Uganda and Tanzania Are Building AI Where the Internet Is Still Scarce

Uganda and Tanzania illustrate a version of artificial intelligence adoption rarely discussed: what happens in economies where connectivity is expensive, electricity is unreliable, most people work in agriculture, and capital is scarce.

Shared conditions:

Both countries have young, fast-growing populations, agriculture-dominated employment and rapidly rising mobile phone penetration alongside limited smartphone and internet adoption.

Mobile money is widespread in both, following the East African pattern, and provides the same foundation for financial services that it does in Kenya: transaction histories supporting credit assessment for people without bank accounts.

Both face high costs for internet access relative to income, though submarine cable investment and terrestrial fibre expansion have improved the situation. Landlocked Uganda depends on transit through neighbours for connectivity, which adds cost.

Uganda:

Kampala hosts a modest but active technology scene, with activity in financial technology, agricultural technology and health.

Agricultural applications are the most economically relevant. Coffee is a major export, and applications include disease detection, yield forecasting and market price information delivered by SMS. Extension advisory services delivered digitally reach farmers who would otherwise see an agricultural officer rarely.

Health applications address severe workforce shortages, including diagnostic support and supply chain management for medicines. Uganda has experience with disease outbreak response, and data analysis for surveillance has practical use.

Uganda has invested in national fibre infrastructure and a national data centre, aiming to host government systems domestically.

Constraints include electricity access, which remains limited outside urban areas despite hydropower generation, periodic internet restrictions during politically sensitive periods, and limited research capacity.

Tanzania:

Tanzania has a larger population and economy, with Dar es Salaam as its commercial centre.

Mobile money penetration is high, and the country was early in implementing interoperability between providers, improving system utility.

Agriculture employs most Tanzanians, and applications mirror regional patterns: crop monitoring, weather services, market information and index insurance. Livestock is economically significant, and monitoring applications apply.

Wildlife conservation is a distinctive area. Tanzania's national parks are economically important for tourism, and camera trap image analysis, acoustic monitoring and anti-poaching patrol optimisation are active applications, often supported by international conservation organisations.

Mining, including gold and increasingly critical minerals, uses geological modelling and operational optimisation.

Tanzania has invested in national ICT infrastructure including fibre backbone and data centre capacity, and connectivity through the port of Dar es Salaam serves landlocked neighbours.

Language:

Swahili is widely spoken in both countries and across East Africa, with tens of millions of speakers. It is among the better-resourced African languages for AI, with genuine research effort behind Swahili datasets and models, though still far behind major world languages.

Local languages beyond Swahili, of which both countries have many, remain largely unserved.

The common constraints:

Compute capacity is minimal in both, with workloads running abroad at foreign currency cost.

Advanced technical education produces small numbers of specialists, and emigration is significant.

Capital markets are shallow, and much technology funding comes from development finance and international donors, which shapes what gets built toward development priorities rather than commercial ones.

The outlook:

For both countries, AI adoption is likely to arrive through mobile-delivered services in agriculture, health and finance rather than through a technology sector in the conventional sense.

That is not a lesser outcome. Applications that raise smallholder yields or extend credit to informal traders affect more lives than most commercial AI products, and they are being built under constraints that would defeat most companies.