Chinese artificial intelligence companies are rapidly escalating competition, releasing a wave of new models and incentives aimed at attracting users, and developers. The surge in activity, particularly around the Lunar New Year holiday, underscores China’s growing ambition to challenge US dominance in the AI sector, and is increasingly focused on practical application rather than purely theoretical advancement.
ByteDance, Alibaba, and Moonshot are among the key players unveiling new capabilities. ByteDance’s Seedance 2.0, a video-generating model, has already caused ripples in Hollywood with its ability to create multi-scene clips featuring realistic imagery and synchronized sound. Tiezhen Wang, an engineer at Hugging Face, described the launch as a “DeepSeek moment,” referencing the breakthrough achieved last year by Chinese start-up DeepSeek with its powerful large language model.
The focus on application-building differentiates the Chinese approach, according to Ritwik Gupta, an AI researcher at the University of California, Berkeley. “Chinese labs are getting better at building models that are useful for making applications. They largely view AI as a tool for building products, in contrast with the US labs, which view it as a race for ‘frontier’ dominance first, product second,” he said.
Alibaba has been particularly prolific, releasing over 400 open-source models since 2023. The company recently launched its Qwen 3.5 model alongside a substantial investment of Rmb3bn ($431mn) in subsidies designed to encourage usage of its AI app, Qwen. The initial response was overwhelming, with users rushing to claim freebies – including bubble tea – through the app, temporarily straining participating stores.
This push towards developer-focused tools is evident in Alibaba’s new model, which is specifically tailored for building AI agents – systems capable of independently operating computers and completing complex tasks. DeepSeek is also expected to release a new model in the coming months, building on the momentum generated by its earlier success.
The rise of Chinese AI is also attracting developers frustrated by limitations and tighter controls imposed by US companies like Anthropic and OpenAI. Moonshot’s Kimi 2.5, launched in late January, is positioned as a high-performing coding system with fewer usage constraints. The company has open-sourced Kimi, allowing developers to run it on various cloud platforms or their own servers.
“Chinese companies are clearly leaning into what developers want and building models to address that,” Gupta explained. “Developers want freedom. They do not want to be locked in by any AI model provider that demands they use their services in a specific way.” Moonshot has demonstrated a willingness to capitalize on opportunities created by its US rivals, releasing migration toolkits to help developers transition from OpenAI’s systems to Kimi when OpenAI retired one of its flagship models last year.
An executive at US-based Perplexity, an AI search engine offering users the ability to switch between different underlying models, noted an increasing preference for Chinese open-source systems among its user base.
However, the rapid expansion isn’t without challenges. The Motion Picture Association, representing major film studios, recently demanded that ByteDance address “infringing activity” after users leveraged Seedance to generate content incorporating copyrighted intellectual property. ByteDance responded with a commitment to strengthen safeguards against unauthorized use of IP.
Despite these concerns, the technological advancements are significant. Wang at Hugging Face highlighted that Seedance represents a substantial improvement over earlier models, particularly in its ability to handle complex scenes. He pointed to a widely shared clip featuring Chinese filmmaker Jia Zhangke interacting with an AI version of himself as an example of the tool’s capabilities, showcasing control over shot angles, framing, and sound.
The competitive landscape is shifting, and China’s focus on practical application, coupled with its commitment to open-source development, presents a compelling challenge to the US’s traditional dominance in the AI space. One analyst, speaking to CNBC earlier this week, predicted that most of the world’s population could be running on a Chinese tech stack within five to ten years, a sentiment that underscores the growing influence of Chinese AI on the global stage.
While the US maintains advantages, particularly in frontier AI research, the rapid progress in China suggests a more balanced competition is emerging, one where cost-effectiveness and accessibility could prove to be decisive factors.
