Global South warned AI gap could deepen 30-year slowdown

Developing nations are being warned that economic stagnation risks could intensify if governments do not bring artificial intelligence into public sector work. Current projections described in institutional guidance say these economies are on track for their weakest growth performance in thirty years, heightening pressure to find new ways to lift productivity.

The advice focuses on practical deployment rather than prestige infrastructure. It says developing countries should concentrate on adapting low-cost AI tools that are localized for their needs, instead of trying to build large data centers and the advanced computing capacity associated with AI models developed in major global powers.

Guidance prioritises low-cost, localised AI over data centres The central recommendation is to bypass the heavy capital spending and major infrastructure build-out required to support cutting-edge computing. The guidance argues that attempting to replicate the large-scale AI stacks built in wealthier countries can impose barriers that are difficult to clear for governments facing tight budgets and competing social priorities.

Instead, it points to a strategy built around scalable, resource-efficient applications that can be adjusted to local languages, administrative systems, and service delivery realities. The stated is to use AI where it can strengthen public administration and service outcomes without requiring the same level of computing investment as frontier models.

Target sectors include health, education, justice and agriculture The guidance identifies several public-interest areas where AI adoption is intended to improve results: health, education, justice, and agriculture. The thrust of the approach is implementation inside government operations, aiming for measurable gains in how services are delivered rather than a broad technology build-out for its own sake.

Framed as a productivity strategy, the proposed use of localized AI is presented as a way to help governments make more efficient use of scarce resources. The guidance links this to a broader development challenge: preventing developing economies from falling further behind in global economic progress if adoption lags.

Energy and water constraints shape the recommended path

A key constraint highlighted is the resource footprint of large-scale AI infrastructure. The guidance notes that building and running major data centers can carry high energy and water demands, which can be difficult to accommodate in settings where infrastructure is limited or where governments must balance competing needs.

By encouraging AI that is designed to be scalable and resource-efficient, the guidance sets out an alternative that seeks to reduce exposure to these infrastructure pressures. It frames the choice not only as a financial consideration, but also as an operational one tied to energy and water availability.

Uncertainty centres on execution and integration

While the guidance sets out a clear direction, it also implies an open question: whether governments can implement these tools effectively across public systems at the speed and scale needed to counter the current growth outlook. The risk it highlights is that without integration into public sector operations, productivity gains may remain limited as growth hits a thirty-year low point.

The message to policymakers is to prioritise workable, locally adapted AI deployments that can be scaled across core services, while avoiding the cost and resource intensity associated with large-scale infrastructure models built elsewhere.