Chinese AI model developers seek fast commercial adoption as training costs fall

Open-source artificial intelligence (AI) models from China are seeing increased adoption across various industries, according to research firm Frost & Sullivan, as major developers step up the commercial application of their systems.

Developers such as DeepSeek, Alibaba Group Holding and Baidu have shifted from pushing their models’ “extreme performance” towards prioritising usability, cost efficiency and broad ecosystem support to accelerate industry adoption, according to Neil Wang, global partner and Greater China chairman at Frost & Sullivan, in an interview with the South China Morning Post on Thursday. Alibaba owns the Post.

According to Frost & Sullivan’s latest report on China’s AI market released on Thursday, the nation has made quick progress in getting more industries to adopt the relevant AI applications.

“AI applications are evolving from general capabilities to scenario-specific deployment,” the report said. It pointed out that the country’s finance, government, telecommunications and healthcare sectors have each achieved an average AI penetration rate in excess of 60 per cent.

The government sector saw the highest adoption rate of 95 per cent, followed by finance with a 78 per cent rate.

Alibaba Cloud, the AI and cloud computing unit of Alibaba Group Holding, has built its Qwen family of models into the world’s largest open-source AI ecosystem for developers. Photo: Shutterstock

The trend has emerged amid a substantial decline of model training costs, which has lowered the barrier to entry for pursuing AI projects, according to Wang. Training costs have seen a 90 per cent drop this year when compared with 2024, he added.

Темы и теги

Китай Наука и Технологии

разработчик модель коммерческий принятие затрата обучение китайский быстрый снижение Frost & Sullivan связь