China is using open-source technology to reduce its reliance on the West

China is putting more focus on open-source software and hardware as it tries to reduce its reliance on technology from the US and Europe. The country has backed its own open-source projects, developed alternatives to platforms such as GitHub and increased its involvement in technologies including RISC-V and Open RAN. Chinese companies are also releasing AI models that can be used by developers outside the country. This is partly about having more control over the technology China uses. It has also become more important as Chinese companies face restrictions on accessing some US technology.

Huawei showed the value of open source

Huawei's experience is a good example. After the US government blacklisted the company, Google removed Huawei's access to its licensed services and apps. Huawei could still use the open-source parts of Android. That allowed the company to continue using Android's underlying code while it worked on its own operating system. Huawei eventually developed HarmonyOS Next and released OpenHarmony, an open-source version of the operating system. China had already been developing alternatives to foreign software before this. Red Flag Linux was developed as an alternative to Windows, while Kylin was later created for China's military and government organisations. But the restrictions placed on companies such as Huawei gave open-source technology a more practical role.

China has been building its own platforms

The Chinese government has made technological independence a priority. Government-backed organisations such as the OpenAtom Foundation have supported open-source projects in China. China also has Gitee, a software development platform similar to GitHub. OpenKylin and openEuler are among the open-source operating systems being developed and used in China. These projects give Chinese companies and government organisations alternatives to foreign software. They also reduce the number of areas where a foreign company can become a single point of dependence. The same thinking is being applied to chips.

RISC-V could reduce dependence on foreign chip technology

RISC-V is an open architecture used to design computer processors. Unlike proprietary architectures, companies can use RISC-V without having to rely on a single company controlling the underlying technology. Chinese companies have become increasingly involved in RISC-V as US restrictions make access to some advanced semiconductor technology more difficult. RISC-V does not solve China's semiconductor problems by itself. Companies still need chip-design expertise, manufacturing capacity and access to the equipment needed to produce advanced chips. But using an open architecture can reduce one part of that dependence. China is also involved in Open RAN, which uses open interfaces between different parts of mobile networks. Chinese telecoms companies are taking part in work around 6G as well. China already has a large presence in 5G equipment. Its involvement in future network standards gives its companies a role in deciding how some of the technology will be built.

Chinese AI models are raising another set of questions

AI has added another part to China's open-source push. Chinese companies have released models including Qwen, DeepSeek, Kimi and GLM. Many of these models are open weight rather than open source. That means the model's parameters may be available to users, but the company does not necessarily release the training data or the methods used to train the model. For a developer who simply wants to download and run a model, the distinction may not make much difference.

 For a company using the model in a commercial product, it can matter. There are also concerns about how some Chinese AI companies have developed their models. Anthropic and OpenAI have accused Chinese AI companies of using distillation to reproduce capabilities from proprietary models. US lawmakers have proposed restrictions on some Chinese open-weight AI models, while government agencies and some state governments have restricted the use of DeepSeek over security concerns. That means companies looking at Chinese AI models have more to consider than performance and cost.

The licence can come with conditions

Another issue is the licence. Open source software can be freely available while still having rules about how it can be used. Some Chinese AI models use common licences such as MIT and Apache 2.0. Others have additional requirements. Moonshot's Kimi K3 licence, for example, requires companies that exceed certain revenue or user thresholds to prominently display the Kimi name. MiniMax's M3 community licence also includes restrictions based on applicable laws and regulations. China has developed its own licences too, including the Mulan Permissive Software Licence and the OpenAtom Model Licence. Both are available in Chinese and English. Where the two versions conflict, the Chinese version takes precedence. For an individual developer, this may not be a major concern. For a company building a product around one of these models, the terms can become more important.

Open source is now part of China's technology strategy

China is not simply developing open-source software because it is cheaper or easier to share. The technology is also useful for reducing dependence on foreign companies. That is why the country is putting effort into operating systems, chip architectures, telecoms standards and AI models. The approach has its limits. Open source does not give China everything it needs to build advanced technology. It still needs hardware, talent, manufacturing capacity and access to equipment. But it does give Chinese companies more room to develop technology without depending entirely on foreign suppliers. For businesses using Chinese open-source software or AI models, that creates a practical issue. They need to look at the technology itself, but also at its licence, restrictions and the company or organisation behind it.