China has taken steps to continue the positive moves it made concerning artificial intelligence in 2025 in the new year. According to reports, DeepSeek published a technical paper, which was co-authored by 19 people, including its CEO and founder, Liang Wenfeng, about “manifold-constrained hyper-connections.”
The paper discussed the general framework for training artificial intelligence systems on a large scale, suggesting a promising direction for the future evolution of foundational models in the country. The release was a reminder to the world, especially during the peak of the festivities, about Chinese AI firms’ sharpened focus on innovation and their willingness to stay ahead in the fast-developing industry.
China makes positive push to continue AI domination in 2026
The release of the paper came at the same time as last year, when DeepSeek announced itself to the global population. The platform began to catch the attention of the world after releasing its large language model (LLM) DeepSeek V3. Weeks later, the company released its reasoning model, DeepSeek-R1, on January 20. The two models matched or surpassed the performance of rival models across a range of benchmark tests, another metric that caught the eye of the global population.
In addition, they were built at a fraction of the cost and computing power that major United States tech firms invest in building LLMs. The DeepSeek unveiling resulted in a massive sell-off on January 27, wiping out nearly $1 trillion in tech stocks, with Nvidia alone recording a $600 billion loss. Meanwhile, analysts have predicted that AI companies in China have the momentum to continue their impressive prowess this year, thanks to Beijing’s policy support.
Aside from favorable policies, the firms have also enjoyed improved funding prospects, greater adoption of AI systems across industries, and a growing number of talents being recruited for these projects.
A domestic AI startup co-founder, who spoke under anonymity, predicted that China could overtake the United States to become the world’s leading AI power in 2027. The co-founder named China’s talent pool as its major advantage in the race.
Xi Jinping highlights healthy competition in the AI industry
During his New Year’s address, Chinese President Xi Jinping mentioned that the market has seen many large AI models competing in a race to the top, while noting that breakthroughs were also being achieved in domestic semiconductor development. He highlighted that all of that has turned China into one of the economies with the fastest-growing innovation capabilities.
Speaking about China’s innovation, Winston Ma, an adjunct professor at the New York University School of Law, with a focus on AI and the digital economy, mentioned that the country is prepped for a policy-driven innovation in 2026. Ma added that things could go well, considering that AI is placed at the center of China’s economic agenda and industrial upgrading plans.
China currently plays host to a strong group of players in the AI industry developing powerful open models beyond DeepSeek, according to Stanford University’s DigiChina Project. The project is currently under the school’s Centre for International Security and Cooperation, and its Institute for Human-Centered Artificial Intelligence, which compiles the annual AI Index report.
They include the Qwen model developer Alibaba Cloud and startups like Moonshot AI, MiniMax, and Zhipu AI, which is known internationally as Z.ai.
According to the report, China’s open-source AI models may have even caught up or pulled ahead of their United States counterparts in capabilities and adoption. For instance, Facebook’s parent company Meta Platforms was said to be using Alibaba Cloud’s open-source Qwen model for the training process for a new model called Avocado.
If you're reading this, you’re already ahead. Stay there with our newsletter.
