can challenge his authority, often by sidelining or purging rivals.

On China’s AI and technological ambitions, Shih notes that the Party is highly cautious about the potential disruptive power of AGI, especially content generation that could undermine Party control. While China enthusiastically supports AI development, including investments in companies like DeepSeek, there is a parallel focus on developing regulatory “brakes” to halt or control AI if it threatens stability or Party authority. The Party’s response to AI advancements involves integrating them into existing cybersecurity and control structures, with Ding Xuexiang overseeing much of this strategic planning, reflecting a cautious but ambitious approach to technological dominance.

Finally, Shih discusses China’s broader economic and political challenges, including its massive local government debt—estimated at over 120% of GDP—and the inefficiencies of its state-led investment system. Despite high debt levels, the Party’s priorities, such as military expansion and technological self-sufficiency, are unlikely to change significantly. He warns that without reforms to reduce unnecessary spending and improve welfare, China’s growth prospects may be limited, especially as demographic decline and diminishing returns on infrastructure investments set in. The interview concludes with reflections on succession planning, elite factionalism, and the importance of understanding China’s opaque political system, emphasizing that true power remains tightly held by Xi Jinping and his inner circle.