What exists today
Sovereign loansIDA, AfDB
Project preparationGIF, ESMAP
GuaranteesMIGA
CoordinationMission 300
Connection subsidiesGPRBA, UEF
01
The broken systemAcross sub-Saharan Africa, the power system is trapped.
From Uganda, Zambia, Tanzania, and Mozambique through the wider Mission 300 reform pipeline, the pattern repeats in every electrification effort. The problem is not just how to build a power line. It is who can finance it for decades while making the revenue system believable.
The loop needs an external entry point.An outside balance sheet can finance the asset. An enforceable revenue covenant can keep the cash cycle intact. One without the other is not enough.
The unfilled function
Most grids belong to everyone, so nobody is fully accountable for them. A single owner can borrow against the wires, keep them repaired over decades, and hand them back to the host country at a fixed date and price.
Announced projects and finished projects are not the same thing. When the institution carries completion risk, it pays out of its own pocket when construction runs late or over budget, which makes the original bid a promise the contractors have to keep.
A power line repays its cost over decades, not months. If nobody absorbs the lean early years of the revenue plan, the money for the wires never arrives. An institution that stays in for the long term lets the loan last as long as the asset.
The ribbon-cutting is the easy part. Most access programs fail quietly after commissioning: scheduled tariff adjustments get delayed, collections enforcement softens, protected funds get redirected. Enforcement means the rules stay in force when it is politically expensive to enforce them.
This is the institution's entire reason to exist.
Why these three questions decide everything
The questions are not rhetoric. Studied electrification history answers the first two, and the third is the point where access programs usually fail without anyone declaring it.
Who finances the wires.
The historical breakthroughs were not financed by private markets. The United States reached nearly universal rural electrification because the Rural Electrification Administration, created in 1935, lent public money for decades to the cooperatives private utilities had refused to serve. South Korea reached near-universal access by the 1980s through state-planned expansion under a single statutory utility. First-time grids have been built on long-dated public capital, or not at all.
Who guarantees payment.
A grid repays its cost over twenty or thirty years. No private lender signs on if tariffs can be frozen and bill collection can collapse in year six, whatever the interest rate, because the expected return goes negative. A guarantee from an institution that cannot be voted out or quietly raided is what makes decades-long private co-investment rational in the first place.
Who enforces the deal after the ribbon cutting.
Most access programs fail quietly after commissioning: scheduled adjustments get delayed, collection enforcement softens, protected funds are redirected to other spending. The asset then decays within a few years and the program is never declared dead. Nobody failed loudly, so nothing got fixed. Someone has to hold authority once the cameras leave and the politics change.
02
The designThe operating machine.
The Facility would own transmission and distribution assets, raise first-loss equity from donors and concessional capital, then borrow as a AAA-rated institution to build the wires.
NOT:No power purchase agreementsNo utility management contractsNo grid software
PHASE 01 OF 05
The revenue covenant secures the deal.
Before concrete is poured, the Facility locks cash flow: a ring-fenced escrow, automatic tariff adjustment within agreed bands, performance-based tranches, a pre-funded lifeline for poor households, and public verification of flows.
PrecedentEscrow in project finance keeps bondholders paid even when utilities face pressure.
Design moveFailure to execute a scheduled adjustment exactly as written releases the next capital tranche only after published verification.
PHASE 02 OF 05
Assets enter as public contracts.
Each asset company (country and asset class) is formed by statute. Its contract and baseline are published before construction: collections, losses, outages, and customer counts on the record.
PrecedentOrissa, India: contracts signed against unverified baselines collapsed; the lesson became mandatory verified baselines.
Design move"No baseline, no build" is a statutory rule, not a preference.
PHASE 03 OF 05
Money flows in a fixed order.
Capex from Facility reserves. Grants buy tariff reductions only through the lifeline mechanism. Bonds are issued by the Facility, on its own balance sheet.
PrecedentIDA's $1-in-every-$3 leverage model for multilateral lending.
Design moveThe Facility's AAA rating is built by donor equity and asset cash flows, not by sovereign pledges.
PHASE 04 OF 05
Construction is contracted, not managed alone.
Engineering, procurement, and construction are opened to competitive bidding with fixed-price discipline, then a five-year operations contract with loss-reduction targets.
PrecedentDelhi cut AT&C losses from above 55% to under 10% using loss-reduction bidding.
Design moveContractors share upside for beating loss-reduction targets.
PHASE 05 OF 05
Accounts are public, line by line.
Every asset company's collections, payments, and flows through the covenant are audited by an independent firm and published machine-readably. Integrity unit reports to the board.
PrecedentGlobal Fund-style independent evaluation and fiduciary safeguards after its 2011 crisis.
Design moveSunlight is priced into contracts: verified transparency is part of the bond covenant.
The capital stack
First-loss equity (25%)
Donors and development finance, absorbing losses first. [IDA leverage model]
AAA bond issuance (50%)
Capital markets at low cost, backed by donor equity and asset cash flows.
Concessional capital (25%)
Low-cost funding for tariff lifelines and poor-household protection.
The revenue covenant in detail
Customer revenue flows into a dedicated account. Operations, debt service, and operator residuals are paid by an automatic waterfall before any political pressure can redirect the money.
Customer revenue flows into a dedicated account. Operations, debt service, and operator residuals are paid by an automatic waterfall before any political pressure can redirect the money.
Tariffs adjust monthly by a published formula tied to fuel costs, the exchange rate, and inflation. No cabinet must sign every change. Adjustment bands are set at the covenant's start and can only be renegotiated in open session.
Capital is released in tranches only after published verification of prior targets: loss reduction, collection rates, and outages. Performance gates replace discretionary disbursement.
A subsidized lifeline tariff for the poorest households is funded before capital is deployed, not cross-subsidised by the utility. The operator's books never need to carry the cost of social protection.
All flows through the covenant are audited by an independent firm and published machine-readably. Board, bondholders, and citizens see the same numbers.
03
The targetsThe countries where the wires come first.
The Facility hunts where the circuit is broken: low consumption, high losses, stranded generation, and a government willing to fix collections.
Wave 01 · Pilot
Uganda
95 kWh per person. Ugandans use less electricity than a typical refrigerator.
2,048 MW of generation installed against roughly 985 MW peak demand: the power exists. What fails is the last mile. Electricity access at 55.3% of the population (2023). The grid cannot carry or bill the power it has.
- Distribution turnaround plus revenue covenant.
- Metering and collections reform front-loads the loop entry.
- Only template proven here scales.
Why Uganda first: the failure is pure distribution economics. Fix the metering and the collections, prove the revenue covenant on a small grid, and the template transfers. The precedent happened here: Umeme cut losses from 38% to 21% and lifted collections from 80% to 99.1% under a regulated 20%-per-year return on capex, and stopped needing subsidies. The Facility's Uganda template is that turnaround, institutionalized.
Wave 01 · Pilot
Zambia
726.7 kWh per person, yet the economy is grid-fragile.
Copper smelters anchor the load, but the 2024 El Nino drought forced eight to twelve hours of daily load shedding. Zambia's grid is an anchor-load economy that cannot guarantee its anchors.
- Anchor-load contracting plus SAPP regional wheeling.
- Smelter deals prove the covenant can constrain industrial demand too.
Why Zambia second: proving anchor-load discipline alongside the pilots makes the model credible for resource economies too.
Wave 02 · Scale
Tanzania
135 kWh per person, the deepest deficit on this list.
The 2,115 MW Julius Nyerere project is expected online, and a surplus of roughly 2.4 GW sits above a 2.27 GW peak. T&D losses hover near 14.5%. Tanzania is the wire-starved case: generation ahead of the grid.
- Transmission asset SPV with verified baselines.
- Prove T&D economics at sub-200 kWh.
Why Tanzania third: the Nyerere surplus needs wires first. If the Facility can write a bankable transmission covenant at 135 kWh per person, it can work anywhere.
Wave 02 · Scale
Mozambique
390.7 kWh per person, with Cahora Bassa flowing to South Africa.
2,052 MW installed, but capacity is dominated by the Cahora Bassa plant, whose output flows mostly to South Africa while Mozambicans sit at 36.9% access. A regional trade case with a domestic binding constraint.
- Domestic anchor wiring plus regional trade.
- Covenant terms that let exports continue while rewiring domestic service.
Why Mozambique last: the regional trade linkage makes the cash flow work. Domestic customers get wired, SAPP interconnectors carry the margin, and both improve.
One map, three layers.
The pilots prove the model. The reform pipeline is the addressable market. The AI demand collision is the reason for speed. Toggle the layers, then pick any country below to read its cited facts and the Facility template that fits it.
Drag to explore. Scroll or pinch to zoom. Colors: red, pilots. Grey, reform pipeline. Amber, AI demand collision.
Pilot countries · 4
Reform pipeline: Mission 300 Energy Compacts · 29
AI demand collision · 4
Pilot country · Layer 1
Uganda
- 95 kWh/person, 55.3% electricity access (gap data).
- 2,048 MW installed against roughly 985 MW peak demand: a stranded surplus bigger than the country's demand.
Facility template Distribution turnaround plus revenue covenant: meter hard, collect first, build second.
Open Uganda's pilot plan →04
The price of admission to the Age of AIThe deal a president can defend
A bridge, not a sale.
Ribbon-cutting on rails
Reliable power is the most visible win in politics. The Facility turns that ribbon-cutting into a durable asset and a defensible fiscal position: power that stays on after the election, without the utility bleeding the treasury.
A solvent utility
The minister gets audited baselines, loss-reduction contracts, and the fiscal space to protect the lifeline. The minister stops managing a liability and starts delivering a service.
Power that is affordable and real
The poorest household pays the full cost of bad power today: candles, batteries, spoiled food. A pre-funded lifeline makes the first kWh affordable, and the revenue covenant keeps the lights on.
Anchor contracts that hold
The industrial anchor gets bankable availability where it matters. Capped obligation: never forced to buy at a loss, always asked to pay on time. The deal works for the smelter, the server farm, and the factory.
A billion that multiplies
The donor gets leverage: roughly three dollars of spending authority per dollar of equity, a replenishment cycle, and evidence the covenant works before scaling. From press releases to power lines.
05
The launchStart narrow. Then scale.
Two credible designs. One recommendation. Begin with the smallest unit that can prove the revenue covenant.
Option A: Greenfield institution
A new chartered Facility, modeled on the Global Fund governance template, with its own balance sheet and bond program.
- Cleanest mission
- Longest launch timeline
- Largest political lift
Option B: Window with a sunset
An independently governed window inside an existing AAA institution, with step-in rights, a statutory sunset to full independence, and its own covenant.
- Faster to launch
- Borrowed AAA rating
- Risk: mission capture
What it takes to launch
The recommended sequence begins with Option B, migrating to Option A when two conditions hold.
Close one corridorOne asset company with public contract terms, built to plan.
Execute one adjustmentOne formula tariff adjustment executed exactly as written.
Audit one loopOne full cash cycle through a published covenant audit.
What is not yet proven
We have now assembled the closest-to-truth answer from 35 operator cases: below 1,000 kWh per person, concessions work only when subsidy, tariff, and losses are engineered together (see "What concessions below 1,000 kWh actually cost"). What remains unproven is field validation in the pilot countries.
Where the data is thin
Loss and collection figures vary by source and vintage. The Facility design requires verified baselines before any contract is signed: no baseline, no build.
Feasibility, with real numbers
What concessions below 1,000 kWh actually cost
35 operators and programs have tried this. Here’s what their record shows.
The evidence base
Thirty-five operator and program cases (national utilities, distribution concessions, rural concessions, mini-grids) drawn from 28 distinct source documents, mostly World Bank, spanning roughly 1960 to 2025 across Africa and Asia. The headline result: 19 of 35 were not viable. Below 1,000 kWh per person, concessions work only when subsidy, tariff, and losses are engineered together. Caveat: figures are as reported in their sources, not inflation-adjusted, and country counts below 1,000 kWh per person range from 43 to 77 depending on source and year.
The unit-economics identity
The feasibility phase must fill in this identity for every concession area, with measured numbers:
Cash revenueconnected customers x kWh per customer per month x realized tariff (US$/kWh) x collection rate
Cash cost(kWh billed divided by (1 minus losses)) x energy purchase or generation cost + fixed O&M + debt service
Viability gapcash cost minus cash revenue, closed by capital subsidy per connection, operating subsidy or availability payment, tariff indexation, or load growth
The fatal triad
A tariff-cost wedge, losses above 30%, and collection below 75% together are fatal. No observed case survives all three, and the record names the compounding cases:
Chad: 40% losses, ~50% collectionGuinea: 30% losses, 42% collectionSierra Leone: losses above 45%Gaza: 41% noncollection
Observed anchors
3 to 30 kWh/monthNewly connected rural household consumption (Madagascar 3-15, Uganda KRECS 30).
55 kWh/person/yearEastern Africa household end-use average.
US$411 per connectionGlobal mini-grid funding average.
~US$286 per connectionSubsidy needed for Senegal's bankable rural concession.
US$10 monthly billApproximate affordability ceiling for unelectrified households.
400-500 subscribers per siteThe minimum Madagascar's operators needed, plus 75% capital subsidy on generation and 100% on distribution, to be profitable.
~20x electricity salesWhat Uganda's KRECS needed to break even, on 30 kWh/month households at US$0.14/kWh.
Orientation figures, not underwriting assumptions.
What worked, and what it tells the Facility
Operational turnaround
Umeme (Uganda)
Cut losses from 38% to 21%, lifted collections from 80% to 99.1%, with a guaranteed 20%-per-year return on capex, and stopped needing subsidies. This is the operational turnaround the Facility's Uganda template institutionalizes, and 20%/yr on capex is the observed price of private capital here.
Insulated remuneration
CIE (Cote d'Ivoire)
Profitable through the concession at ~5% net margin, with losses down from 29% to 15%, because operator remuneration was ring-fenced from broader sector risk. This validates the covenant's insulated-remuneration design.
Capital structure
COMASEL (Senegal)
12.22% FIRR with ~US$286 subsidy per connection and ~60% private financing; 21,800 connections bid against an 8,500 minimum. This validates the Facility's capital structure.
Affordability reformulated
Cambodia mini-grids
Unsubsidized and spontaneous at US$0.40 to 1.25/kWh. Affordability is about the monthly bill, not the per-kWh price, which is why lifeline obligations must be pre-funded, not buried in the utility.
What failed
JIRAMA (Madagascar). Suppressed tariff ~US$0.13 vs generation cost near US$0.30/kWh.
EDG (Guinea). US 10c tariff vs US 27c cost, 42% collection, 30% losses; subsidies of US$317-448m/yr (1.2-2.1% of GDP).
SNE (Chad). 40% losses, ~50% collection; subsidy near US$1,000 per customer per year.
KRECS (Uganda). 30 kWh/month households at US$0.14/kWh needed ~20x sales to break even despite 100% collection.
The go/no-go gates
- Gate 1, demand.
Verified addressable base of at least 400-500 paying connections per service area and measured willingness to pay at a bill at or below US$10/month, with anchor productive loads identified. Fail means standalone solar territory: stop.
- Gate 2, price vs cost.
Realized tariff times collection must cover energy cost grossed up for losses on a 5-year view, or a committed, indexed subsidy or availability payment closes the gap in the contract. Fail without a funded gap means no-go.
- Gate 3, capital structure.
Capital subsidy per connection sized against the observed US$286-411 range (up to 75-100% of capex where loads are thinnest), disbursement-credible, with renewal capex funded, not just the first investment.
- Gate 4, risk allocation.
Operator remuneration insulated from sector cashflow (the CIE model) or a regulated return on capex (the Umeme model); FX, fuel, and government-payment risk explicitly assigned; downside debt-service coverage at or above 1 with equity IRR above the local discount rate.
Believe now vs measure in-field
Defensible now
- The fatal triad.
- Household-only rural demand of 3-30 kWh/month cannot fund a full-cost concession, and capital subsidy of roughly 40-100% of capex is the observed bankable range.
- Operational contracts with regulated returns and ring-fenced remuneration are the repeatable success pattern.
Only field work can answer
- Actual concession-area demand and ramp-up: national per-capita figures are the wrong denominator.
- Real willingness to pay and collection under prepaid metering in the target area.
- The achievable loss trajectory given the existing network.
- The credibility of the specific government as subsidy payer.
06
The charter homeCanadian-chartered, internationally governed.
The fastest route to a working institution is a national charter and an international board: headquarters in Toronto, Global Fund-style governance, and UN agencies as partners and co-financiers, not owners.
| Dimension | UN-led | Canadian-chartered |
|---|---|---|
| Decision speed | Consensus rules carry real gridlock risk; the Green Climate Fund's experience is the warning. | Supermajority voting on a focused mandate. Votes replace vetoes. |
| Covenant enforcement | A UN body cannot hold covenants against its own member states. | An external covenant holder with step-in rights and gating power. |
| Launch path | Universal-membership negotiation takes a decade. | A coalition of the willing charters in two years, then grows. |
| Legitimacy | Universality of membership. | Equal donor and implementer board seats, plus partner-country voting power. |
| Donor confidence | Diffuse accountability across agencies. | Bankable covenants, audited baselines, a credible path to AAA issuance. |
1997
The Ottawa Process
Canada led the landmine ban outside UN consensus machinery, from campaign to treaty in just over a year. Focused coalitions beat universal committees when speed is the job.
2017
Powering Past Coal Alliance
Canada and the United Kingdom built the energy-sector version of the same playbook: a voluntary coalition that moved national plans without waiting for consensus.
Standing
The honest broker
No colonial past in the target region. Deep credibility with both donor capitals and partner governments. The charter holder is not a creditor to the mission.
Borrow the governance that survived a scandal
Equal voice, from day one
The Global Fund split its 20 voting seats evenly between donors and implementers. The Facility starts from the same template.
Supermajority, not consensus
Decisions by vote with supermajority protection, never by consensus. No GCF-style gridlock.
Integrity from day one
The Global Fund's 2011 lesson: build the independent integrity unit and the evaluation office before the first scandal, not after.
Partner power
The countries whose grids are at issue hold real board power, not observer seats.
The Canada case, long term
- Pension capital.
CPPIB, CDPQ, and OTPP need long-duration real assets. Grid concessions with enforced covenants fit the portfolio.
- Engineering exports.
Hydro-Quebec expertise, AtkinsRealis, Stantec, and WSP get a funded global pipeline instead of one-off projects.
- Critical minerals.
Reliable power de-risks the supply chains Canada depends on, with the DRC and Zambia at the center.
- Sovereign AI compute.
Canada's AI compute strategy gains power-secure partner hosts instead of begging for capacity.
- Francophonie standing.
Many target countries are francophone: Senegal, the DRC, Cote d'Ivoire, Madagascar. The institution speaks their language.
Sovereignty safeguards, by contract not by promise
Host call options at depreciated book valueMandatory transfer sunsetsDispatch stays nationalExposure capsGavi-model co-financing rampsMandatory syndication
07
The decisionThe rest of this page is the design. This part is the evidence: has anything like this actually worked?
This isn’t for every country. It’s for governments willing to enforce revenue collection in exchange for grid financing, before their industrial window closes.
Close one corridorOne asset company with public contract terms, built to plan.
Execute one adjustmentOne formula tariff adjustment executed exactly as written.
Evidence
- Global Infrastructure Facility (GIF) Advisory Council Report FY2025: mobilization claims and project numbers. globalinfrafacility.org
- MIGA FY2025 Annual Report: guarantee issuance and energy portfolio. miga.org
- Multilateral Investment Guarantee Agency (MIGA): FY2025 annual report guarantee issuance data. miga.org
- Green Climate Fund Second Performance Review: disbursement rate and governance. ieu.greenclimate.fund
- Delhi discom privatization: loss-reduction evidence. ppiaf.org
- Global Fund Eighth Replenishment: $11.34B pledges and governance. theglobalfund.org
- IDA twentieth replenishment: $1 in equity enabling $3 in lending. worldbank.org
- Ethiopia beyond electricity access: "Getting more from the grid". rise.esmap.org
- INELFE France-Spain interconnector: a 2,000MW HVDC merchant line with EU and EIB funding backed by a 50-50 joint venture. presse.rte-france.com
- EIU analysis of INELFE: France-Spain interconnector project review. infrastructure.eiu.com
- The Industrial Revolution Atlas: electricity, manufacturing and the deep history of development. journal.polyu.edu.hk
- What Drives Electricity Consumption in Sub-Saharan Africa? A Two-Step Analysis of 30 Countries. tandfonline.com
- Restructuring weak distribution companies (discoms) in India: SEforALL evidence. seforall.org
- The Powering Past Coal Alliance: 15 years of success, 10 years to coal phase-out. poweringpastcoal.org
- The Ottawa Process and the Anti-Personnel Mine Ban Convention: a government-led process outside the United Nations. armscontrol.org
Spot-check the claims above against these sources. Where this site cites figures from research in progress, treat the claim as a hedge: we are describing what the evidence shows, not promising outcomes.