China’s AI revenue projected to reach US$13b on breakthroughs, adoption: Goldman Sachs

Goldman Sachs has raised its forecast for China's AI model market revenue by 30 percent, projecting it to reach US$13 billion. The investment bank attributes this growth to rising cost efficiency, breakthroughs in capabilities, and accelerating adoption among corporations. In a research note, Goldman Sachs updated its year-end annual recurring revenue (ARR) projection for mainland Chinese AI models from US$10 billion to US$13 billion, citing aggressive price cuts, technical advancements, and wider corporate adoption.

The report highlights the intense competition among Chinese AI models in terms of performance per dollar, following recent releases like MiniMax's H3 model, priced at 30 to 50 percent of incumbent market levels, and DeepSeek's V4 Flash model, achieving front-end coding capabilities comparable to Zhipu's GLM-5.2. Alibaba Group Holding has also launched its massive 2.4-trillion-parameter Qwen3.8 Max model, which ranks fourth in global front-end coding capabilities among open-weight models.

These cost-effective offerings have driven Chinese open-weight models to command nearly 70 percent of global token consumption on platforms like OpenRouter. Tencent Holdings' WorkBuddy and ByteDance's TRAE have also made significant strides in the workspace agentic tools market, capturing substantial market shares. Chinese developers are expected to shift towards "community licenses" for open-weight models, requiring commercial users to pay fees.

While the market is expected to see pressure on API pricing and gross margins during the second half of the year, Goldman Sachs warns that geopolitical constraints may further impact the industry, with reports indicating that Chinese authorities are considering restrictions on foreign downloads of model weights and overseas transfers of training data.