Nvidia used its AI Day Singapore event on September 22–23, 2026 to line up a roster of Southeast Asian governments, telcos and internet giants standardizing on its open Nemotron models, with Sea Limited becoming the first ASEAN enterprise to adopt the new Vera Rubin platform. The announcements, staged at the Raffles City Convention Centre, span at least six countries and cover public safety, legal AI, smart cities, drug discovery and e-commerce. The through-line is that regional operators are moving off pilots and putting Nvidia-based systems into production against user bases measured in the tens of thousands to hundreds of millions.
Sea, which runs Garena, Monee and Shopee across the region, said the Vera Rubin buildout will let it train and deploy larger models and agents for a user base already numbered in the hundreds of millions. On Shopee, Nvidia-accelerated models generate seller listings and rank product discovery. Monee, Sea's fintech arm, is using AI for fraud detection and credit-risk scoring on customers underserved by incumbent banks. Garena is applying the same infrastructure to matchmaking and player-safety systems inside its games.
Singapore's Home Team Science and Technology Agency, HTX, is running research on Nemotron 3 Super for complex reasoning and agentic workflows and Nemotron 3 Nano Omni for combined vision, audio and language tasks. The agency's stated goal is locally controlled, multimodal AI grounded in Home Team operational data — a public-safety analog to what banks now expect from private model deployments. Fellow Singapore integrator NCS is pairing Nemotron with Nvidia's video search and summarization Blueprint and pushing into humanoid-robotics work, while ST Engineering is building its AI Studio platform on NeMo and cuOpt, starting with a Marine MRO deployment.
Key facts
- 01Sea Limited becomes the first ASEAN enterprise to adopt Nvidia's Vera Rubin platform, scaling AI across Garena, Monee and Shopee for hundreds of millions of users.
- 02Thailand's iApp Technology fine-tuned Nemotron 3 Nano into OpenThai 2.0 Legal, powering the Thanoy legal chatbot that serves roughly 43,000 users.
- 03Singapore's HTX is piloting Nemotron 3 Super and Nemotron 3 Nano Omni for public-safety agentic and multimodal workloads.
- 04Nvidia AI Day Singapore runs September 22–23, 2026 at Raffles City Convention Centre, ahead of GTC Berlin on October 20–22.
The regional adoption pattern is deeper than a single flagship customer. In Malaysia, YTL AI Labs is fine-tuning Nemotron for enterprise and citizen services. In Vietnam, Viettel AI has fine-tuned Nemotron 3 Super for Vietnamese-language and agentic use, and says the resulting model tops both the VMLU benchmark and its internal product benchmark; it is now being wired into a Legal AI agent for internal staff and external customers. FPT Smart Cloud, also in Vietnam, co-developed Nemotron-Personas-Vietnam, a synthetic dataset grounded in Vietnamese demographic data that local developers can use to post-train models.
Thailand's contribution is arguably the sharpest production case. The Big Data Institute and iApp Technology, working through the ThaiLLM Collaboration, adapted Nemotron 3 Nano into OpenThai 2.0 Legal using the NeMo framework and Thai-language legal data. That model has been open-sourced for the Thai developer community and is the engine behind Thanoy, a legal-assistant chatbot that iApp says already serves approximately 43,000 users. In Brunei, Antrique built a Nemotron-based AI platform aimed at the country's food sector.
Smart-city work is anchored on Nvidia's Cosmos world models paired with the VSS Blueprint. Malaysia's ITMAX runs the combination against city traffic operations, and Thailand's AS-TECH uses the same stack for airport passenger flow. Bitdeer AI co-hosted an Open Models AI Codefest to give regional developers GPU capacity to fine-tune Nemotron for local applications, including healthcare workloads. AI Singapore said it is folding Nemotron models and NeMo tooling into its SEA-LION family, an open model line built for Southeast Asian languages.
The strategy is unusually explicit. Nvidia's stated framing is that AI is becoming a matter of national strategy, and the company is positioning Nemotron as the substrate governments use to build sovereign models without training from scratch. The offer to the region: open weights, persona datasets tuned to local demographics, and NeMo tooling to fine-tune on domestic data — with Nvidia silicon underneath. It is the same sovereign-AI pitch Nvidia has taken to Europe and the Middle East, now localized for ASEAN's mix of languages and regulatory regimes.
The commercial case competes with two alternatives. Chinese open-weights models from DeepSeek and Alibaba's Qwen line are already popular with Southeast Asian developers on cost grounds, and Meta's Llama family retains mindshare among startups that want a familiar license. Nemotron's edge in this rollout is the bundled path — model, dataset, framework and reference blueprints wired to Nvidia hardware — which suits government buyers and large enterprises who want an integrator-friendly stack more than they want the absolute cheapest inference. Whether that bundle holds against price pressure from Chinese open models over the next 12 months is the open question.
For Nvidia, the AI Day Singapore lineup is less about any single deal than about locking in the region's institutional AI stack ahead of the Vera Rubin ramp. Sea's early Vera Rubin adoption gives Nvidia a reference customer at consumer scale in ASEAN, and the Nemotron deployments across HTX, Viettel, YTL, iApp and AI Singapore create switching costs — fine-tuned checkpoints, open datasets and integrator playbooks that only pay back if the underlying hardware stays Nvidia. Registration for GTC Berlin on October 20–22 is now open, and the same sovereign-AI script is likely to run there next.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
