China banks on open source in AI cold war with US

The United States has been trying to counteract the popularization of technological solutions from China for years, often taking steps that are contrary to the development of an open market. Further tariffs and the inclusion of large companies on the Department of Commerce blacklists limit competition while favoring solutions from the US.

China is not remaining passive and is increasingly overcoming American sanctions by gaining new market segments. After more than two years of domination by US companies in the arena of artificial intelligence, the time has come for a Chinese attack preceded by many months of preparations coordinated by Beijing.

China’s largest tech giants such as Baidu and Tencent have been working on their own large language models (LLMs) for some time now, and the unexpected debut of DeepSeek, despite numerous hardware sanctions, surprised even the most powerful AI companies from the West.

[ See also: US GPU export limits could bring cold war to AI, data center markets ]

China has not said its last word yet. While the US is closing its LLM models and restricting access to source code, as well as tightening sanctions, China is following a completely different path. Surprisingly, China’s AI sector is opening up, focusing on the open-source model, which is also crucial in the context of cybersecurity concerns.

In recent weeks, Chinese tech giants like Alibaba, Baidu, and Tencent have flooded the market with new AI models that can be freely downloaded, modified, and integrated with their own solutions via publicly available APIs.

China is pushing the boundaries of its own AI development every few weeks, and its results are already a serious threat to Western technology. While in the West AI was an exclusive and lucrative resource available only to select companies, China, thanks to its open approach, bypasses sanctions, decentralizes development, and uses available resources for mass AI development. And thanks to the development of AI in the open-source model, hardware availability is no longer a problem, because models are voluntarily tested and improved by their users, e.g., from Europe.

If open-source AI becomes as powerful as US proprietary models, the ability to monetize AI as an exclusive product will collapse, which is a key consideration for China. The opening of AI models from China is a surprising move that may also disrupt the foundations of OpenAI’s classic AI model development — based on a small group of companies capitalizing on the latest technological advances.

China may be the country that proves that it doesn’t always pay to be first. Its approach could change the balance of power in the development of artificial intelligence. AI will also be an important weapon in the US-China technological war. The coming months will be extremely turbulent in the context of AI development, in addition to the traditional facets of US-China geopolitical and economic friction.

This article appeared originally in Polish on Computerworld Poland.