As large language models (LLMs) become more accessible, developing countries need to build sovereign AI. That was the conclusion reached by participants at CNBC’s East Tech West conference in Thailand.
Sovereign AI refers to a state’s ability to control its own technologies, data and infrastructure, ensuring autonomy and meeting unique security priorities and needs.
Such capability remains scarce, said Kasima Tarnpipitchai, head of AI strategy at SCB 10X.
He noted that many of the world’s best-known large LLMs are built around English.
“The way you think, how you interact with the world and who you become when you speak another language can be completely different,” said Tarnpipitchai.
Hence countries should assume responsibility for their own AI systems, developing technologies for specific languages, cultures and nations rather than simply translating English-based models, the expert said.
Panellists agreed that the ASEAN region, with a population of about 700 million, is especially well-suited to building sovereign AI. About 61% of the population is under 35, and roughly 125,000 people come online each day.
“I think this is really important, and we are focused on how we can democratize access to cloud technologies and AI,” said Jeff Johnson, Amazon’s managing director for the ASEAN region.
Open-source AI
Participants stressed that using open-source models should be a key means of creating a sovereign AI environment.
“In Southeast Asia, especially in Thailand, there is a great deal of outstanding talent. And if this potential is not realized in a way that makes it accessible to society or helps the ecosystem develop, that would be a real missed opportunity,” said Tarnpipitchai.
The expert added that working with open source is a way to create “collective energy” that will help Thailand compete better in AI and advance sovereignty to the benefit of the whole country.
Open source refers to software whose code is publicly available. This allows anyone to review and adapt it.
Among the most prominent players in open AI are the American Meta and China’s DeepSeek.
Cecily Ng, vice president and CEO for the ASEAN and Greater China regions at Databricks, stressed that the emergence of more open-source AI models gives companies and governments a wider range of options than relying on a few closed models.
Localising compute
Beyond language localisation, using local infrastructure and compute to run neural networks is no less important, said Prem Pavan, vice president and general manager for Southeast Asia and Korea at Red Hat.
Panellists noted that in developing countries such as Thailand, cloud providers with local operations can address this need. That includes AWS, Microsoft Azure, Tencent Cloud, AIS Cloud and True IDC.
“We are here in Thailand and across the Southeast Asian region to support all industries, all businesses of every shape and size — from the smallest startups to the largest enterprises,” said Johnson of AWS.
According to him, the company’s cloud-services economic model lets you “pay for what you use” — lowering barriers to entry and simplifying the creation of models and applications.
AI in Southeast Asia
Across Southeast Asia, countries have varied strategies and initiatives to develop AI:
- Thailand. The national AI strategy aims to turn the country into a regional AI hub by 2027. The government seeks to raise awareness of the technology among 600,000 citizens, build a unified digital identification platform for public services, and encourage foreign investment in data centres and cloud services. Collaboration with IT companies has led to the creation of an Advanced AI Center and a Huawei Cloud initiative with the Thai government;
- Singapore. In 2019 the national AI plan was announced and implemented through projects in education, healthcare, security and ecosystem development. In December 2023 an updated strategy, NAIS 2.0, was launched, targeting “growth points” and expanding trusted use of the technology.
- Vietnam. The Ministry of Science and Technology is preparing a national AI strategy to 2030 (Decision 127/QD-TTg) with ambitious goals to enter ASEAN’s top four and the global top 50 in large language model research.
- Indonesia. Since 2020 a project has been in place, developed with business involvement. It provides for talent and infrastructure development. Laws have already been adopted on data protection, electronic transactions and AI ethical principles. Among the challenges: the country faces a shortage of IT specialists.
- Malaysia. In August 2024 the government approved the creation of a National AI Office under the digital ministry to coordinate policy and develop AI ethics and regulation. In 2024–2025 the state attracted significant investment from tech giants, including $2bn from Google for a data centre and a cloud region. A total of $16bn was invested in the sector in 2024.
- Philippines. In July 2024 the trade ministry launched the National AI Strategy Roadmap 2.0 and a Centre for AI Research. The document complements the first 2021 roadmap and reflects current technologies (generative AI, ethics). The project aims to make the Philippines a regional AI R&D hub, developing solutions for agriculture, urban planning and sustainability.
AI initiatives are also under way in Cambodia, Laos, Myanmar and Brunei.
Most ASEAN countries already have personal data protection laws, including Singapore, Malaysia, the Philippines, Thailand and Indonesia. Vietnam is preparing its own privacy law.
Public and private investment is being mobilised. Singapore is directing substantial funds to “deep tech”, launching a $245m fund to support startups focused on AI. In Malaysia and Vietnam the state co-finances joint projects with corporations and training centres.
Google, Amazon, Microsoft, Nvidia, Tencent and Alibaba continue to invest in ASEAN.
Across the region, interest in generative AI is rising, with joint initiatives proliferating alongside foreign partners. A movement toward “responsible AI” is taking shape: the focus is on algorithmic transparency, ethics and data protection.
In June, it emerged that Thailand’s authorities sought to approve an “Artificial Intelligence Act” regulating the use of AI across the kingdom.