Tencent Secured Lease for 100,000 AI Chips From Oracle

The five-year agreement allows Tencent to expand AI compute capacity across Oracle data centers in Asia.

Updated on Oct. 1, 2026 in Data Centers

Isometric editorial illustration of a high-density server blade containing a silicon wafer, representing cloud computing infrastructure.
Tencent has signed a five-year agreement to lease 100,000 AI chips from Oracle, allowing the company to scale its artificial intelligence infrastructure across Asian cloud data centers. AI Illustration. Upload story photo >

Live Poll

Is now a good time for major tech firms to commit heavily to AI hardware leasing?

Tencent has signed a five-year agreement to lease 100,000 AI chips from Oracle to scale its artificial intelligence infrastructure. This arrangement enables Tencent to access large-scale compute resources without the capital expenditure of constructing its own data centers.

Why it matters

By leveraging existing cloud infrastructure, companies like Tencent can rapidly accelerate AI development while shifting capital burdens to providers like Oracle. This model marks a significant shift in how massive AI hubs are sustained and deployed across international markets.

Oracle reported delivering over 300,000 GPUs to cloud customers in Q1 FY27, achieving a 97.9% utilization rate. Older hardware units have been resold or renewed at a 20% premium over previous valuations.

The players

Tencent

A major technology conglomerate focused on cloud services, digital entertainment, and large-scale AI infrastructure.

Oracle

A provider of cloud infrastructure and database software that has increasingly focused on supplying GPU compute power to AI firms.

ByteDance

A technology company operating global content platforms and maintaining AI hubs that rely on outsourced cloud compute.

The details

Tencent will host these chips within Oracle’s established Asia-based data centers, avoiding the capital-intensive process of building private server farms. Oracle manages these high-density environments using long-term data-center leases that typically span 15 to 19 years, while customer agreements generally persist for five to six years. This structure mirrors the strategy used by ByteDance, which already operates AI hubs in Singapore utilizing Oracle’s cloud infrastructure.

Timeline

  1. Q1 FY27: Oracle delivered 300,000 GPUs and hit a 97.9% utilization rate.

  2. October 1, 2026: The deal between Tencent and Oracle was reported.

  3. Fiscal years 2027-2029: Oracle's uncommenced data-center lease commitments are set to begin.

The Tech Race

The expansion follows a broader trend within Oracle's uncommenced lease commitment pipeline, which ranges from $248 billion to $288 billion in future obligations. This deal represents a specific execution within the massive expansion program Oracle manages to scale global cloud capacity.

This agreement provides Tencent immediate access to 100,000 chips, bypassing the multi-year timelines typically required for building new physical server farms. For enterprise users in Asia, this expansion signals increased availability of high-performance AI services powered by Oracle cloud infrastructure.

The takeaway

The rapid deployment of compute capacity via leasing models is becoming the standard for large-scale AI operations. Watch for updates in Oracle's fiscal reports throughout 2027 for confirmation on how these lease commitments impact long-term cloud availability.

What happens next

Oracle's total lease commitments are scheduled to commence implementation throughout fiscal years 2027 through 2029.

Further reading

For more on how infrastructure providers scale global capacity, visit the Data Centers section.

Live Poll

Is now a good time for major tech firms to commit heavily to AI hardware leasing?