The period between early 2024 and mid-2026 has been the most unforgiving in the short history of African venture-backed entrepreneurship. More than two dozen companies that had, between them, raised in excess of $500m in equity and debt, have been wound up, placed into administration, or simply gone silent. The names include a TIME100 company (Gro Intelligence), a Y Combinator graduate (MarketForce), the recipient of Africa’s largest idea-stage pre-seed round (Edukoya), and a clean-cooking operation that had secured funding from a division of the Microsoft Climate Innovation Fund (Koko Networks).

The standard explanation offered for this wave of collapses is often the “funding winter” — the sharp contraction in risk capital that followed the end of near-zero interest rates in 2022. That explanation is accurate as a proximate cause: nearly every failure on the register was unable to secure a follow-on round. But it is insufficient as a diagnosis. A careful examination of the legal and financial circumstances of these closures reveals a set of eight structural vulnerabilities that were embedded in business models, capital structures and governance arrangements long before the liquidity cycle turned. The funding winter did not create these flaws. It exposed them.

1. The full-stack cost trap

Several of the largest failures by quantum of capital destroyed — Copia Kenya ($123m raised), GoLemon, and Twiga Foods — shared an operating architecture that proved lethal in the inflationary and currency-volatile conditions of 2023–2026. Each company attempted to own and operate the entire value chain: warehousing, vehicle fleets, dark kitchens, agent networks and, in the case of GoLemon, direct sourcing from farmers and manufacturers. The strategic logic was that vertical integration would create a competitive moat, enabling quality control, lower consumer prices and superior unit economics at scale.

The arithmetic that broke them is straightforward. A business with substantial fixed costs — such as warehouse leases, fleet maintenance, salaried engineering teams and, critically, diesel for generators and vehicles — must generate enough profit on each order, multiplied across a sufficient number of orders, to cover those overheads. In other words, the total contribution from all customer transactions has to exceed the company’s fixed operating costs.

In Nigeria between 2023 and 2026, for example, that equation stopped working. Inflation eroded household purchasing power, reducing the number of orders consumers placed. At the same time, fixed operating costs climbed sharply. The naira lost more than half its value against the dollar over the period, driving up the local-currency cost of imported fuel, spare parts and software licences. GoLemon, as instance, disclosed that it remained profitable on a per-order basis at an average basket size of ₦43,700 ($32), indicating that each transaction still generated a positive gross margin. Yet the company was unable to increase order volumes enough for those cumulative margins to outweigh its rapidly expanding fixed-cost base. The same economic dynamic, expressed in different sectoral forms, ultimately undermined FoodCourt’s cloud kitchens and Copia’s rural logistics network.

This was not a failure of execution in the conventional sense. It was a structural mismatch between a high-fixed-cost operating model and an economic environment characterised by rapid, unpredictable currency depreciation. A full-stack consumer business in such a market is, in effect, a short position on exchange-rate stability. The asset-light marketplaces that survived the same period — Chowdeck, Glovo — did so because their cost structures contained virtually no fixed physical infrastructure. The insight is not that vertical integration is inherently flawed, but that it is incompatible with the volatility profile of the markets in which it was attempted.

2. Under-capitalisation in multi-jurisdictional fintech infrastructure

The closures of Chimoney (Nigeria/Canada), Gigbanc (Nigeria), Thepeer (Nigeria) and the judicial liquidation of Bizao (Francophone Africa) illuminate a financing gap that was understood by operators but systematically ignored by investors. Each of these companies attempted to build payments infrastructure that operated across multiple regulatory jurisdictions. Each was capitalised with a seed round that, in light of the actual fixed costs of compliance, was structurally inadequate.

Chimoney’s founder, Uchi Uchibeke, provided the most candid post-mortem. “Under $1 million is too thin for a venture-scale fintech across multiple jurisdictions,” he stated, adding that he should have either raised “meaningfully more or bootstrapped properly with a profitable beachhead.” To operate a cross-border payments or open-banking API in even three African countries, a start-up must fund: licensing fees and legal costs (typically $50,000–$200,000 per jurisdiction upfront, plus annual renewal); a compliance team of two to three professionals per region; treasury operations maintaining float in multiple currencies and absorbing foreign exchange losses; and technology infrastructure capable of meeting the reporting standards of multiple central banks. A conservative estimate places the baseline fixed annual cost of such an operation at $500,000–$800,000 before a single transaction is processed or a single customer acquired.

A seed round of $1m–$2m provides, after the costs of the raise itself, perhaps 12–18 months of runway for compliance alone. There is no remaining capital for product development, distribution or the working capital needed for float. The business is mathematically incapable of reaching break-even scale. The rounds that were raised were, in effect, options on a future acquisition that would resolve the capital inadequacy before it became terminal. When the M&A pipeline narrowed — because acquirers also faced tighter funding — those options expired worthless.

The empirical lesson from the 2024–2026 cull is that a seed round below $5m for a fintech operating across more than one African jurisdiction is not a calculated risk. It is a slow liquidation.

3. The pivot as a value-destruction mechanism

Start-up culture venerates the pivot — the strategic change of direction that transforms a failing venture into a success. The African failure data from this period refutes that narrative. Of the 30 companies examined, at least five attempted major strategic shifts in their final phase of operation: Okra moved from open banking to a local cloud platform (Nebula); Kippa abandoned agency banking for an AI-powered edtech product; MarketForce had itself pivoted years earlier from SaaS to a B2B marketplace (RejaReja); MVX shifted from vessel chartering to freight booking and trade finance; and TradeHub attempted a cross-border marketplace before moving to a B2B software tool.

Only Egypt’s TradeHub ended with a return of capital to investors. Every other pivot resulted in a total loss or, in Okra’s case, a partial return of an estimated $4m–$5m from $16.5m raised. In each instance, the pivot was not a calculated strategic move based on a validated new insight. It was an attempt to deploy the last available capital to avoid the reputational and psychological cost of admitting failure. Kippa’s pivot to edtech was announced publicly; its new website subsequently went offline, and its co-founders took jobs in the United States. No capital was returned.

The structural reason pivots destroy value in a funding-scarce environment is a principal-agent problem. The founder bears the full reputational cost of failure but only a fraction of the financial loss. The rational course for a founder with residual cash and no validated path forward is to bet that cash on a new thesis rather than return it. The investors, who have marked the company up in their own portfolios, have their own incentive to avoid a write-down and consent to the gamble. The result is that capital that could have been returned to limited partners is consumed in a low-probability wager.

The sole counterexample is instructive. TradeHub, founded by a former Bosta executive, tried two different business models over 18 months. When neither gained traction, the founders concluded they no longer had “strong enough conviction in a third idea that justified the risk of continuing” and returned the remaining $1.4m. This decision was treated by the industry as a curiosity. It should be treated as a standard of fiduciary conduct.

4. The impact-capital liquidity mismatch

The collapses of Zydii (Kenyan SME upskilling), SolarNow (East African off-grid solar) and CityTaps (smart water meters in Kenya) share a financial structure that proved internally contradictory. Each was funded by investors who described their capital as “patient” — development finance institutions (DFIs), impact funds and foundations that signalled a willingness to accept longer holding periods and lower financial returns in exchange for measurable development outcomes. The companies, relying on these signals, structured their operations with extended runways to commercial viability.

When the global cost of capital rose and limited partners began re-evaluating risk across all alternative asset classes, the patience evaporated. The mechanism was not a change in institutional mandate but the legal form of the capital itself. A significant portion of these investments had been structured as convertible loan notes or equity rounds that required subsequent funding events to reach sustainability. Impact funds, despite their mission statements, are answerable to LPs who expect both impact metrics and financial returns within a defined fund life. When follow-on capital became scarce, fund managers faced a choice between conserving dry powder for portfolio companies nearest to liquidity or doubling down on loss-making ventures with uncertain paths to revenue. They chose the former.

The lesson is that DFIs and impact funds used instruments designed for high-growth venture capital to fund businesses that were essentially development projects. A genuinely patient capital structure for essential services sold to low-income populations would employ grants, revenue-based financing or equity with explicit long-dated milestones and no forced-liquidation triggers. The use of convertible notes and standard venture equity for these ventures was possibly a category error — one that became terminal when the capital cycle turned.

5. The missing governance layer

Of the 30 companies that failed, only TradeHub returned unspent capital to investors. In every other case, the outcome was a total loss or a deeply subordinated recovery for equity holders. This uniformity points to a governance deficit that is pervasive in African venture capital but rarely discussed.

Most of the failed start-ups had boards composed of founders and investor-nominated directors who represented funds that had marked the company’s valuation up in their own portfolios. When performance began to diverge from plan, each director faced a conflict of interest. Acknowledging that the business was no longer viable and returning residual capital would crystallise a loss — damaging the fund’s reported net asset value and the individual’s track record. Approving a pivot, a bridge round or an extension of the cash runway preserved the possibility, however remote, of a recovery. The rational decision for each individual, acting within the incentive structure of their own fund, was to defer the day of reckoning.

The result was that capital continued to be deployed long after the expected value of additional investment had turned negative. Gro Intelligence, which raised $125m, burned through its entire capital base without any governance mechanism that could force an independent review of the decision to continue. MarketForce ($42.9m), Copia ($123m) and Kippa ($14.3m) followed the same trajectory. In each case, the board lacked an independent director with the contractual authority to trigger a return-of-capital vote if agreed milestones were missed.

The solution is well established in developed venture markets: milestone-based tranching of investment rounds, independent board seats with defined fiduciary duties, and liquidation preferences that give investors the ability to force a solvent wind-up. That these provisions were largely absent from the deals that produced the 2024–2026 cull is not merely a founder-side governance failure. It may represent a failure of institutional due diligence by the investors who wrote the cheques.

6. Regulatory and policy single-point failures

Koko Networks, the Nairobi-based clean cooking company that raised more than $100m in equity and debt, deployed 3,000 ethanol fuel dispensers serving 1.5m households. Its core business — selling ethanol cooking fuel — operated at negative unit margins. The entire investment case rested on the assumption that the company would generate carbon credits from the fuel switch and sell those credits into international compliance markets, turning an operating loss into a net profit at the consolidated level.

In 2025, Kenya’s government declined to issue a Letter of Authorisation required for cross-border carbon credit transfers under Article 6 of the Paris Agreement, citing concerns about credit integrity and the share of economic benefit accruing to the state. That single regulatory decision severed Koko’s revenue line at a stroke. In February 2026, 700 staff were laid off. By April, administrators were marketing the company’s intellectual property and an Indian manufacturing plant to secured creditors.

Koko is the most dramatic instance of a broader pattern. Carbon credits are not a commodity produced by private effort alone; they are a synthetic asset whose existence, valuation and transferability depend entirely on government consent at the point of origin. Treating the revenue stream from such an asset as the foundation of unit economics — rather than as contingent upside — amounts to a concentrated bet on a single regulatory permission. The investors in Koko, including the Microsoft Climate Innovation Fund and the AfricaGoGreen Fund, applied sophisticated discount rates to technology and execution risk while effectively assigning a near-zero probability to sovereign regulatory risk. That risk assessment was demonstrably wrong.

The failure carries implications beyond the climate sector. Any business model in which more than 30 per cent of projected revenue depends on a single government licence, subsidy or regulatory authorisation — in a jurisdiction where that authorisation is subject to political discretion — is not a venture-scale business. It is a policy speculation.

7. The willingness-to-pay chasm

Edukoya’s closure, barely three years after it raised a record $3.5m pre-seed round, has become the emblematic failure of the period. The company was founded by a former McKinsey consultant, Google employee and Kuda Bank executive; its cap table included Target Global and the founders of Kuda and Paystack. It offered a polished digital tutoring and content platform aimed at K-12 students. Its post-mortem statement cited “insurmountable challenges in market readiness”.

Beneath that phrase lies a specific economic reality. Supplementary digital education is, in the household budget hierarchy of most African consumers, a discretionary item. When real incomes are squeezed — Nigeria’s inflation rate exceeded 30 per cent for much of the period, eroding purchasing power — the first expenditures to be cut are those perceived as non-essential enrichments. School fees are non-negotiable. A paid online tutoring subscription is not.

Edukoya’s failure is not an isolated phenomenon. Kenya’s Zydii, which provided digital upskilling courses for small and medium enterprises, found that African SMEs overwhelmingly do not have dedicated training budgets; the buyer was the individual employee, who was unwilling to pay. In each case, investor models had conflated stated intent — high percentages of parents or workers expressing interest in surveys — with revealed preference, the actual conversion to a paid subscription at a sustainable price point.

The failures demonstrate that societal need does not automatically translate into commercial demand. An educational crisis affecting hundreds of millions of people may present a compelling development challenge; it may not, without evidence of willingness to pay at scale, present an addressable market for venture capital.

8. The pedigree trap: when elite founders become a liability

Edukoya is also the clearest example of the final pattern, but it is not the only one. Okra was built by a team with deep engineering credentials and backed by TLcom Capital and Susa Ventures. Kippa’s founders had the track record and network to raise $14.3m. MarketForce was a Y Combinator graduate. Gro Intelligence was founded by a former Wall Street trader whose personal story was central to the company’s fundraising narrative.

In each case, the founder’s pedigree unlocked capital at a scale and speed that would not have been available to a less credentialed team. That capital enabled the rapid construction of expensive organisations — large engineering teams, multiple country offices, polished products — before the fundamental assumption of customer willingness to pay had been validated with hard revenue data. The burn rate became a function of the founder’s own professional standards, and the reputational cost of downsizing or admitting a flawed premise became prohibitively high.

The mechanism is a competence trap. Highly capable people build highly capable organisations. Those organisations cost more to run. When the market proves less ready than the pitch deck assumed, the fixed cost base set by the team’s own calibre leaves no room for a low-cost pivot or a patient, capital-efficient search for product-market fit. The venture consumes its funding and collapses.

The uncomfortable implication is that, in markets where genuine demand is unproven, founder quality — as measured by educational background, corporate experience and prior start-up exits — may be inversely correlated with capital efficiency. Investors who bet on the jockey rather than the horse, without first demanding evidence that the horse can run, were buying a lottery ticket with a negative expected value.

The Bottom Line

The companies reviewed in this failure set were not victims of a temporary capital drought from which the market will recover unchanged. They were, in large part, the product of a structural misallocation: the application of venture capital instruments and venture capital expectations to businesses whose economic characteristics — thin margins, regulatory dependency, long paths to revenue, customers with low ability to pay — are fundamentally incompatible with the venture model.

This does not mean that technology-enabled businesses in Africa are uninvestable. It means that different types of business require different types of capital, structured on different terms, with different governance and different expectations of return.

This table is not an exhaustive list of the failed startups analyzed for this publication. *FoodCourt: The company says it has only paused operations.