What Nvidia and Its Partners Announced Around AI Day Singapore 2026

Around AI Day Singapore on September 22–23, 2026, Nvidia and its partners described three related efforts: building more regional AI computing capacity, adopting newer computing systems, and adapting AI models for local uses. The most concrete new capacity target came from Aolani, but its two announced factories are scheduled for the future—not reported as already running. 240

Aolani’s planned AI factories: 22,000 GPUs in early 2027

Singapore-founded AI-cloud provider Aolani announced a compute collaboration with Nvidia on September 21. It expects to deploy 22,000 Blackwell Ultra GPUs across two AI factories in Malaysia and the Philippines in early 2027. Aolani says that expansion would bring its total deployment to more than 48,000 GPUs and its total AI-factory capacity to more than 100 megawatts. Those totals describe the company’s broader deployment and capacity targets, not the size or power draw of either new site individually. 40

Aolani intends to provide Nvidia-powered cloud computing services to customers. Reporting also describes a revenue-sharing and credit-support arrangement intended to tie the infrastructure expansion to customer demand. That is a description of the companies’ commercial and financing approach, not a published customer access policy: the supplied sources do not establish whether the new capacity will be sold on demand, by reservation, or through both methods. 335

Sea Limited’s Vera Rubin plans

Sea Limited was reported as the first ASEAN enterprise to adopt Nvidia’s Vera Rubin platform. Its plans span businesses including Shopee, Monee and Garena, with the aim of scaling AI models and intelligent agents. The cited reporting does not give a production start date or a confirmed workload for each business, so adoption should not be read as evidence that all three are already running on Vera Rubin. 1012

Where Nemotron fits

Nvidia’s model announcements focused on adapting Nemotron to regional needs rather than announcing a single region-wide deployment. In Singapore, the Home Team Science and Technology Agency, or HTX, is researching Nemotron 3 Super and Nemotron 3 Nano Omni for public-safety applications. Nvidia describes potential support for complex reasoning and agentic workflows from Super, and vision, audio and language capabilities from Omni. These are described as research and potential uses, not a confirmed production rollout. 5

In Vietnam, Nvidia’s reporting identifies Viettel AI as fine-tuning Nemotron for Vietnamese-language applications. It also identifies YTL AI Labs in Malaysia as fine-tuning the models for enterprise and citizen services. The supplied source excerpts do not establish a specific named Nemotron deployment in Thailand, so attributing one to a Thai organization would go beyond the available evidence. 11

What the announcements do—and do not—quantify

The 22,000-GPU figure and early-2027 target belong to Aolani’s proposed expansion, not to every regional project discussed around AI Day Singapore. An earlier Nvidia account says Firmus Technologies is expanding its AI-factory footprint across Australia and Southeast Asia and has deployed AI infrastructure in Singapore with ST Telemedia Global Data Centres. That account does not provide a comparable September GPU commitment for Firmus. The supplied evidence likewise does not support a specific AWS regional GPU figure or timetable tied to these announcements. 4023

The distinction matters: Aolani disclosed a numerical buildout target, Sea outlined a platform adoption, and the Nemotron examples range from research to model fine-tuning. None of those descriptions, by itself, establishes that the planned capacity or applications were fully operational during AI Day Singapore. 4012511