China’s tech leaders, including Alibaba’s Eddie Wu, Tencent’s Pony Ma, and Baidu’s Robin Li, have embraced open-source AI, allowing anyone to freely use, modify, and distribute their artificial intelligence software and code. This approach has been met with support from the Chinese government. In January, DeepSeek’s CEO Liang Wenfeng was recognized as the AI sector representative during a meeting with Premier Li Qiang, signaling Beijing’s endorsement of this model.
Although open-source AI is not a uniquely Chinese innovation, the country’s offerings align more closely with the standard definition. DeepSeek’s source code is licensed to permit near-unrestricted use. In contrast, OpenAI—the U.S. firm behind ChatGPT—has tightly guarded its proprietary models' training data and methodologies, despite plans to release a more accessible model in the coming months. Even Meta’s Llama models, though freely available, impose some limitations on commercial applications. Meta, however, shares a similar belief that widespread accessibility is key to setting industry standards across multiple generations.
China's embrace of open-source AI is not solely a philanthropic endeavor; it serves a strategic purpose. Washington’s technological restrictions have blocked Chinese companies from acquiring the most advanced chips from Nvidia, essential for training and deploying AI models. By leveraging open-source models developed by companies with access to these powerful processors, China has found a workaround. Before DeepSeek’s rise, most Chinese AI models—including those developed for military use—were adaptations of Meta’s Llama.
Pooling resources through open-source collaboration provides China with a mechanism to accelerate its progress in AI, closing the gap with the United States. In recent weeks, leading Chinese firms—Baidu, Alibaba, Tencent, and DeepSeek—have unveiled updates and new releases of their open-source AI models. For instance, Jack Ma-founded Ant Group has pioneered techniques to train AI models using Huawei’s less powerful domestically manufactured chips, achieving results comparable to those generated by Nvidia’s processors. If widely adopted, such breakthroughs could bring China closer to President Xi Jinping’s vision of technological self-sufficiency.
The potential of open-source AI to bridge technological gaps has also captured the attention of other regions. The European Union, home to AI startups like France’s Mistral AI, has begun championing a similar approach. In February, European Commission President Ursula von der Leyen announced plans to mobilize €200 billion to invest in “cooperative, open innovation” for AI development.
Beyond economic and technological benefits, China’s generosity with open-source AI enhances its global reputation. DeepSeek’s founder Liang Wenfeng stated in a rare interview last year that contributing to open-source AI “earns us respect.” This sentiment extends to China as a whole. By providing powerful, freely available AI models, the country has strengthened its soft power beyond the Western world. American economist Tyler Cowen recently remarked that China has gained an advantage over the United States “not only in technology but also in vibes.”
China’s commitment to open-source technology extends beyond AI. The government has actively promoted the adoption of RISC-V, an open-source chip architecture backed by Huawei, Nvidia, and others. This initiative reduces reliance on proprietary designs from UK-based Arm and U.S. giants like Intel and AMD, whose technologies could be restricted by Washington at any time. However, there are risks associated with China’s open-source approach. The model challenges the conventional business logic of AI firms, which typically generate revenue by charging for access and licensing proprietary technologies.
DeepSeek, for example, relies solely on integration fees for revenue, as its open-source model prevents it from charging for access. While this might not pose a significant concern for Liang Wenfeng, who prioritizes innovation over profit, it presents challenges for publicly traded firms like Alibaba. The e-commerce and cloud computing giant has pledged approximately $53 billion in AI and cloud investments. A lack of financial returns could negatively impact its stock price and valuation.
Alibaba, which offers both proprietary and open-source AI models, aims to balance these competing interests. At HSBC’s Global Investment Summit, Alibaba Chairman Joe Tsai explained that while customers might utilize free AI models, they will still purchase computing power, data handling services, security solutions, and a comprehensive software suite from Alibaba. This strategy assumes that Chinese businesses—currently lagging in AI and IT adoption—will increase their spending over time.
However, the greatest threat to open-source AI in China may come from within. Beijing maintains strict control over its economy through centralized policies and regulatory oversight. Generative AI products must adhere to “core socialist values” and avoid content that “endangers national security.” The ambiguity surrounding how these rules apply to open-source AI—by nature decentralized and borderless—raises concerns about the future of China’s approach.
As Chinese companies continue to match or surpass their Western AI rivals, the government may reconsider its willingness to share such powerful technology. Reports indicate that some DeepSeek employees have been placed under travel restrictions. Additionally, Gregory C. Allen from the Center for Strategic and International Studies has warned that DeepSeek’s advanced AI training techniques could ultimately benefit U.S. firms more than Chinese ones, given America’s superior computing resources.
Notably, China’s open-source philosophy does not extend to industries where it already leads globally. In sectors like electric-vehicle batteries and green energy, Beijing has taken protective measures. In 2023, the government banned the export of certain rare earth processing technologies. More recently, reports surfaced that it delayed automaker BYD’s plans to build a factory in Mexico over concerns that proprietary technology might leak to the United States.
For now, China’s supply of free and powerful AI models is abundant. However, as geopolitical and economic factors evolve, the era of China’s open-source AI generosity may not last forever.