DeepSeek slashes AI model costs 50% amid China price war

DeepSeek's release of V4-Flash at a steep discount reflects the intensifying competitive pressure within China's AI sector. The 50% cut in token costs and the decision to pause the previously planned dynamic-pricing mechanism signal that the company is prioritizing market share over profitability — a classic feature of the "involution" race that Beijing has publicly cautioned against. This race-to-the-bottom dynamic has historical precedent in China's solar and electric-vehicle industries, where similar price wars eroded margins across the market. However, the government's position on involution is complex: while Beijing warns against it, experts observe that state subsidies to compute and energy purchases effectively enable the practice by keeping unprofitable companies solvent. DeepSeek's move suggests that market pressures are outweighing those official warnings, at least for now.