Goldman Sachs significantly raised its revenue forecast for China's large language models: ARR is expected to increase from $10 billion to $13 billion by the end of 2026, and reach $125 billion by 2030.

In his latest research report on August 3rd, Ronald Keung, an analyst with the Goldman Sachs Asia Pacific Internet team, pointed out that the revenue realization speed of Chinese large model manufacturers has exceeded Wall Street's previous predictions. Meanwhile, a Jefferies research report shows that Chinese models have continuously dominated the top five in OpenRouter platform call volume for 14 weeks, with DeepSeek V4 Flash consuming 7.22 trillion tokens in a single week to top the global list. Goldman Sachs believes that Chinese large models are moving from a "cost-performance narrative" to a "revenue realization narrative" inflection point, primarily driven by the reshaping of token consumption patterns in Agent and coding scenarios.