At the NVIDIA AI Day Singapore event held on September 22-23, at the Raffles City Convention Centre, NVIDIA and regional partners showcased breakthrough artificial intelligence advancements across Southeast Asia. The initiative aims to accelerate the transition of public sector and enterprise AI from experimental pilots to production-scale deployment using localized open models, developer tools, and a broad partner ecosystem.
Moving Public Sector AI from Pilot to Production
Governments across Southeast Asia are increasingly treating artificial intelligence as a core matter of national strategy. The focus has shifted from initial proof-of-concept pilots to production-scale deployments that reflect local languages, cultural priorities, and distinct economic needs. NVIDIA is supplying the necessary technology, infrastructure, ecosystem, and expertise to help regional nations build AI capabilities.
To operationalize this strategy, NVIDIA and its local partners are concentrating on four primary pillars:
- Enhancing core government operations and digital service delivery.
- Developing accessible, AI-powered citizen services while empowering local businesses.
- Strengthening critical national infrastructure and public safety.
- Supporting domestic startups, developers, and researchers to strengthen national AI capabilities and innovation in each country.
Regional Implementations of NVIDIA Nemotron Models
Enterprise and public sector deployments rely heavily on custom adaptations of NVIDIA’s open model families. In Singapore, the Home Team Science and Technology Agency (HTX) is actively researching the application of NVIDIA Nemotron 3 Super and Nemotron 3 Nano Omni models to enhance public safety operations. Nemotron Super facilitates complex reasoning and agentic workflows, while the Omni architecture integrates unified vision, audio, and language capabilities for real-world multimodal data processing.
Concurrently, NCS is deploying agentic AI across enterprise and government verticals utilizing Nemotron architectures alongside the NVIDIA Blueprint for video search and summarization (VSS). ST Engineering is leveraging NVIDIA NeMo tools and cuOpt software to scale its AI Studio platform for industrial applications, including Marine MRO.
Similar developmental pipelines are scaling across the wider ASEAN landscape:
- Malaysia: YTL AI Labs is fine-tuning Nemotron models for enterprise and citizen services, while ITMAX utilizes the NVIDIA Cosmos open world models alongside the VSS Blueprint to optimize urban traffic operations.
- Vietnam: Viettel AI has extensively fine-tuned Nemotron 3 Super for Vietnamese-language applications. The model achieved top rankings on both the VMLU benchmark and internal evaluations, and is slated for integration into a Legal AI agent harness for internal Viettel Group employees and external customers. FPT Smart Cloud has also co-developed Nemotron-Personas-Vietnam, an open dataset grounded in local demographic and cultural data.
- Thailand: The Big Data Institute and iApp Technology—members of the ThaiLLM Collaboration—are utilizing Nemotron foundations. iApp Technology fine-tuned Nemotron 3 Nano with OpenThai 2.0 Legal data using the NVIDIA NeMo framework. The resulting open-source model powers “Thanoy,” a legal-assistant chatbot serving approximately 43,000 users. Additionally, AS-TECH applies the Cosmos and VSS stack to manage passenger flow in airports.
- Brunei: Antrique has constructed an AI innovation platform designed to increase productivity within the nation’s food sector.
Expanding the Open Ecosystem in Singapore and Beyond
Ecosystem collaboration remains a critical driver for regional specialization. AI Singapore is expanding its indigenous SEA-LION open model family—designed specifically for Southeast Asian languages and cultural contexts—to incorporate NVIDIA Nemotron open models and NeMo tooling.
In the biomedical sector, Hummingbird Bioscience and LynxKite are building an explainable Toxicity Knowledge Graph. Powered by the Nemotron 3.5 Lightning model and NeMo Retriever, the system uses in silico simulations to integrate complex public and proprietary datasets, aiming to de-risk and accelerate pharmaceutical drug discovery and development.
Bitdeer AI co-hosted the Open Models AI Codefest alongside NVIDIA, supplying the underlying GPU cloud infrastructure required for developers across the region to train localized applications using Nemotron open models, datasets, and training recipes.
Sea Limited Adopts NVIDIA Vera Rubin in ASEAN First
In a major hardware milestone for the region, Singapore-founded global technology company Sea Limited has become the first enterprise in the ASEAN region to adopt the NVIDIA Vera Rubin platform. Sea will leverage the advanced computing infrastructure to scale its artificial intelligence capabilities across its primary operating divisions.
Sea currently deploys AI across Garena, Monee, and Shopee platforms to service hundreds of millions of users:
- Shopee: AI tools assist merchants with product listing generation, search discovery, and customer engagement, allowing resource-constrained small and medium-sized enterprises to utilize enterprise-grade machine learning systems.
- Monee: Digital financial services apply AI models to automated fraud detection and credit risk assessment, broadening financial inclusion for underserved populations.
- Garena: Digital entertainment and video game operations utilize AI to maintain safe, inclusive, and engaging online player communities.
The integration of the NVIDIA Vera Rubin architecture will provide Sea with the necessary computational throughput to accelerate innovation, scale AI applications more broadly, and expand intelligent agent deployment across its entire regional ecosystem.
