China’s open-source artificial intelligence (AI) models are rapidly gaining ground across key industries, marking a shift in strategy that has turned past setbacks into a new wave of opportunity.
According to research firm Frost & Sullivan, adoption is rising as major developers move away from simply boasting “extreme performance” and instead prioritize usability, cost efficiency, and broader ecosystem support.
Neil Wang, global partner and Greater China chairman at Frost & Sullivan, explained in an interview with the South China Morning Post on Thursday that developers such as DeepSeek, Alibaba Group Holding, and Baidu are now steering their AI systems toward practical applications that accelerate industry adoption.
Frost & Sullivan’s latest report on China’s AI market, also released Thursday, highlighted how quickly the nation has made progress in pushing relevant AI applications into more industries.
“AI applications are evolving from general capabilities to scenario-specific deployment,” the report said, noting that finance, government, telecommunications, and healthcare sectors have each achieved an average AI penetration rate above 60 percent.
The government sector leads with a 95 percent adoption rate, followed by finance at 78 percent. This growth has been driven in part by a sharp decline in model training costs—down 90 percent this year compared with 2024. Wang noted that this has dramatically lowered the barrier for companies to launch new AI projects.
This aggressive pivot toward open-source has helped Chinese AI firms narrow the gap with their U.S. counterparts. Companies enable third-party developers to modify, build on, and distribute their models by making source code openly available, fueling a wave of experimentation and fast-paced adoption.
Start-ups like DeepSeek and MoonshotAI, alongside Big Tech players such as Alibaba Cloud and Baidu, are seeing strong uptake of their models thanks to their affordability and innovative features. For instance, lower training costs have allowed DeepSeek to shift its focus from merely topping benchmark scores to delivering “manageable costs, adequate quality and a more user-friendly toolchain,” Wang said. Its V3.1 model now includes a “flexible hybrid reasoning mode” that automatically switches between “deep thinking” and “fast response” depending on task complexity.
China’s vast market in industrial AI applications gives it “a clear advantage” over the U.S., Wang added, emphasizing that domestic firms are “extremely fast at trial and error and technology iteration,” which accelerates commercial rollout. Still, bottlenecks remain. Many developers are heavily dependent on foreign providers for “foundational development software” such as AI frameworks and compilers, leaving vulnerabilities in the ecosystem.
The momentum behind this current wave of open-source adoption is deeply rooted in China’s earlier struggles with AI development. For years, U.S. export restrictions choked access to high-end semiconductors and advanced chips, particularly GPUs vital for training large models. Washington’s sanctions on Huawei in 2019 and later bans on Nvidia’s most powerful AI chips limited China’s ability to compete on hardware performance. That hardware shortfall left Chinese firms at a disadvantage in producing frontier models that could match U.S. giants like OpenAI, Anthropic, or Google DeepMind.
However, the scarcity forced a strategic rethink. Instead of relying solely on closed, resource-intensive systems, Chinese companies leaned into the open-source model, which democratized access and drew in a wider base of developers.
This approach allowed firms like Baidu, Alibaba, and newer entrants such as DeepSeek to leapfrog some of the hardware bottlenecks by mobilizing vast developer communities around low-cost, flexible tools. It also aligned with Beijing’s push for technological self-reliance, making open-source not just a technical strategy but also a political and economic one.
The growing adoption has forced U.S.-based AI companies to launch open-source. Earlier this month, OpenAI CEO Sam Altman admitted that China’s surge in open-source artificial intelligence played a key role in the company’s decision to release its own open-weight models.
“It was clear that if we didn’t do it, the world was gonna be mostly built on Chinese open-source models. That was a factor in our decision, for sure. Wasn’t the only one, but that loomed large,” CNBC quoted him as saying.
Thus, China has created a model that prioritizes adaptability, affordability, and rapid scaling—traits that are accelerating adoption across industries at a pace even faster than in the U.S, by championing open-source systems.
