Companies Turn to Chinese Open-Source AI for Lower Costs and Data Control

Facing high fees from American AI giants, companies are looking to open-source alternatives produced mainly by Chinese developers. Models from Alibaba, DeepSeek, Moonshot and Z.ai can be downloaded and operated on privately managed hardware, allowing organisations to retain control over sensitive data. Driven Lawyers in Amsterdam, for example, runs a scaled-down Qwen model on German infrastructure to automate legal contract summaries, avoiding the expense of premium U.S. services.

Philip Gast of The Stack notes that many tasks do not require the most powerful models, and that hosting open models on third-party servers can still be cheaper than building in-house hardware. OpenRouter, a global provider, now offers a catalogue of such models, reflecting growing usage worldwide. The shift is driven by both cost considerations and regulatory pressures to keep data within European jurisdictions, especially in sectors like health and defence.