Why Africa is the next AI investment frontier (and why the case is narrower than the headlines)

Executive summary

Clients with capital allocation decisions to make about Africa keep asking the same question, in slightly different forms: is the African AI investment thesis real, or is it the next frontier-market story that does not deliver? We have tried to answer it without overclaiming.

The substrate is stronger than five years ago. The median age in sub-Saharan Africa is roughly 19 and the EU is in its mid-40s, on UN World Population Prospects 2024 ^[1]. Sub-Saharan smartphone adoption reached 51% in 2023, on GSMA Intelligence data, with continued growth projected through 2030 ^[2]. Sub-Saharan Africa held more than a billion registered mobile money accounts at the end of 2024, with annual transaction values exceeding $1.1 trillion, according to the GSMA State of the Industry Report on Mobile Money ^[3]. Subsea capacity has expanded materially with the Equiano landings from 2022 ^[4] and the 2Africa rollout through 2023 and 2024 ^[5]. Six African countries have published national AI strategies ^{[8][9][10][11][12][18]}. Partech recorded $3.2B in startup equity across 427 deals in 2024 ^[6], with independent triangulation from Africa: The Big Deal ^[7].

The headline case is narrower than the public discourse. African AI capital is a small fraction of the US and Chinese totals reported in the Stanford AI Index ^[17]. The Naira devaluation since 2023 has compressed dollar returns on local-currency investments, with the IMF noting persistent FX and inflation pressure through 2024 ^[15]. Senior engineering talent flight to US-headquartered remote employers is documented at four-to-six-times salary differentials in the Endeavor and TLcom Inflection Point report ^[16]. Infrastructure remains uneven. Regulatory direction is improving but quality varies.

What we believe with high confidence: the risk-adjusted return profile in selected African markets, for patient capital with local operational support, is meaningfully better than it was a decade ago. What we are less sure about: the timing of the inflection, the durability of the recent regulatory wave, and whether the pan-African AI champions will emerge from Lagos, Nairobi, Cairo, or Cape Town first.

01 /Demographics and the labour market case

The starting point is not AI. It is people. The UN's World Population Prospects 2024 puts the median age in sub-Saharan Africa at roughly 19, against around 44 in the European Union ^[1]. By 2050, on UN medium-variant projections, sub-Saharan Africa is expected to account for a steadily rising share of the global working-age population, on the order of one in four by mid-century ^[1].

This matters for AI investment for a reason that gets confused often. AI capability does not follow young populations directly. It follows labour markets, consumer markets, and capital allocation decisions made by firms that need to be where the customers and the workers are. McKinsey Global Institute's 2023 work on African economic growth makes a similar point about productivity and demographic dividends being conditional on capability investment, not automatic ^[19].

The right way to think about the demographic case is as a forcing function on capital, not as a guarantee. Multinational employers will increasingly need African talent. African consumers will increasingly need digital services priced for their incomes. AI tooling that solves problems at the price points and infrastructure conditions of African markets will have a domestic addressable market large enough to support venture-backed scale, even if return profiles look different from US peers.

There is a secondary effect that is harder to model and that we keep watching. The diaspora connection between African and OECD tech ecosystems is denser than in most emerging-market histories. Lagos and Nairobi engineers spend two to five years in London, Berlin, or San Francisco, then either relocate back or maintain working relationships that flow capability home. We will not put a number on this because we cannot, but it shows up in our hiring funnel and in the founder profiles of the better-capitalised African AI startups.

02 /Connectivity, mobile money, and the infrastructure substrate

If AI is going to be useful in African markets, it needs three things to work: connectivity, payments, and identity. The first two have moved further than most external observers credit.

GSMA Intelligence's Mobile Economy Sub-Saharan Africa 2024 puts smartphone adoption at 51% in 2023, with continued growth pushing the number above 80% by the end of the decade on their central scenario ^[2]. Mobile internet penetration sits below smartphone adoption because of affordability and coverage gaps, but the trajectory is consistently upward.

Mobile money is the more striking number. The GSMA State of the Industry Report on Mobile Money 2025 reports more than 1 billion registered mobile money accounts in sub-Saharan Africa at end-2024, with annual transaction values exceeding $1.1 trillion ^[3]. That is the largest digital payments rail outside India and China. For AI applications that need transaction signal, behavioural data, or settlement infrastructure, the rails are now in place in ways that they were not a decade ago. Anyone building credit risk models, fraud detection, or merchant analytics in markets like Kenya, Tanzania, Ghana, and Côte d'Ivoire is now working on top of payments infrastructure that compares favourably to many OECD economies on coverage.

Subsea capacity is the third leg. Google's Equiano cable started landing on African coasts in 2022, with the first landing in Togo announced by Google's Africa team and subsequent landings in Nigeria, Namibia, South Africa, and Saint Helena ^[4]. The 2Africa system, led by a Meta-anchored consortium that includes Vodafone, China Mobile, MTN GlobalConnect, Orange, stc, Telecom Egypt, and WIOCC, is one of the longest subsea cable systems ever built and has been progressively bringing landings into service through 2023 and 2024 ^[5]. The aggregate effect is that international bandwidth into West, East, and Southern Africa has been transformed.

The CIO-level implication, which we keep finding clients have not updated their thinking on, is that latency from Lagos and Cape Town to major European cloud regions is now competitive with intra-European latencies. Cloud-hosted AI workloads serving African users from European or Middle Eastern regions are no longer materially handicapped on round-trip time. That changes the architecture options for any pan-African AI deployment significantly.

We are not claiming the infrastructure problem is solved. Power reliability still affects on-prem deployments unevenly. Last-mile fixed broadband remains thin outside major cities. National data centre capacity is patchy. The Microsoft and G42 announcement in May 2024 of a $1 billion investment in a Kenyan data centre is one of several recent signals that this is being addressed at scale ^[20]. But the wholesale picture has shifted enough to make it worth re-running assumptions that were set in 2018 or 2019.

03 /Talent supply, education, and the diaspora effect

We will not repeat the unverifiable per-country STEM graduate figures that some Africa-investment decks circulate. The UNESCO Institute for Statistics keeps the most reliable cross-country tertiary education and graduation data, and the picture they show is a steadily expanding tertiary STEM pipeline across the major African economies, with Egypt, Nigeria, and South Africa among the larger producers in absolute terms ^[14]. The depth of the pipeline at the top of the distribution, which is what matters for frontier AI work, is harder to measure but improving.

A few structural advantages are worth naming.

English-medium instruction in technical fields. Across most of Anglophone Africa, university teaching in computer science and engineering is in English. The vast majority of state-of-the-art ML resources, papers, and online courses are in English. African STEM graduates can engage with that material directly. This is not unique to Africa, but it does materially reduce the capability transfer cost compared with non-English-medium markets. Egypt, Morocco, Tunisia, and the Francophone West and Central African states sit on a different curve and the implications for AI talent show up in our own hiring practice.

Diaspora flow. Young African engineers move back and forth between African and OECD tech hubs at higher rates than the equivalent flows in many comparable middle-income regions. This is partly cultural, partly visa-driven, partly the way networks have built up around hubs like Lagos, Nairobi, Cape Town, and Cairo. The Endeavor Nigeria and TLcom Capital Inflection Point report from 2024 is the best public source we have found on the dynamics of this flow ^[16].

Talent flight is the flip side. The same Endeavor and TLcom analysis documents senior engineering pay differentials of roughly four to six times between local Lagos or Nairobi roles and remote employment with US-headquartered firms ^[16]. This is not abstract. We have seen it in our own hiring funnel and in clients' retention numbers. It is a permanent feature of the landscape rather than a temporary phase, and any African AI capability strategy that assumes it can match Bay Area total compensation locally will struggle.

The honest framing is that talent depth is improving but not yet at the level of, say, India in the early 2010s when global firms first scaled out engineering centres there. The pipeline will keep deepening. The diaspora connection compounds. The pay gap is a constraint that African employers will need to manage with non-cash levers, including local equity, mission, and quality of working conditions, rather than expecting it to close.

04 /Policy and regulation: a faster wave than is widely recognised

Policy is where the change has been fastest and is most under-recognised by external investors who set their priors before 2023.

Mauritius published the first national AI strategy on the continent in November 2018 ^[18]. Rwanda adopted a National AI Policy in April 2023 covering data governance, ethics, and infrastructure ^[12]. Egypt published the second edition of its National AI Strategy in 2025, covering 2025 to 2030, updating the original 2021 strategy ^[8]. NITDA released Nigeria's National AI Strategy in August 2024 ^[10]. South Africa's Department of Communications and Digital Technologies released a National AI Policy Framework in 2024 ^[11]. Kenya launched its first National AI Strategy in March 2025 ^[9].

That is six national AI strategies across six markets that together account for a large majority of African economic activity, all published or formally adopted within a seven-year window with most concentrated in 2023 to 2025. Quality varies. The Mauritius document is more sectoral and public-procurement-focused. The Rwandan policy is more comprehensive on ethics. The Nigerian and Kenyan strategies lean implementation-heavy with named workstreams. The South African framework reads as a lighter-touch coordination document with detail expected in subsequent regulation. Egypt's update is the most ambitious in scope.

Oxford Insights' Government AI Readiness Index 2024 places the leading African markets in respectable but mid-table global positions, with Mauritius, South Africa, and Egypt typically scoring highest among African states ^[13]. The picture for African states overall trails OECD economies but has improved consistently year over year on Oxford Insights' methodology.

The implication for investors is twofold. First, the regulatory direction of travel is consistent enough that investments made in 2026 are unlikely to be wrong-footed by sudden hostile policy shifts in the leading markets. Second, the gap between published strategy and operational regulation is real, particularly on data protection enforcement, cross-border data flow rules, and AI-specific compliance regimes for healthcare and financial services. We treat the strategies as a signal of direction, not as a substitute for ground-level due diligence.

A separate point on data protection. Most of the leading African markets either operate under a comprehensive data protection regime (POPIA in South Africa, the Nigeria Data Protection Act 2023, Kenya's Data Protection Act 2019) or are in the process of operationalising one. The interaction between these regimes and AI training data has not been litigated in most of these jurisdictions. We expect the first major test cases to land in 2026 or 2027 in Nigeria or South Africa.

05 /Capital flow, valuations, and what got funded

Numbers first. Partech Africa's 2024 Africa Tech Venture Capital Report puts total African startup equity funding in 2024 at $3.2B across 427 rounds ^[6]. Africa: The Big Deal, the independent funding tracker run by Maxime Bayen and Max Cuvellier, recorded around $2.2B across companies raising $1 million or more in 2024 ^[7]. The two trackers use slightly different methodologies, with Partech including some grant and debt components in its broader measure. Both show the same directional pattern: 2024 was the third consecutive annual contraction, with totals roughly back at 2019 levels and well below the 2021 to 2022 peak.

Country concentration is the dominant feature. The Big Four of Nigeria, Kenya, Egypt, and South Africa absorb the large majority of capital each year, a pattern documented across multiple years of Partech data ^[6]. AI-specific funding is a smaller carve-out and is harder to track precisely because most rounds are sector-tagged (fintech, healthtech, agritech) rather than capability-tagged. We will not put a fake AI-specific number on it.

Stage distribution is informative. The 2024 Partech data shows the seed segment held up better than later-stage rounds, which is consistent with what we see in our own engagement portfolio: continued early-stage activity, with later-stage rounds harder to close given global venture conditions and African macro headwinds. Median round sizes in 2024 sit roughly flat year-on-year on the Africa: The Big Deal data series ^[7].

Valuations are the part of the market we are most cautious about. In our practitioner experience advising on diligence for institutional investors, headline valuations on the small number of credible African Series A and Series B rounds in 2024 and 2025 looked elevated relative to revenue multiples on later-stage exits in earlier vintages. The vintage may be repriced down. We do not know yet.

The Stanford AI Index 2024 puts global private AI investment at well into eleven figures annually, with the US share alone an order of magnitude above the entire African startup equity total ^[17]. The right framing is not that Africa is competing with the US for AI capital. It is that the absolute base of African AI capital, while small, is large enough to support several pan-African AI businesses through to credible exits over the next five years.

The composition of capital is shifting. Development finance institutions and impact-aligned vehicles have historically been a larger share of African venture funding than in the US or Europe, with commercial venture capital growing as a share over the last few years on the Partech data ^[6]. The composition matters because DFI-backed rounds tend to skew agritech, healthtech, and inclusive finance, while commercial rounds skew fintech, infrastructure, and consumer applications. The shape of the African AI startup population reflects which capital pool is dominant in any given vintage.

06 /Risks worth naming

We will not pretend the case is risk-free. Several things can break the thesis.

Currency volatility is the most acute. The Nigerian Naira lost a large portion of its dollar value following the 2023 FX reform, with the IMF's 2024 Article IV consultation noting persistent inflation and exchange-rate pressure through 2024 and into 2025 ^[15]. For dollar-denominated investors holding local-currency revenue, this has compressed returns materially. Egypt has had similar pressure on the pound. Kenya and South Africa have been more stable. Currency hedging is expensive and incomplete, and the right structural response is usually to underwrite to hard-currency revenue or to accept a wider return distribution.

Talent flight is the second. The four-to-six-times pay gap between local senior engineering roles and US remote roles is documented in the Endeavor and TLcom Inflection Point report ^[16]. The structural answer is not to compete on cash. It is to compete on ownership, mission, and work environment, and to design organisations that can develop talent through several seniority bands without losing the senior layer to remote arbitrage.

Regulatory ambiguity is the third. The wave of national AI strategies through 2024 and 2025 is encouraging ^{[8][9][10][11][12]}, but the gap between policy and operational regulation is real, and several of the strategies will need to be translated into specific compliance regimes for AI in healthcare, financial services, and public sector procurement. Investors who underwrite specific regulatory outcomes should expect to be wrong some of the time.

Infrastructure unevenness is the fourth. Power reliability, last-mile broadband, and data centre capacity are all materially better than they were a decade ago, but they are not at OECD reliability levels and they affect deployment architectures meaningfully. The Microsoft and G42 May 2024 commitment of $1 billion to a Kenyan data centre and digital ecosystem is an example of the scale of investment now being directed at these gaps ^[20], but the gaps will close gradually rather than suddenly.

Political risk varies sharply by market. Some markets sit at the more stable end of the spectrum on standard country risk indices. Others have either had recent political transitions that warrant cautious diligence or have ongoing security concerns in parts of the country. We do not score this in detail in this report. We note it as a material consideration and recommend that any continental allocation strategy treat individual markets discretely rather than aggregating.

Concentration risk is the fifth and most tractable. The Big Four absorb the majority of capital each year ^[6]. An investor with an Africa-AI mandate and a small team can credibly cover Lagos, Nairobi, Cairo, and one of Cape Town or Johannesburg with two or three locally-based partners. Trying to deploy across fifteen markets with the same staffing model produces shallow coverage and poor outcomes. The fund managers we see doing well are the ones who have made this concentration explicit.

A final note on timing. We have been asked, mostly by clients in London and New York, whether the African AI thesis is something to commit to in 2026 or to wait until 2028. The honest answer is that it depends on whether the investor has the operational capacity to support African portfolio companies actively or is hoping to deploy capital into a more developed market structure later. The market structure will keep developing. Several of the better deals will price tighter once it does. Patient capital with operational support has historically captured the better risk-adjusted returns in this kind of frontier and the next decade is unlikely to break that pattern.

07 /What to take from this

The substrate for AI capability development in selected African markets is materially stronger than five years ago. The headline returns over the next decade will not match US or Chinese AI investment in absolute terms. The risk-adjusted return on patient capital with operational support, deployed across three to five named markets rather than fifteen, is the actual case.

For boards setting continental capital allocation in the next budget cycle, the binding constraint is not whether to invest but how. The investors who do well will be those with offices in two or three African cities, locally hired partners with deal authority, and patience to underwrite through currency cycles. Coderex advises operators, investors and ministries on which markets carry weight for their use case, where the regulatory direction of travel is most credible, and how to design African presence at the operational scale that frontier returns actually require.

Expect the seven-market core (Nigeria, Kenya, Egypt, South Africa, Rwanda, Ghana, Morocco) to absorb most disclosed AI-relevant capital through 2027. Expect at least one of Nigeria, South Africa, or Egypt to issue substantive AI-specific regulation in healthcare or financial services before 2028. Expect the announcement-to-deployment gap from the 2023–24 vintage to surface during 2026 and 2027 as that cohort hits the eighteen-month mark, repricing several rounds. Expect the first credible billion-dollar African AI exit in this cohort to land closer to 2028 than 2026.

Methodology

This report draws on publicly available sources verified as of 29 April 2026. Demographic data are from the UN Department of Economic and Social Affairs World Population Prospects 2024 release. Mobile and mobile money data are from GSMA Intelligence and the GSMA State of the Industry Report on Mobile Money 2025. Subsea cable references are to public announcements from Google (Equiano) and the 2Africa consortium press materials.

National AI strategy citations are drawn from the official government publications of Egypt (MCIT), Kenya (Ministry of Information, Communications and the Digital Economy), Nigeria (NITDA), South Africa (DCDT), Rwanda (MINICT), and Mauritius (MTCI). Each strategy was checked for publication date and scope.

Capital flow figures are from the Partech Africa Tech Venture Capital Report 2024 and Africa: The Big Deal annual data, with the Stanford AI Index 2024 used for global comparison. McKinsey Global Institute's 2023 work on African economic growth was used as background for the labour-market and productivity section. The IMF's 2024 Article IV consultation on Nigeria was used for the currency volatility section. The Endeavor Nigeria and TLcom Capital 2024 Inflection Point report was used for the talent flight section.

Limitations. AI-specific funding figures within the African venture totals are not separately tracked at the level of granularity that the report would have liked. The trackers report sector tags (fintech, healthtech) rather than capability tags. We have therefore reported total African startup equity figures and treated AI-related share as estimable but not directly cited. The original draft of this article cited a $612M Q1 2026 number that did not correspond to any verifiable underlying tracker; that figure has been removed from this version and replaced with full-year 2024 data from Partech and Africa: The Big Deal.

Per-country STEM graduate counts were also reframed. The original draft cited specific 2024 counts for Nigeria, South Africa, Egypt, and Kenya that we could not verify against UNESCO UIS or national university commission data with sufficient precision. We have therefore reported the directional pattern from UNESCO UIS without the specific numbers.

This report does not score consumer adoption, Francophone or Lusophone markets in depth, or political risk at the country level. Each of these may be added in subsequent updates.