Oracle and Tencent Sign Reported $7 Billion AI Cloud Deal for 100,000 GPUs

Oracle and Tencent have reportedly struck a five-year cloud infrastructure deal worth approximately $7 billion, granting the Chinese tech giant access to around 100,000 advanced AI chips deployed across Oracle data centers in Southeast Asia.First reported by the Financial Times on September 30, the agreement would mark Tencent’s largest overseas cloud commitment with a US provider and another landmark contract for Oracle’s fast-growing AI infrastructure business.

Under the reported terms, Tencent would pay roughly 30% upfront, approximately $2.1 billion, with the remaining $4.9 billion spread over the life of the contract on an undisclosed schedule, implying an average annual value of $1.4 billion. Neither company had confirmed the deal publicly at the time of writing, and details on GPU models, specific locations and deployment timelines remain undisclosed.

For Tencent, the strategic logic is straightforward. The company is pushing aggressively into large language models, generative AI and agentic applications while competing with Alibaba, Baidu and rising specialists such as DeepSeek. US export controls have severely constrained the availability of advanced accelerators inside China, making offshore leasing arrangements an attractive workaround, securing high-performance compute capacity without importing restricted hardware onto Chinese soil.

For Oracle, the deal reinforces its positioning as a tier-one AI infrastructure provider. Coming on the heels of its reported $300 billion commitment with OpenAI, it confirms Oracle’s ability to attract long-duration, large-scale infrastructure contracts. The flip side is the capital intensity these deals demand: massive upfront expenditure creates real pressure on free cash flow and execution.

The transaction also spotlights Southeast Asia’s growing strategic importance as an alternative deployment zone for advanced computing, particularly for organizations navigating semiconductor restrictions. That said, regulatory uncertainty remains a real risk, Washington could move to tighten rules around remote access to restricted AI computing capabilities hosted outside US borders.

Three broader signals emerge from this deal. AI compute capacity has become a strategic resource, commanding multi-billion, multi-year commitments. Cloud infrastructure is displacing direct hardware ownership for organizations requiring GPU scale. And geopolitical constraints are actively reshaping global data center investment patterns and accelerating the redistribution of AI infrastructure across geographies.

The AI infrastructure race is no longer simply about acquiring GPUs, it is about securing capacity, financing its deployment and controlling access to it across a multi-year horizon.