Tencent Secures 100,000 AI Chips Through $7 Billion Oracle Deal to Bypass US Export Controls

4 Sources

Share

Tencent signed a five-year lease agreement with Oracle for approximately 100,000 advanced AI chips across Southeast Asia data centers in a $7 billion deal. The arrangement allows the Chinese tech giant to access cutting-edge processors unavailable in China due to US export controls, marking its largest overseas lease deal as it accelerates AI model development.

Tencent Secures Largest Overseas Lease Deal with Oracle

Tencent has signed a five-year lease agreement with US cloud provider Oracle to access approximately 100,000 AI chips across multiple Southeast Asia data centers

1

2

. The Tencent Oracle deal, valued at approximately $7 billion, represents the Chinese tech giant's largest overseas lease agreement to date

3

. Under the terms, Tencent must pay roughly 30% upfront, which contributed to a negative free cash flow of RMB 13.8 billion ($2 billion) in the company's second-quarter earnings—the first negative quarterly figure in over a decade

1

. Oracle shares climbed 1.86% to $139.85 during pre-market trading following the announcement

3

.

Source: Benzinga

Source: Benzinga

Circumventing US Export Controls Through International Leasing

The arrangement enables Tencent to access advanced processors unavailable in China due to tightening US export controls

1

. While Chinese tech companies are blocked from directly purchasing advanced AI chips, they remain permitted to lease computing capacity internationally under US rules, potentially gaining access to Nvidia's cutting-edge processors

4

. Chinese technology companies have been scrambling to secure AI chip supplies amid escalating restrictions

2

. A shortage of domestic supply has driven Chinese tech groups to seek computing capacity overseas, with demand intensifying as the US tightened export controls to prevent Nvidia's advanced chips from entering China through intermediate locations

1

. This escalation has led to increased prices, longer terms, and higher upfront payments for data center lease agreements across the region.

Accelerating AI Model Development and Infrastructure Investment

Tencent will use the 100,000 AI chips to advance its AI model development and create agentic tools

1

3

. The company's latest Hunyuan models have demonstrated substantial improvements in performance, closing the gap with top-tier domestic models such as Alibaba's Qwen and offerings from DeepSeek

1

4

. AI infrastructure spending has surged dramatically, with capital expenditure in the second quarter jumping 176% year-over-year to RMB 53 billion ($7.9 billion) for AI infrastructure and pre-payments for computing resources

1

3

. Chinese tech giants and frontier AI labs are tapping into these overseas resources especially for training their models, as domestic chips cannot yet perform such computationally intensive tasks

1

.

Integration Across WeChat Ecosystem and Enterprise Applications

Armed with stronger models, Tencent is implementing AI agents throughout its ecosystem, particularly within WeChat, its superapp that serves more than 1.4 billion users

1

3

. The company is rolling out Xiaowei, an embedded agent in WeChat designed to execute tasks across the platform's millions of mini-programmes, such as ordering food or booking services, based on natural language commands

1

. Beyond consumer applications, Tencent has built a matrix of AI agents for productivity and enterprise use, led by WorkBuddy, its flagship AI-native office agent, which has experienced rapid user growth and leads the PC-based office agent market in China

1

. The company has also hired top-tier talent and restructured its infrastructure and data departments to support its AI push

1

.

Competitive Landscape in Southeast Asia Data Centers

While Tencent is catching up with multibillion-dollar leasing deals, ByteDance and Alibaba remain the biggest clients of Southeast Asia data centers

1

. In December 2025, Tencent reportedly gained access to Nvidia's advanced Blackwell AI chips by renting computing capacity at a Datasection-operated data center near Osaka, Japan, allowing the company to use Nvidia B200 processors through a third-party cloud structure despite US restrictions on chip sales to China

3

. Recent reports indicate China has considered allowing companies such as ByteDance and Alibaba to purchase Nvidia's high-end RTX PRO 5500 chips

3

. Beijing has also been promoting domestically developed alternatives to US technology providers

2

. Watch for how sustained heavy spending on AI plans affects Tencent's margins and profitability in coming quarters

4

, and whether other Chinese tech firms follow with similar international leasing arrangements to maintain competitiveness in the escalating AI race.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved