US enterprises including DoorDash, Airbnb, and Siemens are adopting Chinese AI models to replace American suppliers like Anthropic's Claude and OpenAI's ChatGPT. The shift is driven by cost-effectiveness, as one organization reportedly spent $500 million per month on Claude usage, while AI-intensive companies average approximately $7,500 per employee monthly on AI applications according to Ramp AI Index research. Chinese AI models from startups like Moonshot AI, DeepSeek, and Zhipu offer comparable capabilities at significantly lower costs, with the added advantage of open-weight architectures that allow enterprises to customize models and maintain greater control over sensitive data processing.
DoorDash and Major Enterprises Adopt Chinese AI Models for Cost Savings
DoorDash co-founder Andy Fang stated 上週 on X platform that the company achieved substantial cost savings by using Moonshot AI's models for lower-tier tasks. San Francisco startup Lindy has completely abandoned Anthropic's AI tools in favor of DeepSeek's latest V4 model. Chinese AI company Zhipu released its GLM-5.2 model 上個月, attracting significant attention from Silicon Valley industry players. DeepSeek and Zhipu's leading models demonstrate capabilities exceeding or approaching US counterparts while maintaining significantly lower costs.
Open-Weight Models Enable Enterprise Customization and Security Control
The open-weight characteristic of Chinese models allows enterprises to access model parameters and numerical values completely, enabling customization according to specific business needs. Eugene Cheah, CEO of OpenRouter, Inc. platform Featherless.AI, told media that enterprises have realized they do not need to pursue the "best" model, but can choose faster and cheaper alternatives. The open architecture provides enterprises with deeper understanding of how models process sensitive company data, enhancing cybersecurity control.
Industry Experts Confirm Cost-Effectiveness of Chinese AI Models
Sam Bresnick, researcher at Georgetown University's Center for Security and Emerging Technology, stated that Chinese models are sufficiently practical for most enterprise workloads, creating incentives for companies to transfer operations to more economical models. He questioned: "Why pay premium fees for Anthropic and OpenAI models when Chinese models are generally viable?" Aidan Gomez, CEO of Canadian AI company Cohere, noted that US AI model suppliers face credibility challenges, particularly after the first Trump cabinet temporarily suspended access to Anthropic's Mythos model overseas, exposing potential risks of over-reliance on single entity suppliers.
FAQ
Which US companies are switching to Chinese AI models?
DoorDash, Airbnb, and Siemens are among the major enterprises adopting Chinese AI models. DoorDash co-founder Andy Fang confirmed 上週 that the company uses Moonshot AI's models for cost savings, while San Francisco startup Lindy has completely replaced Anthropic tools with DeepSeek's V4 model.
Why are US enterprises choosing Chinese AI models over American suppliers?
Cost reduction is the primary driver, as one organization spent $500 million per month on Claude usage according to media reports. Chinese models from DeepSeek, Moonshot AI, and Zhipu offer comparable capabilities at significantly lower costs. Additionally, open-weight architectures allow enterprises to customize models and maintain greater control over sensitive data processing, addressing both cost and security concerns.
