Who really wins: The risk of AI wealth concentration in emerging economies

By Paulette Watson MBE

Ghana’s National AI Strategy is one of the most ambitious policy documents to emerge from sub-Saharan Africa in a generation. The targets are bold, the sectors are well chosen, and the political will appears genuine.

But ambition does not determine who captures value. That is decided later, in procurement rooms, data licensing agreements, and talent pipelines.

This is the question business leaders and policymakers must now confront: who is this strategy actually designed to make wealthy?

The concentration risk is structural, not accidental

Artificial intelligence creates value at scale. That is its defining economic property. But in emerging markets, scale tends to concentrate around those who already control infrastructure, telcos, large financial institutions, and multinational technology firms with existing compute power and distribution networks. This is not malice. It is gravity.

Ghana’s strategy rightly identifies fintech, agriculture, and creative industries as priority sectors. But consider what AI-enabled fintech at scale requires: credit data, transaction histories, identity infrastructure. These assets are not held by smallholder farmers or informal traders. They are controlled by incumbents. Without deliberate structural intervention, the AI layer built on top of this data will consolidate existing advantages rather than disrupt them.

The same logic applies to agriculture. Predictive analytics for cocoa yield gains is positioned as transformative. But who owns the agricultural data platforms? Who holds contracts with commodity buyers? If optimisation flows through existing intermediaries, smallholders may see marginal productivity gains, while the bulk of value capture remains upstream.

This is not unique to Ghana. It is the defining governance challenge of AI deployment across emerging economies. Most strategies do not resolve it, because doing so requires constraining the very actors whose capital and infrastructure make implementation possible.

Data ownership is the fault line

Ghana’s commitment to data sovereignty is important. But sovereignty at the national level is not the same as ownership at the individual or community level.

The risk is a model where Ghana retains nominal control over data flows, while the economic value derived from that data accrues to platform operators, often foreign, who process, model, and monetise it at a layer removed from the original data subjects. This is not hypothetical. It reflects the global trajectory of digital platform economics.

The proposed sovereign data infrastructure, cloud hubs, localised models, and Twi-language tools provide a necessary foundation. But foundations alone are not enough. The real governance questions are more difficult: Who licenses access to national datasets? On what terms? And how is value shared with the communities whose data underpins these systems?

Without clear answers, infrastructure risks enable extraction rather than ownership.

The inclusion calculus

The antidote to AI wealth concentration is not redistribution after the fact. It is designing participation into the system from the outset, across talent, data, and access to opportunity.

Women represent less than 25% of Ghana’s STEM workforce. In AI, the gap is even wider. This is not simply a diversity issue. It is a structural signal that the AI economy, as currently configured, will be built by and optimised for a narrow demographic, with predictable consequences.

Initiatives building large-scale talent pipelines for women and girls across Africa, including efforts to train one million women in AI, are therefore not peripheral. They are central to national competitiveness. If the talent base shaping AI systems is narrow, the systems themselves will be narrow. And narrow systems in diverse economies fail, both commercially and socially.

What business leaders should do now

For business leaders, the question of value concentration is not theoretical. It is central to long-term risk and opportunity.

AI systems built on biased or incomplete data fail in credit scoring, health diagnostics, and agricultural advisory. The cost of failure is tangible: regulatory exposure, reputational damage, and lost market trust. Inclusion, therefore, is not a parallel ethical consideration. It is a core business strategy.

Practically, this requires a shift in approach. Data sourcing must be tested for representativeness before scaling. Engagement with public procurement pipelines should prioritise those embedding inclusion from the outset. Talent strategies must extend beyond the traditional STEM graduate pool, which is both too small and too homogeneous to meet the demands this strategy is creating.

The real measure of success

Ghana’s AI Strategy will not ultimately be judged by its targets, but by its distribution.

A country that achieves AI-driven GDP growth while concentrating that growth among incumbents and multinationals has not built an AI-powered society; it has built an AI-powered elite.

The language of inclusion is already present in the strategy. The task now is to translate that language into structure: data ownership frameworks, procurement conditions, and talent investments that make inclusion systemic rather than aspirational.

The ambition is clear. The architecture is emerging.

The question now is not whether Ghana can build an AI economy, but who that economy is ultimately designed to serve.

Paulette Watson MBE is the author of She Disrupts: A Black Woman’s Journey in STEM and AI Industries and leads Global BeMeDigitalInclusion, an initiative focused on advancing one million African women and girls into AI careers. She collaborates with governments, institutions, and industry on AI, digital transformation, and inclusive talent development across Africa.

Post Views: 67