A Launch Base Africa analysis of funding data across 2025 and 2026 shows disclosed investment into African artificial intelligence companies has fallen by more than 90 per cent, even as billions of dollars of debt and infrastructure capital pour into electric mobility, clean energy and fintech. The divergence threatens to hollow out the continent’s early‑stage software pipeline and risks a brain drain of the technical talent that the 2025 funding cycle briefly promised to support. The retreat from African AI coincides with an unprecedented boom in US artificial intelligence investment, where startups have absorbed over $220bn in the first two months of 2026 alone, leaving African ventures almost invisible to global tech investors.
In 2025, African AI startups raised at least $170m across more than two dozen rounds. Investors from US venture funds to European development banks and local African fund managers backed everything from Arabic large language models and serverless AI infrastructure to health diagnostics and agritech platforms. By the first half of 2026, the picture has changed dramatically. Disclosed AI funding has collapsed to approximately $16m, and the number of rounds has roughly halved. The investor base has thinned to a handful of local angel networks, impact‑focused vehicles and a few small specialist funds. No large US venture capital firm has led an African AI round so far this year.
At the same time, capital is flooding into hardware‑heavy sectors. Spiro, the electric motorcycle battery‑swapping operator, has raised more than $415m since 2023, including a $215m equity round announced earlier this month. Kenya’s Cold Solutions raised $19m for temperature‑controlled logistics. Fintech lenders and remittance platforms are tapping structured credit and stablecoin‑backed facilities at scale. Africa‑focused debt and infrastructure investment has never been busier, but the capital is bypassing artificial intelligence almost entirely.
The AI funding retreat, in numbers
A review of publicly disclosed deals tracked by Launch Base Africa shows that at least 28 AI‑focused African startups raised money in 2025. The rounds included Intella’s $12.5m Series A, KERA Health’s $10m seed, Thunders’ $9m seed and Cerebrium’s $8.5m seed, alongside a host of smaller pre‑seed and seed deals. Investor participation was broad: Y Combinator, Gradient Ventures (Google’s AI fund), 500 Global, Bowery Capital and other US firms featured alongside European DFIs such as Bpifrance and Digital Africa, Japanese and Middle Eastern funds, and a wide range of Africa‑based venture firms from Atlantica Ventures and LoftyInc Capital to Enza Capital and ANZA Capital.
By the first half of 2026, the AI funding map has shrunk. Only around 14 rounds have been disclosed, with total committed capital of roughly $16m. Even when adjusting for the shorter reporting period, the annualised run‑rate points to a decline of more than 90 per cent from the 2025 total. The drop is not simply a product of a few large 2025 rounds falling out of the comparison; the number of active investors has contracted sharply, and the types of AI being funded have narrowed to niche, applied verticals such as medical diagnostics, legal tools and logistics optimisation.
Who has retreated — and who remains
The retreat is broad‑based but uneven. US venture capital, which provided the largest growth‑stage cheques and lent global credibility to the African AI story, has almost entirely disappeared from disclosed deals. None of the major US firms that backed African AI in 2025 — Y Combinator, Gradient Ventures, 500 Global, Bowery Capital, Authentic Ventures — have participated in a new African AI round in 2026. The only US‑affiliated capital flowing into the sector comes from philanthropic and impact programmes, such as the Madica fund backed by Flourish Ventures, the Morgan Stanley Inclusive Ventures Lab, and the Google Black Founders Fund.
European development finance institutions, which used equity and grant programmes to seed AI startups in 2025 through instruments such as Digital Africa’s Fuzé fund and Bpifrance’s direct investments, have sharply reduced their AI exposure. Their focus has shifted to infrastructure debt and climate‑linked financing, leaving only occasional grant‑level support for early‑stage software companies.
Local African fund managers, too, have pulled back. In 2025, the likes of Atlantica Ventures, LoftyInc Capital, P1 Ventures, Enza Capital, and ANZA Capital all made AI bets. None has yet disclosed a new AI investment in 2026. The local investors still writing AI cheques are, for the most part, smaller and more specialised: Egypt’s A15 and M Empire angels, South Africa’s 3 Capital Ventures, and the pan‑African 54Collective. Angel syndicates and family offices have filled a few pre‑seed gaps, but their capacity is limited.
For all the rhetoric surrounding AI, relatively little capital is currently flowing into African AI startups. Many of the international funds that helped drive the sector’s early momentum have disappeared from recent deal activity, while local investors have shifted into defensive mode, preserving capital and doubling down on existing fintech bets.
Why the capital is flowing elsewhere
Two powerful forces are pulling capital away from African AI. The first is the gravitational pull of the US artificial intelligence market, which has entered a period of frenzied investment. In the first two months of 2026 alone, US AI startups secured a record $220bn in funding, driven by mega‑rounds for frontier models and enterprise infrastructure. Anthropic closed a $65bn Series H that valued it at $965bn, the highest for any private AI company. Elon Musk’s xAI raised $20bn. ElevenLabs tripled its valuation to $11bn with a $500m Series D. Roughly 68 per cent of this capital has concentrated on enterprise AI infrastructure and safety‑focused frontier models, not consumer applications. In this environment, a $5m seed round in Nairobi or Cairo is barely a rounding error. US venture firms that dabbled in African AI in 2025 have refocused entirely on their home market, where opportunities are vastly larger, returns can be generated faster and the technical ecosystem is deeper. There is no strategic imperative to look abroad.
The second force is the simultaneous restructuring of development finance away from early‑stage equity and towards infrastructure debt. DFIs such as the IFC, British International Investment, the Netherlands’ FMO and France’s Proparco have spent this year so far reallocating capital from venture‑style equity experiments back to long‑dated, non‑dilutive instruments. The reason is straightforward: debt secured against physical assets such as electric buses, solar plants or cold‑chain depots offers predictable cash flows and measurable climate impact, while equity in pre‑revenue AI companies does not. The result is that even the institutions that might have plugged the gap left by retreating US venture firms are not doing so. They are instead writing $50m cheques to e‑mobility companies and $94m to renewable energy platforms.
The gap is not being filled by local equity. Many of the African funds that were active in AI in 2025 are now fully deployed or in the late stages of their investment periods, with successor vehicles struggling to close in a difficult fundraising environment. The result is a structural void at the early‑stage equity level, precisely where AI startups need capital to move from prototype to product.
The consequences for African AI
The immediate effect is a thinning of the pipeline. African AI startups that raised seed rounds in 2024 or 2025 are now looking for follow‑on capital with fewer options available. The ambition to build foundational models tailored to African languages and contexts — a theme that generated headlines when Widebot raised $3m in late 2025 — has not been matched by continued investment. No new African large language model company has closed a round of similar scale in 2026, and Widebot itself has not announced a follow‑on.
Instead, the AI that is getting funded is narrow and applied: a legal‑tech tool in Kenya (Hakimu, $200k), a restaurant management platform in South Africa (Certus AI, undisclosed), a medical diagnostics company (AI Diagnostics, $4.75m). These are sensible businesses, but they are not the kind of platform bets that build continental AI capacity.
Talent is another concern. The 2025 funding cycle allowed several African AI startups to hire machine‑learning engineers and data scientists at globally competitive salaries. As funding dries up, those roles are at risk. Without a local employment base, skilled AI practitioners are likely to seek remote work for US or European firms, accelerating the brain drain that African tech has long struggled to contain. The fact that US companies are raising hundreds of billions of dollars to hire AI talent only intensifies the magnet.
The bigger picture
The divergence between hardware‑heavy, debt‑funded sectors and starved software innovation is not unique to Africa, but it is particularly acute there because of the continent’s heavy reliance on DFI capital and the thinness of its local venture industry. What is happening in 2026 is less a story of AI being singled out than a reflection of a broader shift in investor appetite towards infrastructure and structured finance — a shift that is rational for investors but carries long‑term consequences for the startup ecosystem.
“If you can get a 6–8 per cent return from senior secured debt in an electric bus fleet, why would you take an equity risk on a pre‑revenue AI startup?” said a Johannesburg‑based fund manager, who asked not to be named. “The market is telling founders to go build infrastructure, not code. And the Americans are telling them, ‘we have all the AI we need at home’.”
Whether that signal is temporary or permanent will depend on whether Africa’s local venture funds can return to fundraising, and whether international AI investors ever decide the continent is a strategic, not just a philanthropic, opportunity. For now, the data is clear: African AI is being left behind, caught between a US investment boom that ignores it and a DFI machinery that has switched from software to steel.