World Bank sees Thailand gaining from AI supply chains

Thailand ranks fifth among the developing economies expected to benefit most from investment flowing into semiconductors, electronics and data-centre equipment as artificial intelligence reshapes global supply chains, according to the World Bank.

China tops the group, followed by Mexico, Malaysia, Vietnam and Thailand. Developing economies already account for more than half of worldwide exports of goods associated with AI systems.

The findings form part of the World Bank’s 2026 work on artificial intelligence, which examines how developing economies can capture the economic gains from the technology without having to match the spending or computing capacity of the world’s largest technology companies.

The Bank argues that AI could give developing countries an unusually rapid route past economic constraints that have persisted for generations. If managed effectively, the technology could lift global economic growth to its strongest rate since the boom years of the 2000s, while spreading internationally faster than previous waves of technology.

The threat to jobs also differs sharply between richer and poorer economies. World Bank findings reported this week put the share of jobs vulnerable to AI-driven automation at about 4.5% in developing economies, compared with 14.2% in high-income countries.

Southeast Asia moves deeper into the AI supply chain

Malaysia offers one example of how Southeast Asian countries are moving into higher-value parts of the semiconductor industry.

After decades focused heavily on chip manufacturing and assembly, the country is beginning to develop domestic design capabilities. Malaysian company SkyeChip in 2025 unveiled the MARS1000, the country’s first locally designed edge-AI processor, built using 7-nanometre process technology.

The shift comes as spending on AI infrastructure accelerates far beyond the scale available to most Southeast Asian economies.

Alphabet, Amazon, Meta, Microsoft and Oracle are expected to spend about US$750 billion on capital expenditure in 2026, largely to support AI infrastructure. The estimate covers five of the biggest US cloud and technology groups and is equivalent to 38% of their combined revenue, according to S&P Global Ratings.

That spending is greater than the annual gross domestic product of several Southeast Asian economies individually, including Singapore, Thailand, Vietnam, the Philippines and Malaysia.

Indermit Gill, Chief Economist of the World Bank Group and Senior Vice President for Development Economics, said developing economies should not allow the scale of investment by global technology companies to deter them.

“AI has thrown developing economies a lifeline, and they should seize it,” Gill said.

Countries do not need to build their own large AI models or hyperscale data centres to benefit, he argued. Smaller and cheaper tools can instead be adapted to local conditions and applied to areas where developing countries have immediate needs.

The World Bank sets out three broad stages for countries with different levels of technological capacity:

• Adopt existing technology by using smaller, affordable and readily available AI tools.
• Adapt the technology to local data and languages, including applications in healthcare diagnosis, education and agriculture.
• Innovate by moving into more advanced domestic AI development once the necessary infrastructure and capabilities are in place.

Reliable electricity and high-speed internet remain basic requirements. The Bank also identifies literacy, numeracy and other foundational workforce skills as necessary before countries can make wider use of the technology.

Thai small firms take up AI chatbots

Thailand is already showing relatively strong use of one of AI’s most accessible applications.

The World Bank’s Enterprise Survey on AI Adoption covered companies in manufacturing, construction and services with at least five employees across seven countries. The Thai sample included 360 businesses.

Among small Thai companies employing five to 19 people, 20% reported using AI chatbots in their operations, compared with 25% in the United States.

The relatively high rate of initial adoption does not mean that AI has transformed how most Thai companies operate.

The World Bank found that businesses are still mainly using the technology for basic support tasks rather than restructuring their core operations around it. Information searches, drafting and editing documents, and translation were among the most common uses.

How companies adapt imported technology to local conditions could determine how useful those tools eventually become.

The Bank cites the experience of a British AI model developed to screen Covid-19 patients and later tested in Vietnam. Its performance fell when applied to Vietnamese patients because of substantial differences in the demographic characteristics of the populations involved.

Southeast Asian developers have responded to the wider problem of local relevance by developing SEA-LION, or Southeast Asian Languages in One Network, a family of large language models designed around the languages and contexts of the region.

The aim is to give AI systems a better understanding of Southeast Asian languages, laws and cultural settings instead of relying entirely on models developed using data from other parts of the world.

Governments test AI while tightening oversight

Governments in Southeast Asia are also experimenting with AI in public services while developing rules intended to control its risks.

Singapore’s tax authority has introduced an AI assistant that allows taxpayers to check outstanding amounts and change payment arrangements. The service saved members of the public almost 12,000 hours in 2024, according to the World Bank material.

ASEAN has meanwhile introduced regional guidance on AI governance and ethics as member states work towards common principles for the technology.

Individual governments have also intervened when they see specific risks. The Philippines temporarily suspended access to the Grok model over concerns about sexually explicit material generated by the system.

For governments, the World Bank identifies public trust as a central requirement for wider AI adoption. Rules need to protect personal information and reduce the risk that automated systems reproduce bias when used in public-sector decisions.

Without adequate safeguards, the Bank warns, AI could deepen existing inequalities rather than narrow them and erode trust in the institutions deploying it.

Its prescription for developing economies nevertheless remains centred on practical adoption rather than attempting to compete immediately at the technological frontier: start with accessible AI, adapt it to local needs and move towards more advanced development as infrastructure and skills improve.