Sea Limited Becomes the First ASEAN Enterprise to Adopt Nvidia's Vera Rubin Platform as AI Day Singapore Showcases Regional Builds

Nvidia used its AI Day Singapore event to argue that Southeast Asia's AI build-out has moved from experimentation to production — and to name the region's first flagship customer for its newest hardware. Sea Limited, the Singapore-founded company behind Shopee, Garena and Monee, has become the first enterprise in ASEAN to adopt the Vera Rubin platform, according to Nvidia. Sea plans to use the compute to develop and deploy AI models and intelligent agents at greater scale across its businesses.

AI already runs through Sea's core products: Shopee uses it to assist sellers with product listings, discovery and customer engagement, while Monee applies AI to fraud detection and credit risk assessment and Garena uses it in gaming. Moving to Vera Rubin gives the company additional capacity as those workloads expand — and a symbolic first for a region that has mostly been an adopter of hyperscaler infrastructure rather than an early customer for brand-new Nvidia platforms.

The rest of the announcement read like a map of regional government and enterprise deployments, all built on Nvidia's Nemotron open models. In Singapore, the Home Team Science and Technology Agency (HTX) is researching Nemotron 3 Super and Nemotron 3 Nano Omni for public-safety applications requiring complex reasoning, agentic workflows and multimodal understanding of operational data. Systems integrator NCS is pairing Nemotron models with Nvidia's video search and summarization technology for agentic AI across enterprise and public-sector use, and ST Engineering is building its AI Studio platform on NeMo tools and cuOpt routing software.

Elsewhere in the region, Malaysia's YTL AI Labs is fine-tuning Nemotron models for enterprise and citizen services, Viettel AI is adapting them for Vietnamese-language applications, and AI Singapore incorporates Nemotron into its SEA-LION model family for Southeast Asian languages and cultures. Vietnam's FPT Smart Cloud has released Nemotron-Personas-Vietnam, an open dataset built on Vietnamese demographic and cultural data.

Thailand provided the most concrete end-user example: the Big Data Institute and iApp Technology used the NeMo framework to fine-tune Nemotron 3 Nano with Thai-language legal data, producing OpenThai 2.0 Legal — the engine behind Thanoy, a legal-assistant chatbot that Nvidia said has about 43,000 users. Nvidia also highlighted Cosmos open world models in smart-city projects, from traffic operations in Malaysia to passenger-flow management at Thai airports.

The numbers are modest by hyperscaler standards — a legal chatbot with tens of thousands of users is not a frontier deployment — but the direction is the point. Nvidia is seeding its open model stack across ten ASEAN markets' public sectors, creating a developer base that defaults to Nemotron and, by extension, to Nvidia hardware underneath.

For Southeast Asia's governments, the pitch is sovereignty: open-weight models that can be fine-tuned on local languages and deployed locally, rather than depending on US or Chinese frontier APIs. For Nvidia, every HTX pilot and OpenThai model is another wedge into a region where GPU demand is growing faster than local supply of AI talent — and where the next Sea-scale customer is presumably being cultivated.