Oracle and Chinese tech giant Tencent have agreed to a five-year lease worth $7 billion across the former’s data centres in south-east Asia, the Financial Times reported.
This marks Tencent’s largest overseas lease deal, and Oracle is providing access to 1,00,000 ‘advanced AI chips’ that are otherwise not available in China.
This development comes as the US tightens restrictions on China’s access to advanced AI chips, particularly NVIDIA’s high-end accelerators.
The country has restricted the export, reexport and in-country transfer of advanced computing chips and systems to China and certain Chinese-headquartered entities.
Yet, the Tencent deal illustrates a significant complication in those controls: the US can restrict where an NVIDIA chip is shipped without necessarily preventing a Chinese company from remotely accessing that chip’s computing power once it is installed in another country.
Under the current export-control framework, a Chinese company cannot simply have a restricted NVIDIA accelerator shipped to a data centre in China.
But if the accelerator is legitimately installed in a data centre in a country where its export is permitted, a company can potentially rent that data centre's computing capacity and send workloads to the GPU remotely.
A recent Reuters legal analysis describes almost exactly this scenario: a Chinese company could potentially become an infrastructure-as-a-service customer at a Malaysian data centre and remotely use advanced computing hardware it could not receive in China.
The report notes that US rules currently do not comprehensively prohibit this kind of ordinary remote access.
Tencent is not alone. ByteDance and Alibaba are already among the largest Chinese users of south-east Asian data-centre capacity.
Chinese companies have increasingly reserved large amounts of capacity in Malaysia, Singapore, Indonesia, Thailand and other regional markets as domestic access to advanced GPUs has become more constrained.
Tencent develops the Hunyuan family of AI models and has been expanding into agentic AI, with products such as WorkBuddy, CodeBuddy and Yuanbao designed to execute multi-step tasks rather than simply generate responses.
Its latest Hy3 and Hy4 models have been trained and optimised specifically for reasoning, coding, and agentic workloads.
- Coforge Revenue Rises 54% as AI Engineering and Deal Bookings Gain Momentum
- Tech Mahindra Partners With SpaceXSI to Bring AI Into Software Engineering
- Infosys AI Revenue Rises 29% in Q2, FY27 Growth Guidance Cut
- LTM Launches BlueVerse Industrial AI Lab in Chennai
- Chariot Launches Speech Recognition Model for 10 Indian Languages
- Meta Expects Up To $145 Billion In Capex As AI Investments Expand

