Microsoft has quietly become one of the most important channels through which OpenAI models reach the Chinese market, even as OpenAI and Anthropic avoid selling their models directly in China due to concerns over intellectual property protection and potential misuse.
According to a Bloomberg report, Microsoft is selling GPT models through Azure to some of China’s largest internet companies. This gives Microsoft a unique position among American AI vendors: it is providing Chinese firms access to models that their original developers are not willing to offer directly in the country.
Major Chinese Companies Are Using OpenAI Models Through Microsoft
The scale of the arrangement is significant. ByteDance, the parent company of TikTok, has reportedly been Microsoft’s largest AI customer in recent years, with much of its usage relying on OpenAI models. The company is also on track to spend more than US$1 billion annually on Microsoft’s AI and cloud services.
Other major Chinese companies, including Ant Group, Meituan, and Tencent, also buy AI models through Azure. However, Ant Group has stated that it develops its own models and that its core products do not rely on external AI systems.
Azure AI Revenue in China Is Growing Rapidly
Inside Microsoft, the company’s AI growth in China has been viewed as a major commercial success. Azure’s AI revenue in China reportedly grew faster than in any other sales territory, roughly tripling in the financial year that ended in June 2025 after rising by about 400% the previous year.
Judson Althoff, Microsoft’s former chief commercial officer, described the company as the one bringing together the two major AI hubs: the US West Coast and eastern China. Microsoft President Brad Smith has also told US lawmakers that the company’s China business represented around 1.5% of Microsoft’s total revenue in 2024.
Why OpenAI Models Reach China Through Microsoft
The reason lies in Microsoft’s unique contract with OpenAI, which allows it to set its own terms for selling GPT models internationally.
While OpenAI and Anthropic have chosen not to sell directly into China, Anthropic’s models are not part of Microsoft’s China offering. This leaves Microsoft acting as a key intermediary, giving Chinese companies access to advanced American AI models despite the original developers’ reluctance to serve the market directly.
Concerns Over Distillation and Model Misuse
The main concern is that Chinese companies could use outputs from OpenAI models to train competing systems, a technique known as distillation. Bloomberg reported that OpenAI has privately urged Microsoft to take stronger action to prevent Chinese customers from using GPT models in this way.
Microsoft says it uses automated monitoring systems and sells only to established companies rather than individual developers. However, sources cited by Bloomberg said Chinese buyers do not face additional scrutiny, while tracking synthetic data generated by AI models remains difficult.
To reduce regulatory and political exposure, Microsoft does not host OpenAI models inside China. Instead, Chinese customers access them over the internet through data centers located outside the country, including in Singapore.
Microsoft Is Selling American Models to China and Chinese Models to the West
The situation becomes more complex when looking at Microsoft’s broader AI strategy. In January 2025, Microsoft added DeepSeek’s R1 model to Azure AI Foundry. It has also confirmed that it is testing a fine-tuned, Azure-hosted version of DeepSeek-V4 as a lower-cost option for Copilot Cowork, an enterprise agent currently powered by OpenAI and Anthropic models.
This places Microsoft in a sensitive but highly profitable position. The company is selling advanced American AI models to Chinese businesses while also offering Chinese models to Western companies, capturing value from both sides of the global AI market.
Can Microsoft Maintain This Balancing Act?
Despite the commercial opportunity, Microsoft’s role could face growing political and regulatory pressure. In Washington, lawmakers increasingly view China’s AI ambitions as a threat to American industry and technological leadership.
OpenAI’s private concerns could also become more visible if fears grow that GPT outputs are being used to train rival Chinese models.
For now, Microsoft holds a position no other American AI company has: it is the main commercial gateway for OpenAI models in China, while also bringing Chinese AI models to Western enterprise customers. That gives Microsoft a powerful role in the global AI race, but it also places the company at the center of one of the most sensitive technology tensions between the United States and China.
Read the article in
