Commoditising Intelligence: Chinese Open-Weight Models and the Valuation Logic of the American AI Boom

Frontier-capable Chinese open-weight AI models challenge the scarcity assumptions supporting American model valuations and infrastructure investment. This paper asks under what conditions that competitive shock could materially reprice the US AI complex. It combines model price-performance evidence, routing and enterprise-spending data, company disclosures, macroeconomic indicators, and financial-stability analysis within a seven-link causal framework. Capability convergence and model-layer price pressure are well supported, but enterprise evidence shows selected workload migration and multi-model use rather than broad replacement. On OpenRouter, the short-run price-volume relationship is close to unit elastic. Clouds, accelerator suppliers, applications, and adopters can capture value displaced from proprietary laboratories. Model commoditisation therefore need not reduce AI expenditure or infrastructure demand. A broad correction requires technical substitution to lower enterprise spending and expected capacity utilisation. A systemic crisis further requires asset impairment, refinancing stress, and wider credit contraction. Debt-financed infrastructure and opaque contractual commitments make that tail risk more credible, but capacity scarcity, cloud growth, and strong hyperscaler balance sheets weigh against it. The leading outcome is rent compression and value migration, not an AI-wide demand collapse.