NVIDIA used its AI Day Singapore event on 22–23 September to set out how governments and companies across South-East Asia are building on its open Nemotron models, adapting them for local languages and public services rather than training systems from scratch. NVIDIA’s own summary names deployments in four countries.
In Singapore, the Home Team Science and Technology Agency is researching Nemotron 3 Super and Nemotron 3 Nano Omni for public safety work, and AI Singapore is extending SEA-LION, its open model family for South-East Asian languages, to include Nemotron models and NVIDIA’s NeMo tools. Sea Limited, founded in Singapore, is described as the first enterprise in the ASEAN region to adopt NVIDIA’s Vera Rubin platform.
Beyond Singapore, Malaysia’s YTL AI Labs is fine-tuning Nemotron for enterprise and citizen services, and Vietnam’s Viettel AI for Vietnamese-language and agentic applications. In Thailand, iApp Technology has fine-tuned OpenThai 2.0 Legal on Thai-language legal data, released it as open source, and runs it behind a legal-assistant chatbot that NVIDIA says already serves approximately 43,000 users.
Why it matters: sovereign AI in the region increasingly means adapting open weights from an American chip company. The language and data work stays local; the model lineage and the hardware do not. Every figure here is NVIDIA’s own, and that user count is the only one it gives.