Aolani to Deploy 22,000 Nvidia Chips in Southeast Asia

Singapore-based AI cloud provider Aolani has agreed to deploy 22,000 Nvidia Blackwell Ultra graphics processing units across new data centers in Malaysia and the Philippines. The infrastructure is scheduled to come online by early 2027, significantly expanding the region's capacity for high-performance artificial intelligence workloads.

This move will push Aolani's total GPU deployment past 48,000 units and bring its total power capacity above 100 megawatts. According to TradingKey, the partnership relies on a revenue-sharing model designed to lower the financial barrier for building such large-scale computing facilities.

Revenue sharing reduces capital risk

Traditional data center projects often require massive upfront capital, which can be a barrier for mid-sized cloud providers. In this arrangement, Nvidia provides credit support and shares in the recurring revenue generated by the cloud services. This aligns the vendor's income with actual customer usage rather than one-time hardware sales.

However, the financial stability of this model depends entirely on sustained demand. If customers do not utilize the GPUs at high rates, the revenue stream may not cover the significant operating costs. Neither company has disclosed the specific revenue split or the total financing amount, leaving the exact financial risk profile unclear to outsiders.

Southeast Asia attracts major cloud investment

Aolani is part of a broader trend where global tech giants are pouring billions into the region. Major providers have committed over $50 billion to AI infrastructure in Southeast Asia, with Malaysia emerging as a primary hub. Google and Microsoft have each pledged roughly $2 billion to expand their local data center presence.

The Philippines represents a newer, smaller market for these investments. Aolani aims to serve local enterprises and startups in both countries, providing access to advanced AI computing resources that were previously difficult to obtain regionally. This expansion reflects a growing demand for localized data processing to meet latency and regulatory requirements.

Delivery and power constraints remain

Despite the strategic alignment, the project faces significant execution risks. The timeline is tightly linked to the delivery of the Blackwell Ultra chips and the availability of sufficient power supply in the target locations. Data center construction is complex and often subject to regulatory delays.

Success will hinge on the ability to secure financing and maintain high GPU utilization rates. If equipment delivery slips or power capacity falls short, the 2027 deadline may be missed. The lack of public details on contract terms makes it difficult to assess how resilient the project is to market fluctuations.