By Joanna Tan · AI in Action · September 25, 2026

Most conversations about sovereign AI in Southeast Asia focus on one half of the equation: who builds the model, and where the data centre sits. Thailand has spent 2026 quietly running both halves of the play at once — building a homegrown large language model, and then handing AI access to as many of its citizens as possible. Taken together, the two efforts make for one of the region’s more complete case studies in what “AI in action” actually looks like at national scale.

A Thai-language model with real customers

The infrastructure layer is ThaiLLM, launched in April by the National Science and Technology Development Agency (NSTDA) alongside the Ministry of Higher Education, Science, Research and Innovation, NECTEC, and the Digital Economy and Society Development Fund. It comes in two sizes — an 8-billion-parameter model built for speed and a 30-billion-parameter version for heavier workloads — trained on more than 100 billion tokens of Thai government documents, educational material, and legal texts. Rather than shipping that data overseas for training and inference, NSTDA runs ThaiLLM on the domestic ThaiSC supercomputer, which is the crux of the sovereignty argument: government, business, and personal data can be processed without leaving the country.

What makes ThaiLLM more than a research exercise is who has already built on top of it. KBTG, Kasikorn Bank’s technology arm, and SCB 10X, Siam Commercial Bank’s innovation unit, are both running production fine-tunes of the 8B model for financial-services use cases, alongside NECTEC’s own Pathumma variant and the open-source OpenThaiGPT project. That’s sovereign AI infrastructure being tested against a genuinely demanding customer — two of the country’s largest banks — rather than sitting in a lab.

Then: give it to five million people

The adoption layer arrived a few months later. On 31 August, Thailand’s Ministry of Digital Economy and Society, through the Office of the National Digital Economy and Society Commission, launched TH-AI Passport — a platform called AiPASS that bundles more than 30 AI models, including tools built on GPT, Claude, and Gemini, with structured, gamified learning under a “Learn, Earn, Plern” (play) model. Citizens aged 15 and up can register, work through courses, and unlock deeper access to premium AI tools as they build skill, with backing from Microsoft, Google, and NVIDIA’s training programmes.

The uptake surprised even the ministry: more than 1.3 million people pre-registered within two weeks of sign-ups opening, against a stated target of reaching five million users. “AI will have value only when it is in the hands of people who know how to use it,” Digital Economy and Society Minister Chaichanok Chidchob said at launch, framing literacy, not just access, as the goal. The programme is explicitly pitched as closing a measured gap: Thailand’s working-age generative AI usage sits around 12.4%, versus a global average closer to 17.8%, under the country’s 2022-2027 national AI action plan.

Why the pairing matters

Plenty of APAC governments have announced sovereign compute ambitions this year; fewer have paired that with an equally deliberate push to get ordinary citizens and mainstream institutions actually using the resulting tools. ThaiLLM answers the “can we build and control our own AI” question; TH-AI Passport answers the “will anyone actually use it” question that often gets less attention in sovereignty debates. A sovereign model that only serves a handful of ministries is a smaller win than one embedded in daily banking apps and reinforced by millions of citizens building baseline AI fluency in parallel.

For a region where Singapore, Malaysia, Indonesia, and Vietnam are all sketching their own sovereign AI roadmaps, Thailand’s two-track approach — sovereign infrastructure on one side, mass literacy and access on the other — is worth watching as a template, not just a headline.

Source: “ThaiLLM launched to strengthen Thailand’s national AI capacity,” The Nation Thailand, 20 April 2026; “TH-AI Passport draws 1.3 million registrations ahead of AiPASS launch,” The Nation Thailand, 29 August 2026.

Joanna Tan covers AI investment, policy, and the business of AI for AIAPAC.ai Insights.