The Second Threshold of Artificial Intelligence: Industrial Coordination, Employment Asymmetry, and Strategic Lockout in the United States and China

Artificial intelligence competition is commonly framed as a race for more capable models, larger compute infrastructures, and progress toward autonomous reasoning. This paper argues that such developments constitute only a first threshold. A second threshold is reached when artificial intelligence becomes an operating layer for physical production, coordinating factories, warehouses, logistics networks, energy systems, procurement, maintenance, finance, and demand within a continuously learning industrial architecture. Crossing this threshold may transform not only productivity, but also employment, geopolitical power, and the conditions under which later competitors can continue technological development. The paper develops a comparative conceptual framework for the United States and China. China possesses greater manufacturing depth, denser supply chains, and a much larger body of labor still embedded in industrial and logistics systems. These strengths may also increase the domestic employment shock of industrial automation. By contrast, the United States has already experienced substantial deindustrialization. Industrial AI may therefore rebuild production in sectors where domestic factories and jobs have already disappeared, replacing imports and overseas labor before displacing an equivalent number of existing American manufacturing workers. The same technological advance may thus produce manufacturing expansion in the United States but labor substitution and intensified overcapacity in China, particularly where productive capacity already exceeds effective demand. The framework further argues that the first actor to cross the second threshold may acquire a qualitatively different form of first-mover advantage. Superior intelligence, compute, industrial resources, and access to software and data infrastructures could allow a threshold leader to degrade the training environment of pre-threshold rivals through data poisoning, synthetic-content pollution, benchmark manipulation, supply-chain compromise, or attacks on model-development pipelines. Nightshade-style concept poisoning provides a limited technical precedent for the broader possibility that apparently usable training material can induce systematically false associations. Such active disruption may later be reinforced by standards, certification regimes, platform dependence, and trust requirements, converting an initial capability gap into durable strategic lockout. The Wuhan robotaxi controversy illustrates a complementary mechanism: automation becomes politically consequential when it replaces visible labor and when technical failure is converted into a public governance problem. The paper therefore distinguishes front-stage automation from less visible back-stage automation and proposes observable indicators for testing the theory. It concludes that the decisive AI race may be won not by the country with the best chatbot or the largest present manufacturing base, but by the actor that can industrialize AI, absorb the resulting employment shock, preserve effective demand, and prevent competitors from completing the same transition. Keywords Artificial intelligence; industrial coordination; technological forecasting; automation; employment displacement; manufacturing; United States; China; data poisoning; strategic lockout; geopolitical competition