Businesses Explore Cheaper Chinese AI Models Amid Rising US Costs

Businesses are increasingly turning to less expensive artificial intelligence models developed by Chinese companies as the cost of leading U.S. AI services rises. This shift is driven by the need to manage escalating expenses associated with token usage and AI operations, prompting some consumer-facing companies to explore open-source alternatives from China. DoorDash, for instance, is experimenting with DoorDash CLI, an AI agent tool that can be accessed via terminal. Andy Fang, DoorDash's co-founder and CTO, stated on X this week that Moonshot AI's models offer superior quality at a lower cost.

Moonshot AI is not the only Chinese AI provider gaining traction. Cursor, an AI coding startup, utilized Moonshot's Kimi model to develop its Composer 2 coding agent. Additionally, the startup Lindy has reportedly transitioned from using tools provided by Anthropic to DeepSeek's V4 models, according to the Financial Times. These companies are joining larger entities such as Airbnb and Siemens, which are also investigating the integration of Chinese AI providers like Alibaba and DeepSeek to mitigate rising AI expenditures.

According to Yasir Atalan, deputy director and data fellow at the Center for Strategic and International Studies, this trend is influenced by three key factors: cost, capability, and the availability of open-source models. Atalan told Fortune that high-performance models from U.S. companies appear costly in comparison to their Chinese counterparts. The appeal of open-source models is particularly strong for countries outside the U.S., as it allows them to avoid sharing sensitive enterprise data.