The World Bank's AI plan starts with the cost of getting connected

World Bank, Digital Progress and Trends Report 2025, Figure 1.5: the 4Cs of AI foundations. CC BY 3.0 IGO. Cropped and rasterised from the original PDF; illustration unchanged.

A useful AI service needs more than a model: the person using it must be able to connect, afford the device and data, and do something worthwhile. That is the organising idea of the World Bank Group's digital strategy launched in Seoul on 20 October. Its significance for African readers is the connection between those layers, not a promise that a new AI service has already arrived locally.

A 21 October TechReviewAfrica report prompted this explainer. The primary release sets out four priorities: connectivity, affordable devices and data, digital services and skills, and AI readiness. Privacy, cybersecurity and fraud protection run across them. These are the institution's stated plans; PING has not measured their delivery.

Think of access as a chain

A network reaching a neighbourhood and a resident being able to use it are different achievements. Consider a hypothetical shopkeeper who can get a signal but cannot comfortably pay for a suitable phone and recurring data. A new business application does not resolve those costs by itself. Equally, a cheaper handset is not sufficient if the service is unreliable or the task remains difficult to complete. The practical unit is the usable service, not one impressive component.

The Bank's approach spans international links, national networks and connections to homes, businesses, schools and clinics. It also identifies entry-level smartphones, data costs, digital identification, payments and skills. That makes the announcement broader than a data-centre plan. For readers assessing a future local project, the useful question is which part of that chain it changes, and what still has to work alongside it.

Shared computing is a separate announced initiative

Alongside the strategy, the Bank announced an AI Infrastructure and Readiness Initiative for emerging-market governments. It describes shared, pay-as-you-go GPU capacity as an alternative to standalone national data centres, with an initial focus on Africa and Latin America. GPUs provide specialised computing for developing and running AI applications. Cassava Technologies and NVIDIA are named as knowledge partners, not as proof of completed facilities or signed supply contracts.

The release gives no country allocation, operating site, capacity, customer price, application route or operating timetable for that initiative. A developer in Lagos or Nairobi therefore cannot use this announcement as an access entitlement or a price quotation. Nor does the regional focus establish a particular university's, business's or public agency's eligibility. Those questions need an actual programme or service document when one is available.

The task matters more than the label

The strategy emphasises locally adapted, task-specific AI, with examples including crop-disease identification, support for health workers and inventory management. These illustrate intended uses, not validated products or demonstrated results. For a hypothetical stock-management tool, a meaningful assessment would examine whether it handles the shop's real products, gives understandable corrections and fits the way staff already work. Calling a system small does not answer those questions.

Look for the bridge from strategy to service

For African institutions and UK-based partners following the announcement, the next useful evidence is specific: who is served, what they can access, how it is paid for, and how performance and safeguards will be assessed. The answers may differ between projects. A regional ambition should not substitute for local terms, and an attractive demonstration should not substitute for evidence from the intended users.

The strongest reading of this announcement is a joined-up direction: connect people, make access affordable and make services useful, while building AI capability. Whether that direction becomes a better everyday experience depends on the projects that follow. PING will distinguish those concrete commitments and outcomes from the strategy itself. More explainers are on PING's Networks desk.