A growing number of companies are turning to Chinese artificial intelligence models as rising operating costs make premium American AI services increasingly expensive, according to a report by The Financial Times. The shift is being driven by businesses seeking more cost-effective solutions without significantly compromising performance.
The report says organisations ranging from Silicon Valley startups to European industrial firms are adopting Chinese AI models for routine business tasks. Companies including DoorDash, Airbnb and German engineering company Siemens are among those reported to be using AI systems developed by Chinese firms.
A key factor behind the trend is the industry’s move from flat-rate subscriptions to usage-based pricing. As businesses expand their AI deployments, operating costs have risen sharply, prompting many to reconsider whether they need the most advanced—and most expensive—AI models for everyday applications.
Instead, companies are increasingly relying on Chinese models developed by firms such as DeepSeek, Moonshot AI and Z.ai for tasks involving text generation, coding and data processing. According to the report, these models now offer performance that is competitive with leading American alternatives while costing substantially less to operate.
DoorDash has reportedly redesigned its AI infrastructure to reduce expenses. Co-founder Andy Fang said the company now uses Moonshot AI’s Kimi K2 model for routine workloads, while reserving Anthropic’s more advanced models only for complex tasks that require higher reasoning capabilities.
Some companies have gone a step further by replacing American AI models entirely. San Francisco-based startup Lindy reportedly switched from Anthropic’s models to DeepSeek V4, with founder Flo Crivello describing the move as transformative because it both reduced costs and improved day-to-day performance.
Narrowing Performance Gap
The report suggests that the performance gap between leading American and Chinese AI models has narrowed considerably in recent months. Venture capitalist Marc Andreessen has publicly stated that some Chinese models are now viewed within the industry as capable of matching—or even surpassing—the performance of leading American systems in certain use cases.
Another advantage highlighted in the report is that several Chinese models are released as open-weight systems. Unlike proprietary cloud-only models, open-weight AI allows organisations to deploy the models on their own infrastructure, keeping sensitive data within private servers while customising the software to suit their specific requirements.
European Firms Seek Greater Independence
In Europe, the shift is also being influenced by strategic concerns. The report notes that recent US export controls on some advanced AI technologies have prompted European companies to reassess their dependence on American providers.
As a result, many businesses are exploring a broader mix of AI platforms, including Chinese alternatives, to reduce operational risks, improve cost efficiency and avoid relying too heavily on a single country’s technology ecosystem.
