The hardware is funded. The brain isn't.

For a decade, electrifying rural Africa had a certain flow to it. Win the grant, buy the hardware, pour the concrete, cut the ribbon — and then operate the site with whatever was at hand: a Victron tab here, a SparkMeter tab there, fuel logs in a spreadsheet updated on Fridays, faults arriving as WhatsApp voice notes. It worked, more or less, because nobody ran more than a handful of sites, and nobody was checking too closely.

That order is ending. The $750M DARES program — the largest distributed energy deployment in World Bank history — is building up to 1,350 mini-grids for 17.5 million Nigerians, and it pays differently: 20% of every grant, $24,000 per site, is held back for twelve months and released only against verified operational performance. The moment a site is commissioned, a clock starts. The spreadsheet stack that survived three sites does not survive thirty sites and an auditor.

Here is the part you can't see from a desk in San Francisco: the operating software this program structurally requires did not exist when the program was designed. It still doesn't. The insight lives at the intersection of three facts that no one has synthesized into a product — the compliance architecture of a performance-based grant, the actual tool stack Nigerian operators run today, and the state of edge AI, where a $50 computer at the site can now run the predictive models that three years ago needed cloud infrastructure rural Nigeria can't reliably reach.

And the window is locked open. The Electricity Act 2023 is constitutional-level law with 80%+ of states aligned; NERC issued 85 mini-grid licenses in eighteen months — more than the years before combined; the first DARES grant agreements are signed and Lotus Bank has committed ₦100B in debt financing. Debt is the tell: equity tolerates uncertainty, lenders cannot. The sector's capital is shifting to money that requires auditable performance data as a condition of deployment. Someone has to generate it.