Africa's AI Adoption Map: What the Public Indices Actually Say About 14 Markets

Executive summary

The major public indices that score African countries on AI readiness do not agree with each other on the order, and they do not measure the same thing. Oxford Insights scores government and policy. GSMA scores connectivity and payments rails. Stanford counts research output and investment. The Tony Blair Institute writes a continent-specific lens. UNESCO publishes country-level Readiness Assessment Methodology reports. The AU and Smart Africa publish strategy. Briter Bridges and Partech track venture capital. Each one tells a partial story. Read together they tell a more honest one.

This report synthesises the published scores for 14 African markets across these indices. We do not produce a Coderex composite score. The numbers come from the cited sources. Our contribution is the reading.

The Oxford Insights Government AI Readiness Index 2024 ^[1] places Mauritius highest in Africa, with Egypt, South Africa, Rwanda and Senegal in the next band, and Kenya, Morocco, Nigeria, Tunisia and Ghana clustered in the middle. Ethiopia, Zambia, Tanzania and Uganda sit at the lower end of our 14-market sample. The spread between top and bottom inside Africa on this index is wider than the spread between most African leaders and the global median.

GSMA's Mobile Money State of the Industry 2024 ^[3] documents more than a billion registered mobile money accounts in Sub-Saharan Africa by end-2023, with transaction values that exceed those of several European countries. The GSMA Mobile Connectivity Index ^[2] places South Africa, Morocco and Tunisia at the top of their respective regional groupings. Partech ^[6] reports African startup tech equity funding at roughly $3.2 billion in 2024, concentrated in Nigeria, Kenya, Egypt and South Africa. The ITU ICT Development Index ^[13] cross-checks the connectivity reading. Read together these indices explain why the same country can look like a leader on one and a laggard on another. They are not measuring the same thing.

Coderex's view: for a multinational client planning a continental AI rollout, the strategic question almost always reduces to three to five named markets. The other markets are monitored, not deployed into. That is the framing the cited indices, read together honestly, actually support.

01 /Why one index is never enough

Anyone briefed on African AI readiness from a single index will get a distorted picture. Oxford Insights' Government AI Readiness Index ^[1] is the most cited, partly because it covers 188 countries and updates annually. It scores three pillars: government, technology sector, and data and infrastructure. A country that has passed an AI strategy and stood up a data protection authority will score well on the government pillar even if the underlying telecoms infrastructure is weak.

GSMA scores the rails. The Mobile Connectivity Index ^[2] looks at infrastructure, affordability, consumer readiness and content. The Mobile Money index ^[3] tracks the deepest digital payment system in the developing world. A country that ranks low on Oxford because of policy thinness can rank high on GSMA because of years of operator-led infrastructure work. Kenya is the cleanest example. Côte d'Ivoire is a quieter one. Senegal is too.

Stanford's AI Index 2025 ^[4] is built for a different question: where is frontier AI capability being produced. South Africa and Egypt show up in cross-country indicators on AI publications and patents. Most African markets do not appear at the global benchmark scale, which is itself a finding rather than an absence.

The Tony Blair Institute's State of AI in Africa 2024 ^[5] tries to fill the gap with a continent-specific lens. It is more generous to markets like Rwanda, Ghana and Senegal than the global indices are. UNESCO's Readiness Assessment Methodology country reports ^[9] do something different again: they document what is in place and what is missing in specific countries, sector by sector.

If a board is making a continental investment decision based on a single index, they are almost certainly mispricing the market. The fix is to read at least three of these in parallel and look at where they disagree.

02 /The seven markets with a workable case across multiple indices

Nigeria, Kenya, Egypt, South Africa, Rwanda, Ghana and Morocco are the seven markets where a multinational client can build an AI-led product or operation today with reasonable confidence that the supporting ecosystem will not be the limiting factor. The reasoning differs by market and the index disagreements matter.

Nigeria. Lagos has the deepest engineering and ML talent pool in Africa by absolute headcount. Stanford's AI Index 2025 ^[4] indicators on talent flow and investment activity place Nigeria in the African top tier. The Oxford Insights 2024 score places Nigeria in the middle band on government readiness, dragged down by data infrastructure and lifted by tech sector strength ^[1]. Partech's 2024 funding report ^[6] shows Nigeria as one of the four largest recipients of African tech equity. Currency volatility and power reliability are the binding operational constraints, not talent.

Kenya. Nairobi has the most operationalised data protection regime in East Africa. The Oxford Insights 2024 score places Kenya in the middle band overall but it ranks higher on the data and infrastructure pillar than on the government pillar ^[1]. Mobile money penetration in Kenya remains the global benchmark; M-Pesa data flows are the substrate for a lot of fintech and identity-adjacent AI work ^[3]. Capital availability is thinner than Lagos but the regulatory clarity is better.

Egypt. Cairo has strong technical talent. The Oxford Insights 2024 score places Egypt in the upper band of African countries on technology sector capacity ^[1]. Stanford's indicators show Egypt with the second-highest research output footprint in Africa ^[4]. The binding constraint is regulatory friction on cross-border data movement and a less predictable enforcement posture. Tony Blair Institute commentary ^[5] flags this as the dominant cited concern from operators.

South Africa. Cape Town and Johannesburg both have credible AI talent. POPIA gives South Africa the most operationalised data protection regime on the continent. Oxford Insights places South Africa in the top African band on the government and data pillars ^[1]. GSMA Mobile Connectivity Index ^[2] places South Africa first in Sub-Saharan Africa. Capital availability has tightened over recent years as some local funds have wound down. The binding constraint is currency and macro pressure rather than capability.

Rwanda. Kigali punches above its weight on the government pillar. Oxford Insights places Rwanda in the second band of African countries on overall readiness, lifted heavily by government engagement ^[1]. Tony Blair Institute commentary ^[5] is more positive on Rwanda than the global indices because the policy environment is genuinely more conducive than the talent pool depth would suggest. The talent pool is the binding constraint; English-language ML resource access is workable but the absolute pool is small.

Ghana. Accra has a small but growing ML community. Oxford Insights places Ghana in the middle band on overall readiness ^[1]. The regulatory environment is workable but does not have specific AI provisions yet. AI4D scoping work ^[10] has highlighted Ghana as one of the more promising secondary markets in West Africa. Capital is the binding constraint.

Morocco. Casablanca and Rabat have respectable technical capacity. Oxford Insights places Morocco in the middle band ^[1]. GSMA Mobile Connectivity Index ^[2] places Morocco at the top of North Africa alongside Tunisia. The market gets overlooked because most ecosystem reports default to Anglophone Africa, and ML resource access is mostly in French. We rate it higher than its press coverage suggests, and the cited indices broadly support that view.

The seven-market list is not a ranking. It is a list of markets where the public indices, read in combination, do not give a board a reason to walk away.

03 /The seven cluster markets

Tanzania, Uganda, Senegal, Côte d'Ivoire, Ethiopia, Tunisia and Zambia look very different from each other once you look past the regional grouping. Treating them as a single cluster is a research convenience, not a strategy.

Senegal. Dakar has stronger Francophone tech depth than Anglophone press tends to capture. Oxford Insights places Senegal in the second African band on overall readiness, particularly on government ^[1]. Tony Blair Institute commentary ^[5] is positive. The binding constraint is capital flow and the language drag on global ML resource access.

Côte d'Ivoire. Abidjan is one of the quietly growing financial and tech centres in Francophone West Africa. Oxford Insights places Côte d'Ivoire in the middle-lower band ^[1]. Mobile money penetration has grown materially per GSMA SOTIR ^[3]. The market behaves more like a secondary deployment market than a primary one.

Tunisia. Tunis has a small but capable engineering community. Oxford Insights places Tunisia in the middle band ^[1]. The regulatory environment is in flux and the macro situation has been difficult. We have seen practitioners advise caution on long-horizon investment until the regulatory direction is clearer.

Tanzania, Uganda. Both score in the middle-to-lower band on Oxford Insights ^[1] and middle band on GSMA Mobile Connectivity ^[2]. Mobile money is well established in both. Beyond mobile money, the AI ecosystem moves project by project rather than at any platform level. AI4D scoping reports ^[10] have flagged both as markets where targeted programme funding can move the needle disproportionately.

Ethiopia. Almost every published index places Ethiopia near the bottom of the African field on AI readiness. Oxford Insights places it among the lowest scores in our 14-market sample ^[1]. The market is large and the underlying demographics are strong. The binding constraints are connectivity, regulatory environment and political risk. We do not currently see a credible operator case for direct enterprise AI deployment in Ethiopia at scale, though the country features in continental and AU-level strategy ^[7].

Zambia. Lusaka has interesting healthcare data work, partially funded through IDRC and AI4D programmes ^[10]. Oxford Insights places Zambia in the lower band ^[1]. Commercial AI capacity is limited.

The headline for the cluster: do not treat them as a single market, do not assume that doing well in Anglophone Africa transfers to Francophone Africa or vice versa, and do not expect any of these seven to be primary deployment markets in the next two years on the basis of the published indices.

04 /Capital flow and where it goes

Partech's 2024 Africa Tech VC Report ^[6] places African tech equity funding at roughly $3.2 billion for the year, the third consecutive annual contraction and roughly the level the continent last saw in 2019, before the 2021–22 venture spike. Nigeria, Kenya, Egypt and South Africa absorbed the large majority. Briter Bridges' tracking ^[11] is consistent with that order of magnitude though it differs on category boundaries between fintech-led AI work and pure-play AI.

The capital flow story has two implications. First, the four-market concentration in funding is much sharper than the readiness indices would suggest; the indices give Mauritius, Rwanda and Morocco respectable scores but the capital does not flow there at scale. Second, the absolute level of African AI investment is small in a global context. Stanford's AI Index 2025 ^[4] documents private AI investment at hundreds of billions globally with Africa registering at low single digits of percent. The base is small. The trajectory matters more than the level.

The World Bank's Digital Economy for Africa programme ^[12] is one of the larger DFI-coordinated efforts on the supply side. It is not a venture capital substitute. It is infrastructure spend with a longer payback than commercial capital tolerates.

The capital concentration matters because most African AI work today is built on top of fintech, healthtech and mobility platforms that themselves depend on rails the GSMA reports document. Capital flowing into the four-market core funds those platforms, and the AI work that sits on top inherits the rail. A market with strong readiness scores but no platform layer (Rwanda is the cleanest example, Morocco a partial one) absorbs capital differently. The deals are smaller, the time to operating leverage is longer, and the exit assumptions are less mature. None of this is in the indices. It is in the capital flow data when read against the Mobile Money report ^[3] and the Stanford ^[4] global investment context.

A second-order effect: African companies that raise outside the four-market core often do so in dollars from London, New York or Paris-based investors, which exposes them to the same currency volatility we flag in section 5. The Mo Ibrahim Foundation's 2024 forum report ^[15] on financing Africa's digital and energy transitions covers this dynamic at the level it deserves. Boards modelling AI investment in Africa should read it alongside the readiness indices, not as a substitute.

05 /What the indices all miss

Five things, consistently.

The diaspora effect. Lagos and Nairobi capability is materially lifted by engineers and researchers who spent five to ten years in US, UK or European labs and then returned. Stanford's AI Index ^[4] does not capture this directly. Oxford Insights does not. Tony Blair Institute commentary ^[5] gestures at it. The effect is most concentrated in Nigeria and South Africa, then Kenya, then Egypt.

Francophone language drag. The vast majority of state-of-the-art ML resources, papers, libraries, online courses and frontier-lab documentation is in English. African STEM graduates in Anglophone systems engage with that material directly. Senegalese, Ivorian, Tunisian, Moroccan and Cameroonian graduates do not always. None of the cited indices weight this directly.

Political risk. Several markets in the 14 carry political risk that is non-trivial and that does not show up in policy-oriented scoring. Oxford Insights' government pillar ^[1] captures whether a country has an AI strategy. It does not capture whether the cabinet that wrote the strategy will still be there in 18 months.

Currency volatility. Nigeria's currency has eroded dollar returns on local-currency investment materially over recent years. Egypt's currency situation has been similarly difficult. The indices do not score this.

Power and infrastructure reliability. Oxford Insights and GSMA score infrastructure at the policy and access level. Neither scores how often the power actually goes off in Lagos or how that affects on-prem deployment. Operators know. Boards reading indices often do not.

These five gaps are not arguments against the indices. They are arguments for triangulating between indices and operator commentary rather than relying on a single league table.

06 /The continental strategy layer

The African Union Continental AI Strategy ^[7] was endorsed in 2024. It sets direction at policy level on harmonisation, capacity building and ethics. The Smart Africa Manifesto ^[8] sits alongside it as a more operational layer. Neither moves a country up the Oxford Insights rankings on its own. Both matter for understanding where the policy environment is heading rather than where it currently is.

UNESCO's Readiness Assessment Methodology country reports ^[9] are a useful counterpart because they go deep on individual countries rather than ranking them. Where they exist for our 14 markets, they give a more textured picture than any global league table.

The IDRC and AI4D Africa scoping reports ^[10] document the supply side at programme level: what is being funded, where, by whom, and on what theory of change. They are not readiness scores but they explain why some markets are moving faster than the indices alone would predict.

07 /How to use this map

For a multinational client planning a continental rollout, the practical takeaway is to do the work the indices do not. Pick the three to five markets where the strategy actually carries. Then read the country-specific UNESCO RAM report ^[9], the relevant Tony Blair Institute commentary ^[5] and the most recent Partech ^[6] and Briter ^[11] capital flow data for those markets in detail. The Oxford Insights overall score ^[1] is a screening tool, not a planning tool.

For a policy-side reader, the takeaway is similar in shape but different in content. The indices are useful for benchmarking direction of travel. They are less useful for setting strategy because they reward the things that are easy to measure (a published AI strategy, a data protection law on the books) more than the things that are hard to measure (whether either is being enforced).

For an investor, the cleanest signal in the published data is the gap between Partech-style capital flow concentration ^[6] and Oxford Insights-style readiness scoring ^[1]. Where capital is concentrated and readiness is high, the trade is competitive. Where readiness is high and capital has not yet flowed, the trade is harder to access but cheaper to enter. Rwanda, Morocco and parts of Senegal show up in that second category at present.

Coderex advises operators, investors and ministries on which three to five markets carry weight for their specific deployment or policy intervention, where the index disagreements matter most, and what diligence beyond the league tables is worth doing. Boards making a continental decision on a single index are almost certainly mispricing the market; the value is in reading the disagreements honestly.

The 2025 editions of the Oxford Insights Government AI Readiness Index and Stanford's AI Index will publish later this year. Expect the seven-market core to hold. Expect Côte d'Ivoire and Senegal to move up modestly on the Francophone-aware indices as Tony Blair Institute and IDRC AI4D country work matures. Expect Ethiopia and Zambia to remain near the bottom of African readiness scoring without a structural break in connectivity or political risk. Expect at least one of Nigeria, South Africa or Egypt to formally enforce a data protection or AI provision before the next index cycle, which will reshape the data and infrastructure pillar scoring.

Methodology

This report synthesises published, public scores from the listed indices. Coderex did not generate any composite or proprietary readiness score. Where we describe a market as being in a particular band, that band is taken from the cited index, not from a Coderex-internal model.

We covered 14 African markets: seven with full treatment (Nigeria, Kenya, Egypt, South Africa, Rwanda, Ghana, Morocco) and seven with cluster treatment (Tanzania, Uganda, Senegal, Côte d'Ivoire, Ethiopia, Tunisia, Zambia). The full-treatment selection follows the markets that all of the cited indices cover with usable data; the cluster selection follows markets where index coverage is thinner but where AI ecosystem activity is documented.

Limitations: index editions update at different cadences and not all of the cited indices were on a 2024 or 2025 release at the time of writing. Where multiple editions exist we cite the most recent we could verify. Where indices disagree on rank order or band membership for a given market, we say so in the body rather than averaging across them. Operator-level commentary that is not in the cited indices is marked as such in the relevant paragraph rather than absorbed into the index numbers.

We did not interview operators specifically for this report. Future updates of this map may include a primary-research layer.