Africa's Startup Funding Totals Describe Ten Deals, Not an Ecosystem

Executive Summary

A monthly total of African startup funding describes the largest rounds of that month, not the market a typical founder faces. In July, August and September 2026 the ten largest deals made up 86.94%, 98.35% and 85.52% of each month's disclosed funding, according to Nairametrics' monthly reports, and across the first half of 2026 they took 60% to 65% by value on two different providers' counts. I conclude that the total works as a scoreboard for those rounds. The evidence is not one-directional: concentration varies from month to month, and the one published half-year median rose. A founder or investor should read deal count, median round size and top-ten share beside every total.

Introduction

The conventional reading says African startup funding is recovering whenever the monthly total rises. That reading mistakes a sum for a distribution: a total records how much money moved, and nothing about how many companies it reached.

Take September 2026. Nairametrics reported US$260.3 million across 54 disclosed deals, up 88.21% on September 2025 and down 40.19% on August's US$435.2 million, with M&A excluded. Both movements were presented as news about the ecosystem. Neither says what a founder raising a first institutional round should expect, because US$222.6 million of the month, 85.52%, sat in what the report counts as its top ten (it lists eleven entries, two tied at tenth).

This article treats a funding total the way an operator treats a revenue line: as an average that hides its distribution until the distribution is shown. I write in the first person because the thesis is mine, and I rank the evidence so that it can be tested. Funding concentration means the share of a period's startup funding raised by its largest deals; the top-ten deal share is the portion of a period's total raised by its ten largest deals.

The figures come from Briter's Africa Venture Pulse report and from Nairametrics' monthly analyses, which for August name The Big Deal as the data provider. Trackers differ in scope, so I compare shares within a source and never mix totals across sources. The monthly figures here are Nairametrics' analysis, not the provider's own deal-level series.

How Does a Monthly Funding Total Get Built?

The mechanism is arithmetic. A total adds every disclosed round, so the largest-deals effect, one or two rounds setting the top-line figure, is built in. In August 2026 one round, Moove's US$250 million, was 57.4% of the month's US$435.2 million (a calculation from the two reported figures). Without it August was US$185.2 million. September's US$260.3 million is then about 40.5% above August, not 40.19% below it (a calculation on the same basis). The direction of the monthly comparison reverses when one round is removed.

The same total also blends instruments. The July report, citing Africa: The Big Deal, put debt at US$75 million of about US$102 million, or 74%, and equity at US$25 million, the lowest monthly equity total recorded since April 2019. In September, debt was US$112.1 million, 43.07% of the total. A rise led by debt is a different statement from a rise in equity, and a total reports them as one number.

Large rounds also have identifiable sources. State-backed fund programmes are one of the sources of capital that can sit behind larger rounds, and Egypt's and Nigeria's programmes use public money as an anchor or first-loss limited partner, a design meant to pull private capital into funds. Neither programme has reported realised returns, so a rise in a total cannot yet be traced to it.

A total is a sum, and a sum is set by its largest terms.

How Concentrated Were July, August and September 2026?

Three consecutive months of Nairametrics reporting give the following picture. Shares are as reported; the deal counts are those each report states.

For comparison, the August report puts the top-ten share in August 2025 at 76.19%. The ten deals that set each of these months were a small part of the deal count: eleven of 54 in September is about 20% of deals carrying 85.52% of value (a calculation).

What the other deals averaged is a calculation, not a reported statistic. In September, US$260.3 million less US$222.6 million leaves US$37.7 million across 43 deals, about US$0.88 million each. In August the remainder is US$7.2 million across 20 to 22 deals, about US$0.33 million to US$0.36 million each. In July, US$13.35 million across 37 deals is about US$0.36 million each. These are averages that a single larger deal can move; they are not medians, and I use them only because the reports publish no median.

Two figures near 98% appear in these reports, and they measure different things. August's top-ten share was 98.35%, a share of one month's total. September's top-ten money was up about 98.40% on the US$112.2 million of September 2025, a twelve-month change in what the ten largest deals summed to. Neither describes how far ten deals can move a total inside a month, and I do not use that framing.

The reports also disagree with themselves in places, and the disagreement should be visible. The September report lists eleven entries in its top ten, two tied at tenth, and the eleven listed amounts sum to US$222.6 million, so the 85.52% share rests on eleven deals. For July, the August report gives 47 deals, while the July report counts 44 rounds of US$100,000 or more. I use the first because it is the report that states a top-ten share.

In each of the three months, the ten largest deals carried at least 85% of the reported value, so the total described those deals and little else.

Where Does the Data Not Support the Thesis?

A contrarian case that hides its weak points is an advertisement. Four findings cut against the strong form of my argument.

Is the concentration constant?

No. The top ten took 60.09% of H1 2026 funding in Nairametrics' half-year count, up from 47.95% in H1 2025, and 65% in Briter's report, up from 48%. Two providers agree on direction and differ by five points on level. "Ten deals" is accurate for a month and approximate for a half-year, where the top ten were roughly three-fifths to two-thirds of value, not all of it. Nairametrics' own report states the share as 60.09% in one section and 61.69% in another; the US$948 million top-ten figure supports 60.09%.

Is part of the monthly share a sample-size effect?

Yes. Ten deals are a larger part of a month with 47 to 54 deals than of a half-year with 205 (Briter) or 241 (Nairametrics) deals: about 20% of September's count (eleven of 54) against roughly 4% to 5% (ten of 205 or 241) (calculations). A smaller denominator pushes the monthly share up on its own. The monthly figures are therefore best read against each other, and the half-year figures against the half-year.

Did the typical round shrink?

On the only published median, no. Briter reports that the median deal size rose from US$500,000 in H1 2025 to US$1.7 million in H1 2026, which it links to a larger number of large deals. A median is hard for a few big rounds to move, so this is the strongest evidence against reading concentration as a collapse for everyone else. The caveats are that it covers disclosed deals only and includes debt, and that it cannot be separated by stage. My ex-top-ten averages above, from a different provider, sit well below US$1.7 million. I cannot reconcile the two from published material, and that gap is itself the reason to ask for medians by stage.

Do the providers agree on the total?

No. Briter counts US$3.3 billion across 205 disclosed deals in H1 2026; Nairametrics counts US$1.58 billion across 241 startups. My state-of-the-market baseline records trackers whose totals for the same period differ by more than double, which is why I compare within a source only.

The thesis survives as a claim about measurement and weakens as a claim about founders' experience, where the only published median moved the other way.

What Should Replace the Total?

The replacement is a set of four readings printed beside the total, so that the total is never read alone.

• Disclosed deal count. The number of rounds a provider reports in a period. In H1 2026 Nairametrics counted 241 deals against 301 in H1 2025, a fall of about 20% (a calculation), while its total rose from US$1.47 billion to US$1.58 billion, about 7.5% (a calculation).
• Median round size. The middle round in a period, which a few large deals do not move. Briter's H1 2026 figure is US$1.7 million.
• Top-ten share. How much of the total rests on ten deals. The three months above ran from 85.52% to 98.35%.
• Rest-of-market average. The total less the top ten, divided by the remaining deals. It is a derived stopgap, labelled as a calculation wherever used, for months in which no median is published.

A total belongs on the page only after these have been printed beside it.

Where Does This Argument Stop?

This is an argument about a measure. It does not say how a founder raises capital, and it does not argue against any fund. What a typical early round looks like is covered in the pre-seed piece, which reports pre-seed rounds of US$100,000 to US$500,000 per round, citing BusinessDay (12 January 2026), and states its own average of about US$165,000 across 281 startups as a blend that is not a typical round. Why venture funds need very large outcomes is the subject of the fund-return mathematics in the Deal Flow piece of 22 October, and I do not repeat it here.

The exit side shows a related habit of measuring by the largest outcomes. AVCA's Nairobi Summit argued that funds should model the exits that actually occur rather than IPOs and unicorn valuations. Country concentration is a further layer: Nigeria raised US$364.1 million, 83.66% of August, and US$11.3 million, 4.34% of September, while Egypt supplied four of September's listed top ten and 28.66% of the month across nine deals. A reader who wants to know how any of these sources fit into a founder's financing path should start with the capital map this piece argues the total misdescribes.

Why This Matters for Founders and Capital Allocators

For a seed-stage founder, a rising total is not evidence that more capital is available at their stage. The decision-relevant numbers are the median, the count and the rest-of-market average, because those describe the rounds a company like theirs can raise. A founder who benchmarks against the top-line total will misjudge how much institutional capital exists for a first round.

For investors and limited partners, the total supports the ecosystem story and the concentration supports the portfolio question: which part of the reported capital is available to new managers and new companies, and which is committed to a few large borrowers and issuers. Asking a provider for count and median alongside any total costs nothing and changes the reading. For ecosystem bodies and governments that cite totals, the practical change is to print the top-ten share beside the figure.

What Would Change My Mind?

A thesis is only useful if it can lose. I would revise this one if any of the following appeared in a named provider's data. The thresholds are my own tests, not findings.

• The top-ten share falls below 50% for three consecutive months while the disclosed deal count is stable or rising.
• A provider publishes medians by stage and the seed-stage median rises across several months while the top-line total is dominated by a few rounds.
• The rest-of-market average rises across three consecutive months on a stable deal count.

Any of these would show that capital is reaching the broad base of companies even as a few large rounds dominate the total, which is the claim I am arguing against.

Key Takeaways

• A monthly funding total is a statement about the largest rounds in the month: in August 2026 one US$250 million round was 57.4% of it (a calculation), and removing it reverses the direction of the September comparison.
• Concentration is high but not constant: the top ten were 85.52% to 98.35% of the three most recent monthly totals and 60% to 65% of the first half, so "ten deals" holds for months and approximates for halves.
• The one published median moved against the strong form of the thesis, rising from US$500,000 to US$1.7 million between H1 2025 and H1 2026 according to Briter, so the case rests on measurement, not on a collapse for typical founders.
• Providers disagree on level, with H1 2026 totals of US$3.3 billion and US$1.58 billion, so shares should be compared within a source and never across sources.
• The replacement for the total is four readings printed beside it: disclosed deal count, median round size, top-ten share and the rest-of-market average.

Conclusion

A total that ten rounds can set is a scoreboard for those rounds. The three most recent months show it plainly, the half-year data softens it, and the median that Briter publishes limits how far it can be pushed. What follows is practical: print the count, the median and the top-ten share next to every total, and read the total last.