Leapfrog 2.0: How Africa Could Win the AI Race That Nobody Expects Them To

---

In 2007, a Kenyan farmer named Wanjiru had never set foot in a bank. The nearest branch was a couple of hours away. The fees were too high. The paperwork assumed a financial history she did not have. Then a basic Nokia phone changed everything. With M-Pesa, launched that same year by Safaricom, she could send and receive money in minutes. No branch, no credit history, no legacy system to navigate, because there was no legacy system at all. Africa did not fix its banking problem. It simply skipped it.

That was not a fluke but a pattern, and as artificial intelligence reshapes every major industry on earth, the question worth asking is whether Africa is about to do it again. The clear assumption is no. AI needs compute, infrastructure, capital, and data, and only wealthy economies have all of that in abundance. But that same assumption may be overlooking similar instances such as M-Pesa.

Winning the AI race here does not mean producing the next OpenAI or matching Silicon Valley’s frontier model research. That is not the game. It means something more specific. Building locally owned and operated AI companies that solve African problems, capturing the value those companies create instead of exporting it to foreign tool providers, and growing AI-enabled productivity in agriculture, health, finance, and logistics fast enough to outpace what mature economies can manage while building off of legacy systems. By that definition, Africa has a real structural position. Not a guaranteed one. The advantages only convert into something real if governments treat the next decade of infrastructure investment as what it actually is, foundational.

The Clean Slate: Why Having Less Can Mean Building Better

Consider what the United States and Europe are actually dealing with. Banks running on COBOL code from the 1960s. Healthcare systems with patient records spread across dozens of incompatible databases. Supply chains built around machinery that was never designed to talk to any other machine. Retrofitting AI onto that kind of foundation is expensive, slow, and full of compromise. You end up building for what your existing systems will allow, not for what AI can actually do.

Africa, across large parts of the continent, doesnt have that problem in the same shape. The clean slate is real but uneven. South Africa, Egypt, and parts of Nigeria carry meaningful legacy infrastructure, especially in banking and telecoms, that looks closer to what the US and Europe deal with. The leapfrog opportunity sits where those systems were never built in the first place. Rural agriculture, primary healthcare, smallholder logistics, informal retail across most of East and Sub-Saharan Africa. That is still a large addressable surface, but the nuance matters. The point is not that the entire continent has a clean slate. It is that the clean slate exists in the sectors where AI is most likely to create new value rather than just optimize old value.

The World Economic Forum documented in 2025 that Africa has already proven it can move straight to more advanced models without unwinding what came before, pointing to mobile telephony and mobile payments as the clearest examples (World Economic Forum, 2025). The Brookings Institution put it more directly, arguing that Africa's track record of skipping outdated systems creates specific openings for AI-native adoption across agriculture, healthcare, and financial services (Brookings Institution, 2025).

A 2025 McKinsey analysis found that African institutions have in some cases moved to cloud infrastructure faster than counterparts in developed markets, precisely because they had no decades of on-premise investment to protect (McKinsey, 2025). A few specific examples make the point concrete. Standard Bank, Africa's largest bank by assets, has migrated significant portions of its core operations to AWS, which lets it deploy machine learning models for fraud detection and credit scoring without rebuilding the underlying stack each time. Safaricom runs M-Pesa on a cloud-native architecture that processes billions of transactions monthly. That is the exact kind of foundation that makes layering AI services on top, things like real-time risk scoring or personalized financial products, a software problem rather than an infrastructure one. Flutterwave and Andela built cloud-first from the start, meaning AI features are not retrofits but extensions of how their products already work. The implication is direct. When the underlying infrastructure is already cloud-native, the marginal cost of adding AI-enabled services is dramatically lower than it is for a US bank running parts of its core on mainframes.

To make this concrete, consider a hypothetical agritech startup in Kenya building an AI advisor for smallholder maize farmers. It does not need to integrate with a legacy agricultural extension bureaucracy because there largely isnt one at the scale needed. It builds directly on top of M-Pesa for payments, uses satellite imagery and weather data through cloud APIs, runs a Swahili-trained language model for the farmer-facing interface delivered over basic feature phones via SMS or USSD, and prices the service per season at a few dollars. A US equivalent would have to navigate USDA data systems, existing extension service relationships, established agribusiness software contracts, and farmer expectations shaped by decades of prior tools. The Kenyan startup builds AI-native from day one because there is no other version of itself to displace.

Sperling's analysis of technology leapfrogging identifies two conditions that consistently predict success: the absence of legacy systems and a population with unmet needs that a new technology addresses better than any available alternative (Sperling, 2025). Africa satisfies both across a wide range of sectors. The open question is whether the surrounding infrastructure can be built fast enough to capture that before the window closes.

The Demographic Edge That Cannot Be Manufactured

More than 60 percent of Africa's population is under 25, making it the youngest continent by a significant margin (Mo Ibrahim Foundation, 2025). Africa also has some of the highest birth rates in the world. By 2035, it is projected to have the largest working-age population globally. Every year, millions of young people are entering a labor market where AI is not a new technology disrupting what they know. It is the baseline.

That matters more than it sounds. The real challenge of AI adoption in mature economies is not technical. It is cultural. Workforces trained on legacy systems, managers who built careers before machine learning existed, organizations whose internal processes were designed for a different era. These are not problems you fix by buying software. They take a generation to work through. Africa does not have that problem in the same way. AfriCatalyst put it plainly in their 2025 analysis: Africa's young , and growing, population creates a foundation for AI adoption that other regions cannot engineer into existence if those young people get real access to tools and training (AfriCatalyst, 2025).

Abdullahi, in African Business, connects this to actual job creation, documenting how technology sectors across the continent are already absorbing young talent at rates that would not have seemed credible ten years ago. The mobile connectivity investment that has happened is starting to convert demographic potential into real economic activity. The question is whether AI-relevant infrastructure investment follows at the pace the window demands.

This is where government is the central actor, not a supporting one. A young population without broadband, without technical education, and without a job market connecting AI skills to real employment is not a competitive advantage. It is a demographic dividend that expires unused. The Mo Ibrahim Foundation argued in 2025 that governments need to treat AI infrastructure the same way previous generations treated roads and ports: as foundational public goods that the private sector will not build on its own (Mo Ibrahim Foundation, 2025). That framing is right. The private sector follows returns. It does not lay fiber in rural Tanzania if there is no business case for it.

The Condition That Makes or Breaks Everything

Leapfrogging is not automatic. M-Pesa worked because of a specific combination: regulatory openness, a motivated actor in Safaricom, and a population with an urgent problem that the technology solved better than anything else available. Remove any one of those and the story ends differently. The infrastructure requirements for AI adoption are also significantly more demanding than those for mobile money.

AI needs reliable electricity. It needs connectivity that reaches beyond major cities. It needs data that reflects African contexts, languages, and realities, not datasets scraped from Western internet usage and used to train models built for entirely different populations. And it needs local development capacity: engineers and institutions that can build tools suited to African problems rather than importing tools designed elsewhere.

The current gap on those requirements is large and worth quantifying. Roughly 600 million Africans, about 43 percent of the continent's population, still lack access to electricity, and rural electrification rates fall below 30 percent in much of Sub-Saharan Africa (International Energy Agency, 2024). Internet penetration sits around 40 percent continent-wide, compared to 90 percent or more in North America and Europe, and broadband speeds in many African markets remain a fraction of OECD averages. Data center capacity tells the same story. Africa hosts under 2 percent of the world's data center supply despite being roughly 18 percent of the world's population. These are the numbers any honest version of the leapfrog argument has to start from.

Business of Tech Africa made the point directly, warning that leapfrogging without these foundations does not produce competitive advantage. It produces dependency, where African economies consume AI tools built, owned, and controlled elsewhere, with limited ability to shape how those tools work or who captures the value (Business of Tech Africa, 2024). Nguyen and Abebe (2023), in a continent-wide study of AI adoption, found that progress has been concentrated in a small number of urban centers with reliable infrastructure, and largely absent in rural areas where most Africans live. That is not a side note. It is a description of what leapfrogging without sufficient foundations actually looks like on the ground.

The investment picture is shifting. Citi and other major financial institutions have flagged Africa as one of the most heavily invested emerging market regions in the world, with capital flowing into digital infrastructure at rates that would have seemed unlikely five years ago. The African Union adopted a Continental AI Strategy in 2024 calling for coordinated investment in infrastructure, datasets, computing platforms, and human capital. The World Economic Forum reported that Africa's digital payment networks surpassed 1.1 billion mobile users and over one trillion dollars in transactions in 2024 (World Economic Forum, 2025). These are not proof the preconditions are in place. They are early signals that they are being built.

The Counterargument, and Why It Cannot Be Skipped

Nobody has a clean answer on this yet. AI as a mainstream technology is maybe five years old and Africa's relationship with it is even newer. Ask economists, ask professors, most will hedge, not because it is a bad question but because the data is too thin to be certain either way. That cuts both ways though. The skeptics do not have a clean answer either. What we do have is a clear set of material constraints worth naming directly.

The most direct pushback came from the head of the World Bank. At the institution's 2025 Fall Meetings, President Ajay Banga said: "Everybody's talking about big AI. What does big AI need to be successful? It needs computing power, lots of it. It needs electricity, lots of it. It needs data, lots of it. You tell me how many emerging market countries have these four, and I'll give you a medal" (Banga, as cited in Connecting Africa, 2025). That is a real challenge, not a fringe one.

The answer came from the floor of the same event. Driss Bengeloune, a digital innovation manager at MTN Benin, pushed back: "What made mobile money such an African innovation is that it emerged out of necessity, in a context where we don't have the same access to traditional banking. I believe that AI will be the same in Africa. We can use AI to resolve some local problems that we have" (Bengeloune, as cited in Connecting Africa, 2025). That exchange is the tension worth sitting with. Banga is right that the infrastructure is not there yet. Bengeloune is right that necessity has driven African innovation before.

The OECD's 2025 Africa Capital Markets Report adds numbers to Banga's point. In 2024, while global AI investment exceeded 100 billion dollars, Africa saw only one notable AI deal, valued under 100 million (OECD, 2025). That gap is enormous, and the leapfrog argument only holds if it is taken seriously rather than waved away.

So how does the leapfrog argument still hold given that gap. Three things are worth being specific about. First, what closing the gap actually requires. By most credible estimates, Africa needs roughly 30 to 50 billion dollars per year over the next decade for digital infrastructure alone, covering grid expansion, fiber, data centers, and the cloud regions that frontier models need to run efficiently. That is not unreachable, particularly if blended finance vehicles combining DFI capital, sovereign commitments, and private investment continue scaling. Second, which sectors are attracting what investment exists. Fintech still dominates, taking the majority of African venture capital, with health tech, agritech, and logistics following. AI investment within those sectors is starting from a low base but is the fastest-growing slice. Third, which countries are ahead and why. Kenya, Nigeria, South Africa, Egypt, and increasingly Rwanda are pulling ahead because they combine three things. Relatively coherent regulatory environments, established mobile money rails that reduce payment friction, and a critical mass of technical talent. The leapfrog will not happen evenly across 54 countries. It will happen first in five or six, and the question for the rest is whether they pursue policies that put them in the next tier.

A few of the most optimistic sources used here also have a bias worth flagging. The World Economic Forum, Brookings, and Mastercard tend to frame Africa's AI story as a global investment opportunity rather than as growth in African agency. That framing is not wrong, but it makes the leapfrog sound more market-driven and more inevitable than it is. It is worth keeping African governments and institutions at the center, not as recipients of foreign capital but as the primary decision-makers. And Business of Tech Africa and Nguyen and Abebe deserve to be treated as actual co-authors of the argument, not sources cited once to look balanced.

The point is not that Africa wins automatically. Africa has real structural edges that no mature economy can manufacture. Those edges only become advantages if the right investment decisions get made in the next decade.

Conclusion: Arriving Differently

In 2007, Wanjiru did not get a better bank. She got something that made the bank question irrelevant. That is what a real leapfrog looks like. It is not simply closing the gap with a new branch of a bank or offering new technology, it is arriving somewhere the incumbents never went.

Africa is not weighed down by the legacy debt forcing established economies to spend billions retrofitting what they built before they can build what comes next. Its workforce is not aging into a technology that arrived after their habits were already formed. In sector after sector, the conditions for building AI-native from scratch are more favorable here than anywhere else. The question is not whether those conditions exist. It is whether the decade ahead produces the investment in power, connectivity, data, and local capacity that turns structural position into actual competitive outcome.

Two specific investments matter most. The first is energy infrastructure, particularly distributed renewable generation paired with grid reliability upgrades. AI workloads, even modest inference at the edge, demand consistent power, and 600 million Africans without electricity is not a problem AI tools can route around. The second is technical education at scale. That means university capacity expansion for computer science and engineering, plus government-subsidized vocational pathways that connect to actual employer demand. Cloud infrastructure and data center capacity will follow capital, but capital follows the engineers and the power. Get those two right and the rest becomes a financing problem rather than a structural one.

For policymakers, the demographic window is not permanent. A generation without quality technical education and reliable broadband ages out of its timing advantage within twenty years. For business leaders, Africa is not primarily a consumer base for AI tools built elsewhere. The conditions for building competitive, locally owned AI companies are stronger than the current investment numbers suggest. And for young Africans: the timing is genuinely in your favor. Whether that means anything depends on what gets built next.

---

References:

Abdullahi, A. (2024, September 19). Unlocking Africa's potential: Creating jobs for youth through tech. African Business. https://african.business/2024/09/technology-information/unlocking-africas-potential-creating-jobs-for-youth-through-tech 

AfriCatalyst. (2025, August 1). Artificial intelligence is rewriting the future of work: What does that mean for Africa? AfriCatalyst. https://africatalyst.com/artificial-intelligence-is-rewriting-the-future-of-work-what-does-that-mean-for-africa/ 

Brookings Institution. (2025, April 17). Leveraging AI and emerging technologies to unlock Africa's potential. Brookings.  https://www.brookings.edu/articles/leveraging-ai-and-emerging-technologies-to-unlock-africas-potential/ 

Business of Tech Africa. (2024). The true cost of "leapfrogging" technologies in Africa. Business of Tech Africa. https://businessoftech.africa/the-true-cost-of-leapfrogging-technologies-in-africa-2/ 

Connecting Africa. (2025, October 16). World Bank chief 'not certain' about AI leapfrogging. https://www.connectingafrica.com/ai/world-bank-chief-not-certain-developing-nations-can-leapfrog-with-ai 

International Energy Agency. (2024). Africa Energy Outlook 2024. IEA. https://www.iea.org/reports/africa-energy-outlook-2024 

Mastercard Center for Inclusive Growth. (2025, August). AI in Africa: Harnessing the transformative power. Mastercard. https://www.mastercard.com/news/media/ue4fmcc5/mastercard-ai-in-africa-2025.pdf 

McKinsey & Company. (2025, May). Leading, not lagging: Africa's gen AI opportunity. McKinsey. https://www.mckinsey.com/capabilities/quantumblack/our-insights/leading-not-lagging-africas-gen-ai-opportunity 

Mo Ibrahim Foundation. (2025). The key to harnessing Africa's AI future: Leveraging the demographic dividend and investing in infrastructure. Mo Ibrahim Foundation. https://mo.ibrahim.foundation/news/2025/key-harnessing-africas-ai-future-leveraging-demographic-dividend-and-investing 

Nguyen, H., & Abebe, G. (2023, December 28). Empowering Africa: An in-depth exploration of the adoption of artificial intelligence across the continent. arXiv. https://arxiv.org/html/2401.09457v1 

OECD. (2025). Harnessing AI in finance for financial inclusion in Africa. Africa Capital Markets Report 2025. https://www.oecd.org/en/publications/africa-capital-markets-report-2025 

Sperling, S. (2025). Five rules for technology leapfrogging in Africa. Science. https://www.science.org/doi/10.1126/science.adz9028 

World Economic Forum. (2025, July). Africa is harnessing technology to leapfrog towards growth. https://www.weforum.org/stories/2025/07/africa-leapfrog-moment-harnessing-technology-green-growth-and-regional-integration-for-global-value-chains/