Sri Lanka, Nepal and Myanmar: AI at the Margins

Sri Lanka, Nepal and Myanmar illustrate what constrains artificial intelligence adoption at the margins of the global economy, where the binding limits are political and infrastructural rather than technical.

Sri Lanka:

Sri Lanka has relatively strong education outcomes and literacy rates by regional standards, and built a modest information technology services export sector serving international clients, concentrated in Colombo.

The sector produces software engineering, financial services technology and business process work, and several Sri Lankan companies have achieved regional scale.

The country's severe economic crisis, including sovereign default, foreign exchange shortages and extended power cuts, damaged the sector badly. Businesses could not import equipment or pay foreign providers, and emigration of skilled professionals accelerated sharply.

Recovery has been partial. The technology sector's fundamentals, educated workforce, English proficiency, established client relationships, remain, and the constraint is macroeconomic stability rather than capability.

Applications where AI is used include financial services, apparel manufacturing, a major export industry where quality control and demand forecasting apply, and tourism.

Nepal:

Nepal is landlocked, mountainous and among the region's poorer economies, with connectivity and electricity infrastructure shaped by extremely difficult terrain.

The technology sector is small but growing, with Kathmandu hosting outsourcing operations and a startup community. Nepali engineers work extensively for foreign clients remotely, and this has become a meaningful source of foreign exchange.

Remittances from Nepalis working abroad dominate the economy, and digital payment adoption has grown substantially, providing the basis for financial technology applications.

Applications with genuine local value include disaster risk modelling, Nepal is highly earthquake-prone and faces landslide and flood risk, agricultural advisory in difficult terrain, and health services delivery where geography makes access hard.

Electricity has improved substantially with hydropower development, and Nepal has significant untapped hydropower potential, which in principle could support data infrastructure, though transmission and investment constraints are severe.

Myanmar:

Myanmar's situation is defined by political crisis. Following the military takeover and subsequent conflict, the technology sector, which had grown rapidly during the preceding decade of opening, has been severely damaged.

Internet shutdowns have been imposed extensively, both nationally and in specific regions, making any internet-dependent business unviable in affected areas. Surveillance capability has expanded, with technology imported for monitoring purposes, and this is the most consequential AI-related development in the country.

Many technology workers have left, and international companies withdrew.

Applications that continue are limited, and the humanitarian situation dominates.

Common conditions:

All three depend on foreign cloud infrastructure, with foreign exchange constraints making that costly.

All three lose skilled workers to emigration, and all three have limited advanced technical education.

For all three, the practical AI question is not model development but whether basic digital infrastructure, electricity, connectivity, payments, functions reliably enough for anything to be built on it.

The outlook:

Where AI reaches these economies, it will arrive through mobile services, agricultural advisory, disaster preparedness and remote work rather than through domestic technology industries.

Myanmar is a reminder that technology development is conditional on political stability, and that the same tools can be deployed for surveillance as readily as for services. The distinction is made by governance, not by technology.