The Continent That Skipped the Queue

The countries winning the AI era aren't the ones that missed the last revolution — they're the ones with nothing left to protect. Africa is finding out what that's worth.

10 June 2026

Every conversation about Africa and technology starts from the same position. What the continent missed. The industrial revolution that didn’t arrive. The infrastructure that wasn’t built. The capital that went elsewhere.

That framing is not just incomplete. It is looking at the wrong variable.

The countries moving slowest into the AI era are not the ones that missed the last revolution. They are the ones that won it. Legacy infrastructure is not an asset right now. It is a weight. The continent with the least to protect is finding out it has the most room to move.

The debt that comes with winning

Europe built the most sophisticated regulatory architecture in the world for the digital era. It also built it around companies it did not create, industries it was not leading, and a competitive position it was already losing. The AI Act, fully applicable from August 2026, is the outcome of that position: a framework designed to make AI safe enough to allow, built by a bloc that does not yet own the companies making AI worth regulating.

The published record on this is unambiguous. As Relefy documented earlier this year in Three Continents, Three Priorities, One AI Race, Europe grows the talent and the early-stage companies. American capital owns them at scale. The gap is not in capability. It is in who controls the outcome.

That dynamic is the product of decades of choices made by governments and institutions trying to protect what they already had. Protecting something requires having built it first. Africa, in most sectors, did not build the legacy systems that now need protecting. That is not a disadvantage. It is an open field.

Building without a blueprint

In 2019, Deepankar Rustagi, a founder operating in Lagos, Nigeria, looked at Africa’s informal retail sector and saw something most investors had decided wasn’t worth the complexity. Millions of small retailers, operating outside formal supply chains, unable to access credit, invisible to the financial system. Not a market failure. An unserved market.

OmniRetail, the B2B e-commerce and supply chain company Rustagi built, did not digitise existing retail infrastructure. There was none to digitise. It built the infrastructure from scratch, connecting over 150,000 retailers across Nigeria, Ghana, and Côte d’Ivoire to inventory, logistics, and financial services through a platform-based model that required no legacy system to negotiate with and no incumbent to displace.

By 2023, OmniRetail was EBITDA positive. By 2024, net profitable. In April 2025, it closed a $20 million Series A co-led by Norfund, the Norwegian Investment Fund, and Timon Capital, marking Norfund’s first direct equity investment in an African startup. Last week, Norway’s Deputy Minister of Foreign Affairs, Andreas Kravik, led a delegation to OmniRetail’s Lagos headquarters to meet the founders and investors behind it.

The delegation arrived after the building was done. That sequencing is the point.

Kola Aina, Founding Partner of Ventures Platform, the Lagos-based early-stage VC firm that backed OmniRetail from the beginning, has described Africa’s challenges as its greatest opportunities — and the founders solving them at scale as proof of what contextual, market-creating capital can unlock. Ventures Platform itself closed a $64 million first tranche of its second pan-African fund in November 2025, backed by the IFC, British International Investment, Standard Bank, and, for the first time, the Nigerian government through its Investment in Digital and Creative Enterprises program, iDICE — marking the first time Nigeria has invested directly in a VC fund. The institutional layer is forming around builders who moved without waiting for it.

The AI layer and why it is different

The argument so far could apply to any sector in any underserved market. What makes the AI moment specifically significant for Africa is a shift in the cost structure of building.

According to the Stanford University AI Index 2025, the cost of running AI at GPT-3.5 performance levels dropped 280-fold between November 2022 and October 2024. Hardware costs are declining 30% annually. What required a well-capitalised team three years ago can now be built by a small team with a modest budget, anywhere, on open-source models that are free to use.

This is what changes the calculus. The previous technological revolutions required physical infrastructure that took decades and billions to deploy. AI capability, at a level sufficient to build genuinely useful products, is now accessible to anyone willing to build with what exists.

Dacod Magagula understood this. A South African founder who grew up in under-resourced schools — the only student in a class of over 70 with access to a computer — Magagula co-founded FoondaMate in 2020 with Tao Boyle. The premise was direct: build an AI study assistant that works on WhatsApp and Facebook Messenger, the platforms African students already use, so that no download, no smartphone upgrade, and no reliable internet connection is required beyond what they already have.

FoondaMate is built on Meta’s open-source LLaMA models. It operates in 10 languages across more than 30 countries, including markets in Latin America and Southeast Asia. More than 3 million learners have used it. Students have submitted over 100 million questions through the platform. Among South African students who use FoondaMate, university eligibility rates are 30% higher than among peers who do not, according to the founders, as cited in Meta’s case study on the platform.

The infrastructure cost was near zero. What it required was the decision to build for a context no one else was building for.

That last sentence carries the weight of the entire argument. Alex Okosi, Managing Director of Google Africa, made the risk explicit at the Africa Soft Power Summit in Nairobi: Africa must not become merely a consumer of AI tools designed and owned elsewhere — if African languages and realities are not embedded in the models being built, the continent ends up depending on tools built without it. The risk is not missing adoption. It is adopting tools that don’t fit the context, then building an economy on top of them.

FoondaMate did not wait for someone else to build an AI tool for African students. It built the African-context layer itself, on open infrastructure, before the window closed.

What the queue was protecting

History tends to tell the story of technological revolutions as races. The countries that moved first won. The countries that moved late fell behind. That framing makes sense when the infrastructure required is physical: factories, railways, power grids. You could not skip that queue. You had to build it in sequence.

AI infrastructure is not physical in the same way. The models are open. The distribution channels exist. The cost of capability has collapsed. What remains scarce is not capital or compute. It is the willingness to build for contexts that the original model developers did not have in mind.

Africa did not miss the queue. The queue was protecting something Africa never had. That is not a consolation. It is a structural advantage, available to anyone willing to use it now, while the people who built the previous revolution are still arguing about how to regulate the next one.