I grew up in Nairobi listening to the BBC World Service with my grandfather. Every evening, the same ritual: his armchair, the crackle of shortwave, a voice from somewhere else telling us stories about the world. When I was old enough, I got a Walkman. Then cassette tapes I could choose myself. That shift, from receiving stories to selecting them, felt like the most natural kind of freedom.

Decades later, I am building Hey Kahani, a screen-free audio storytelling platform for South Asian diaspora children. The premise sounds simple, but the reality is more specific: the oral traditions of South Asia and sub-Saharan Africa: The Panchatantra, the Jataka tales, the grandmother's stories told across a thousand Kenyan and Ugandan kitchens are almost entirely absent from the training data of every frontier AI model.

The 2026 International AI Safety Report, led by Yoshua Bengio and written by over 100 independent experts, puts a number on this: AI models correctly answered 79 percent of questions about everyday US culture and 12 percent of questions about Ethiopian culture.¹ Twelve percent. A peer-reviewed study published the same year analysed six major LLMs and found that of Africa's 2,000-plus languages, just 42 receive meaningful support. Over 98 percent are entirely unsupported.²

The stories that shaped me, that shape billions across the Global South, do not exist in the AI systems that will increasingly mediate how the next generation learns, works, and understands the world. Hey Kahani is, at one level, a children's platform. At another, it is a structured archive of oral tradition. The long-term asset is stewardship of a dataset the global AI industry does not have and cannot easily replicate. That is not a side thesis. It is why I was invited to moderate at the Uganda DeepTech Summit in Kampala, April 30–May 1, 2026.

The Baseline: Governance vs. Reality 

In Kampala, I sat in rooms where some of the sharpest minds in African AI policy spoke with unusual candour. Before we can talk about "AI for Good," we must establish what the data shows. Kwame Opoku, founder of the African AI Governance Index (AAGI), has built the single, open-access platform the continent has needed for years, tracking governance readiness in real time across all 54 AU member states.³

AAGI's live dashboard tells a specific story. Approximately 20 to 22 of 54 African countries have published a national AI strategy, with a further 8 to 21 in the drafting stage.⁴ While 49 countries endorsed the Africa Declaration on AI at the inaugural Global AI Summit in Kigali in April 2025, that is a political signal, not a governance reality. AAGI's AI Risk Monitor has already verified 158 AI-related incidents across 11 African countries — harms accumulating faster than frameworks are being built to address them.⁵

The infrastructure picture is equally precise. Africa accounts for 0.6 percent of global installed data centre capacity while housing 18 percent of the world's population.⁶ AAGI's Infrastructure Tracker maps 223 data centre facilities across 38 countries, a market worth approximately $3.49 billion in 2024, with capacity concentrating in just four countries: South Africa, Nigeria, Kenya, and Egypt.⁷ As Opoku observed at the summit and in his published work: we are writing governance frameworks for an AI economy the continent does not yet physically own.

Sub-Saharan Africa ranks ninth globally on AI readiness, with an average score of 32.70, compared to a North America average of 81.51.⁸ North and Southern Africa lead on policy maturity. East Africa (Rwanda, Kenya, Uganda) represents the next competitive tier. Rwanda is already breaking ahead. Uganda is not starting from the bottom of this picture. It is competing in the second group, with a narrow window to move up.

The $500 Billion Objective

Uganda has set a precise national ambition: grow GDP tenfold from approximately $50 to 60 billion today to $500 billion by 2040.⁹ Achieving this requires annual growth of 14 to 18 percent across fourteen consecutive years. No country achieves that through incremental improvement; it requires a structural shift. AI is that shift, not as a fifth sector, but as the enabling layer for agriculture, tourism, minerals, and health.

The logic is Uganda-specific. Agriculture employs 72 percent of the workforce and generates crop disease patterns, soil profiles, and smallholder farming data that no external AI provider holds. Health systems carry disease burden data of genuine global scientific value. Data that existing AI systems, trained on Western clinical datasets, systematically misread. In every case, Uganda's competitive advantage is not compute power; it is contextual data that the world's largest AI companies will eventually need.

The single most binding constraint is the funding gap. Uganda currently spends 0.17 percent of GDP on science and technology, against its own Vision 2040 target of 2.5 percent.¹⁰ This necessitates strategic sequencing: concentrate scarce resources at the application layer, not the infrastructure layer.

The Capital Stack: Two Markets, One Logic

Conflating capital markets is a primary reason investment stalls. The practitioners on my panel at the summit were consistent on this.

Upstream infrastructure (connectivity and compute) requires patient, blended public capital. The African Development Bank has deployed over $850 million in such structures across the continent, including the $618 million iDICE programme in Nigeria and the Rwanda Innovation Fund, both instruments designed to crowd in private investment once public capital de-risks the entry point.¹¹ Uganda has not had an equivalent conversation with AfDB at scale. The gap is not eligibility; it is governance readiness.

The scale of what is available, if the conditions are right, is significant. In February 2026, AfDB and UNDP jointly launched the AI 10 Billion Initiative at the Nairobi AI Forum: a financing and partnership platform targeting up to $10 billion by 2035, structured around five interlinked enablers: data, compute, skills, trust, and capital, and projected to unlock up to 40 million jobs and a $1 trillion increase in Africa's GDP.¹² The AfDB's own roadmap identifies 2025 to 2027 as the critical "ignition window", the phase in which pilot projects, policy frameworks, and compute infrastructure must be initiated or the economic gains shift to faster-moving regions.¹³ The EAC, simultaneously, is developing a Regional AI Fund designed to mobilise blended finance and crowd in private capital to scale homegrown AI innovations from pilot to market.¹⁴ Uganda, as a founding EAC member leading the bloc's AI strategy development, is positioned to benefit directly from all three instruments but only if its governance and institutional readiness meet the threshold these mechanisms require before capital is deployed.

Capacity building is the bridge between capital availability and capital deployment, and it is where the private sector has a role that DFIs alone cannot fill. Investment readiness organisations like Open Startup, whose BRAIN programme convened this summit, are doing the translation work: preparing founders to speak the language of institutional capital, structuring startups for the due diligence that AfDB-backed funds require, and connecting Ugandan AI ventures to the regional and global networks where deployment-stage funding flows. Malloron Nair of Disraptor made the operational point explicit at the summit: most Ugandan businesses have not yet completed the AI readiness work that makes them investable, which means the capacity building layer is not a complement to the capital stack — it is a prerequisite for it.

Dr. Bienvenu Soglo, who spent two decades shaping AI and digital policy at Intel's government affairs desk across Africa, provided the most technically grounded insight at the summit: dependencies are sequential, not concurrent. You cannot deploy AI without data, you cannot generate useful data without connectivity, and you cannot build connectivity without resolving the political economy of spectrum allocation. His prescription for Uganda was specific: start with the connectivity layer, govern that data before building AI systems on top of it, and design for frugal AI. Not all AI requires massive compute. The applications that will actually reach a Ugandan farmer are those designed for the infrastructure that exists today, not the infrastructure that might exist in five years.

Downstream, AfricInvest's Dali Lakhoua, whose firm has deployed $2 billion across Africa and backed deep AI companies including InstaDeep, acquired by BioNTech¹⁵, was direct: the countries that produce investable AI companies build governance institutions first, attach conditions to infrastructure deals second, and create the legal environment that makes an AI investment thesis viable third. The sequencing that works is DFI capital building upstream conditions, ecosystem builders like Open Startup creating the investment-ready pipeline, and venture capital deploying once both are in place.

The Sovereignty Question

The governance insight that sharpened most clearly in Kampala is this: Africa's challenge is misframed when treated as 54 national legislative processes running independently. It is an opportunity to exercise collective market leverage that no single African country possesses alone — and a deadline that is no longer theoretical. The EU AI Act's core obligations for high-risk AI systems take effect in August 2026.¹⁶ The Malabo Convention and related AU instruments are moving toward enforceability across 44 African nations.

The AU Continental AI Strategy, endorsed by the AU Executive Council in July 2024 and now in Phase I implementation (2025–2026), provides the architectural container focused on establishing governance structures, mobilising resources, and creating national AI strategies.¹⁷ Dr. William Wasswa, a member of the AU's High-Level Panel on Emerging Technologies (APET) and one of the strategy's architects, confirmed at the summit that the shift from guidelines to structured policies to legal frameworks with enforcement mechanisms is real and accelerating.¹⁸

The evidence on what makes governance actually work comes from the ground up. i4Policy worked with young people, civil society, and government officials together to co-design Rwanda's Startup Act before it was announced, using a participatory methodology applied across more than 50 countries.¹⁹ Policies designed with the people they govern are implemented by those people. Policies imposed from above are resisted by them.

The hardest version of the sovereignty question came from Walid Naffati of the Africa AI Commission. Two points the room had been politely avoiding. First: many African administrations have not yet digitalised their core processes. The data that AI systems would govern does not yet exist in usable form. Second, and sharper: if the most realistic path to connecting the 73 percent of Ugandans not yet online runs through LEO satellite — in practice, Starlink — are we routing national citizen data through infrastructure owned by a single American entrepreneur, without any governance framework that specifies what that means for Uganda's citizens? Walid also stressed that a robust national cybersecurity strategy is a non-negotiable prerequisite for any AI roadmap. It’s relevant to note that without securing the networks and databases that fuel AI systems, the technology could actually increase a nation’s vulnerability. Any economic gains or GDP growth achieved through AI could be wiped out in a matter of days by a major cyberattack on critical infrastructure or the compromise of sensitive data.

Dr. Soglo's answer was pragmatic: negotiate specific data governance obligations into every infrastructure agreement before it is signed — data classification frameworks that distinguish sensitive from non-sensitive data, and contractual audit rights that make sovereignty real rather than declared. Ambrose Ruyooka, Uganda's Permanent Secretary for ICT, put it most precisely: sovereignty means control over data regardless of where it is stored, auditability of AI systems built from that data, and the internal capacity to operate both.

Uganda holds an asset most of its peers do not. It is one of five African countries alongside Egypt, Rwanda, DRC, and Zimbabwe with full participating membership in ISO/IEC JTC 1/SC 42, the international committee where global AI standards are written before they are imposed.²⁰ That seat is leverage of a categorically different order than commenting on frameworks after adoption.

Three Actions, In Order

1. Attach governance to infrastructure: Contractual requirements for data sovereignty, audit rights, and interoperability must be negotiated before infrastructure deals close. Once a contract is finalised without these conditions, the leverage is permanently gone.

2. Institutional mandates, not advisory bodies: Convert national AI task forces into enforcement mechanisms with legal mandates and independent budgets. A strategy without an institution to enforce it is merely a document.

3. Application-layer concentration: Fund agriculture and health first. These sectors offer a defensible data advantage that foreign providers cannot replicate at the contextual specificity required.

The stories my grandfather listened to existed in the air and in memory. Most are gone now. Those that remain exist because someone chose to preserve them. Not for nostalgia, but because they understood that a culture without its own stories eventually starts living inside someone else's.

Africa is at that inflection point with AI. The systems being built now will mediate how the next generation of Africans learns, works, and understands themselves. We are ready. The architecture exists. What remains is the will to act as a continent, rather than fifty-four countries waiting to see what the others do first.

Hemish Dave is the CEO and Founder of Hey Kahani (heykahani.com) and a board member of Open Startup International. He moderated sessions on "Capital, Ecosystems & the Conditions for Scale" and "The State of AI Governance in Africa" at the Uganda DeepTech Summit, 2026. All speaker attributions reflect positions expressed at the summit and have not been formally reviewed by the individuals named.

—

Hemish Dave | Founder, Hey Kahani | Board Member, Open Startup International