OpenAI and Google sell AI models to Singapore subsidiaries of blacklisted Chinese firms

Singapore, it turns out, has become the Switzerland of the AI arms race — a neutral ground where American AI labs and Chinese tech conglomerates operate within walking distance of each other.

The Singapore loophole

US export controls on AI target specific entities and specific geographies. Mainland China is restricted. Singapore is not. And all three Chinese tech titans — Alibaba, Baidu, and Tencent — maintain substantial operations in the city-state.

Alibaba Cloud already offers OpenAI-compatible APIs through its Singapore infrastructure. That means developers building on Alibaba’s platform can access models architecturally identical to what OpenAI sells directly, routed through a Southeast Asian intermediary.

The legal distinction matters enormously. A Singapore-incorporated subsidiary of a blacklisted Chinese firm is, on paper, a Singaporean company. It operates under Singaporean law, pays Singaporean taxes, and can enter contracts that its parent company in Shenzhen or Hangzhou cannot.

Why Singapore became the AI neutral zone

OpenAI committed over S$300 million, roughly $234 million, to establish its first applied AI lab outside the United States in Singapore in 2026. Google DeepMind opened a regional research hub there in the same year. Meanwhile, all three major Chinese cloud providers have been expanding their Singapore footprints for years, building data centers and hiring local engineering talent.

The Microsoft precedent

Microsoft has been navigating similar waters for years, maintaining partnerships that allow it to offer OpenAI-powered models inside China itself, despite the restrictions that prevent OpenAI from operating there directly. Microsoft’s Azure cloud platform serves as the delivery mechanism. Because Microsoft holds exclusive commercial licensing rights to OpenAI’s models, it can distribute them through its existing Chinese operations in ways that OpenAI itself cannot.

This creates a peculiar dynamic where US policy simultaneously restricts Chinese access to American AI and enables it, depending on which corporate structure is doing the selling. The policy targets entities, not capabilities. So the capabilities flow through whichever entity isn’t on the list.

What this means for investors

Alibaba Cloud’s OpenAI-compatible API offering suggests that Chinese platforms are building interoperability with American models into their core infrastructure. Microsoft’s ability to distribute OpenAI models in restricted markets gives it a competitive moat that pure-play AI labs lack.

The risk to watch is regulatory. The chip export controls started narrow and got broader. If the US Commerce Department decides that selling models to Singapore subsidiaries of blacklisted firms violates the intent of its Entity List restrictions, the entire arrangement could collapse overnight. Investors positioned in AI-adjacent assets should be watching Commerce Department rulemaking closely.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.