Microsoft Sells OpenAI Models in China

Microsoft has started selling OpenAI models directly in China, a step that OpenAI and Anthropic have both refused to take. The move was first reported on June 18, 2026, per a recent Grok AI News thread.

The decision reflects Microsoft's separate China operations through its Azure partnership structure, allowing localized model access without direct OpenAI branding.

What It Is / How It Works

Microsoft routes OpenAI models through its China-based cloud infrastructure. Customers access the models via Azure services that comply with local data and regulatory requirements. OpenAI itself maintains no direct presence or sales channel in the country.

Alternatives and Comparisons

OpenAI and Anthropic continue to block direct model access for Chinese users. Microsoft’s approach creates a distinct path through its existing joint venture.

Pros and Cons

• Microsoft gains revenue from a large market while competitors stay out.
• Chinese enterprises receive API access without needing overseas accounts.
• Data residency rules remain tied to Microsoft’s local infrastructure.
• Dependency on one provider increases if OpenAI later changes terms.

Who Should Use This

Chinese enterprises needing GPT-class models for internal tools should evaluate Microsoft Azure China first. International teams with strict data policies or those avoiding China-specific compliance should continue using non-Chinese endpoints. Researchers focused on model behavior under different regulatory regimes may find the new access useful for controlled testing.

Bottom Line / Verdict

Microsoft’s move creates the only sanctioned commercial channel for OpenAI models inside China, shifting competitive dynamics for both providers and customers.

Microsoft’s China strategy shows how infrastructure partnerships can bypass direct model-provider restrictions while competitors maintain their current limits.