Explained: Why China's lead in open source AI models is a concern for US

Over 80 per cent of surveyed information technology decision-makers reported that at least a quarter of their company's AI platforms are based on open source

Don't want to miss the best from Business Standard?

China surpassed the US in the global market for 'open' artificial intelligence (AI) models, the Financial Times reported, citing a new study by the Massachusetts Institute of Technology (MIT) and Hugging Face.

 

According to the report, downloads of new Chinese open models reached 17 per cent over the last year, higher than the 15.8 per cent share of American models for the first time. This provides Beijing with an important advantage in how this powerful technology is shaped and used worldwide.

 

Let's take a look at what open AI models are and why China's rise has become a matter of concern for the US.

What are open AI models?

Open AI models are systems that anyone can download, change, and use. They usually come with the model’s learned patterns (weights), some code or data, and instructions on how to use them. Because they are open, startups can build apps faster, researchers can test new ideas, and developers can create products without spending huge amounts of money. They offer a more public and accessible way of building AI compared to fully closed models.

What are the benefits of open-source AI?

According to a recent IBM study, over 80 per cent of surveyed information technology decision-makers reported that at least a quarter of their company’s AI platforms or solutions are based on open source. It also revealed that enterprises adopting open-source ecosystems are more likely to achieve goals. Here are some of the key advantages of open AI models:

• Accessibility: Open AI lets anyone use and learn AI without needing big money or special skills.
• Collaborative innovation: People from everywhere work together to improve open AI models and create new ideas.
• Cost-efficiency: Open-source AI is usually free, helping companies avoid high development or subscription costs.
• Customisation: Organisations can change open AI models to fit their own needs and tasks.
• Transparency: It is easy to inspect, helping people trust these models and making it simpler to find problems, biases, or security issues.

Why are open models important?

Open models matter because they help small companies innovate quickly and provide alternatives to expensive, tightly controlled AI tools such as Anthropic and OpenAI. China's rise assumes significance given that these models also shape the kind of information people around the world receive. So, whichever country’s models are widely used gains soft power in the AI space.

Why is China ahead?

China’s share of open-model downloads has grown quickly for several reasons. According to the Financial Times, Chinese companies release new models very often, sometimes every week, unlike US labs that update every few months. Because China cannot access Nvidia’s most advanced chips, the government encourages companies to open their models so the community can improve them.

 

China also has many strong AI researchers who are creating powerful and efficient models. DeepSeek’s R1 model surprised Silicon Valley because it matched US models despite using far less computing power.

How does US' approach differ?

Most of the US companies, including AI giants OpenAI, Google, and Anthropic, prefer closed models. They keep their best technology private and earn money through subscriptions. Even Meta, which used to lead in open models with Llama, is now focusing more on closed, advanced systems.

 

While the US government has encouraged American companies to release open models with 'American values,' the progress has been slow. So far, only the Allen Institute for AI has released a fully open model called Olmo 3.

Who leads Chinese open models?

According to the report, DeepSeek and Alibaba’s Qwen models dominate China’s open-model downloads. These two companies produce most of the Chinese models used around the world, the data from MIT and Hugging Face revealed.

China's rise may shift narratives

Studies showed that Chinese open models often reflect the government's viewpoints. They typically avoid sensitive topics like Taiwan or Tiananmen Square and follow official narratives. Concerns remain that as these models spread globally, they can influence the type of information users receive, even indirectly.

 

This was highlighted after DeepSeek's meteoric rise. When it came to political questions, DeepSeek's Chinese version mostly refused to answer or followed strict government narratives.

 

Additionally, China's growth in the open-model space could set global AI standards. Many countries may adopt Chinese models simply because they are free, fast to release, and widely available. Since US companies are focused on expensive, closed systems, they risk losing influence in the open AI ecosystem.