The constraint was never capital. It is intermediation.

The standard framing of African infrastructure — a $100bn-plus annual financing gap, a continent starved of capital — has become an obstacle to understanding. It describes a shortfall accurately and diagnoses its cause badly. Global institutional allocators hold more capital seeking infrastructure-like duration than at any point in modern history. Very little of it lands in Africa. The reason is not appetite; it is the absence of a manufacturing process that converts political intent into de-risked, standardised, investable assets.

Look at what actually happens between a project announcement and a financial close. Feasibility, environmental and social assessment, land assembly, tariff modelling, offtake negotiation, currency-mismatch resolution, guarantee structuring, lender due diligence. Each of those steps has an owner in a mature market. In most African jurisdictions, several have no owner at all. The result is a pipeline that looks enormous in registries such as AUDA-NEPAD's African Infrastructure Database — 1,800-plus projects across 55 countries — and thin in the World Bank PPI database, which records only transactions that actually reached financial close with private operating risk transferred.

That divergence between the registry and the closed-deal record is the single most informative statistic in the sector, and almost nobody publishes it. It is the conversion rate of African infrastructure. When we track it across the datasets available, the picture is consistent: the volume of announced projects has risen sharply since 2018 while the number reaching close has been broadly flat.

The institutions that understand this are behaving accordingly. Africa50's asset-recycling work, the Alliance for Green Infrastructure in Africa, PIDG's project-development arm and the AfDB's Sustainable Energy Fund for Africa are all, in different vocabularies, attempts to industrialise preparation rather than to add capital. They are chronically small relative to the task. Preparation funding is a rounding error against commitment volumes, which is precisely backwards.

There is a second-order effect worth naming. Because preparation is scarce, the projects that do get prepared are disproportionately large. A 400MW hydro scheme can absorb $30m of development cost; a 40MW one cannot. This is why the pipeline skews to megaprojects that take a decade, and why distributed assets — which would move access numbers faster — remain underrepresented in every dataset we track.

The policy conclusion is unglamorous. Fund preparation facilities at an order of magnitude above current levels, standardise contracts through instruments like the African Legal Support Facility's model agreements, and publish conversion data so the failure is visible. None of that requires new capital. It requires accepting that the bottleneck is administrative, and that administrative bottlenecks are cheaper to fix than capital ones.

A pipeline of 1,800 projects means nothing if fewer than 40 reach financial close in a year.