China's open-source AI revolution: Game-changer for innovation

CNBC explores in a new article that China is increasingly embracing open-source AI models, a trend that experts say is accelerating AI innovation and adoption across the country.

Industry insiders even compare it to an "Android moment" for the sector, signaling a major shift in how AI is developed and used.

At the forefront of this transformation is AI startup DeepSeek, whose R1 model, launched earlier this year, has made waves by challenging the dominance of American tech giants. The release has raised questions about the massive investments made by companies like OpenAI and Google in large language models and data centers. While R1 attracted attention for its impressive performance and cost-effectiveness, some analysts believe the model’s real impact lies in driving the widespread adoption of open-source AI.

“DeepSeek’s success proves that open-source strategies can lead to faster innovation and broad adoption,” said Wei Sun, principal analyst of artificial intelligence at Counterpoint Research. “Now, we see that R1 is actively reshaping China’s AI landscape, with large companies like Baidu moving to open source their own LLMs in a strategic response.”

Baidu released the latest version of its AI model, Ernie 4.5, along with a new reasoning model, Ernie X1, making both available for free to individual users. Baidu also announced plans to make the Ernie 4.5 series open-source by the end of June. Experts view this as part of a larger shift in China away from proprietary licensing toward more open-source approaches.

Lian Jye Su, chief analyst at Omdia, noted that Baidu’s embrace of open-source is significant. “Baidu has always supported its proprietary business model, but disruptors like DeepSeek have shown that open-source models can be as competitive and reliable,” Su said.

DeepSeek’s R1, distributed under an ‘MIT License,’ is one of the most permissive open-source licenses, allowing for modification and distribution without restrictions, including for commercial purposes. While DeepSeek’s model is free, the startup charges for its Application Programming Interface (API), which allows integration with other companies' applications at a lower cost than competitors like OpenAI and Anthropic.

In contrast, OpenAI and Anthropic’s models are considered closed-source, with their datasets and algorithms not publicly accessible, generating revenue by charging for access.

By Naila Huseynova