DIGITAL PUBLIC INFRASTRUCTURE IN AFRICA · FIELD GUIDE V2.1 · DATA AS OF 2026-09-18
Africa has the code, the political will, and the capital. What it lacks is the artefact in between.
A field guide to the African DPI market for people who have never met any of it, built on UNDP's June 2026 flagship policy paper and fifty-eight dated anchors. New to this? Start here. The finding, said three ways by UNDP itself: the binding constraint is not the availability of capital but the readiness of bankable projects. This edition renders the full evidence base — adoption data for all 55 AU member states, the money flows, the live programmes and the risks — as an interactive, drillable document.
Evidence graded: A primary institutional > B reputable secondary > C author's analysis · ⚠︎ flagged items not for reuse without re-verification · Private working analysis — not investment, legal or procurement advice
§0 · Start here
This section tells you what this document is, what question it answers, and how to read its evidence. It takes two minutes. Nothing here assumes you have met any of this before.
A research file on national digital infrastructure in Africa — the shared systems that let a person prove who they are, move money, and let government institutions exchange data.
It exists because in June 2026 the UN Development Programme launched a continental programme for exactly this, and published a 28-page policy paper explaining why it was needed . Buried in that paper is a sentence that describes a business problem rather than a policy one. This document takes that sentence seriously and works out what follows from it.
Fifty-plus countries want to build national digital infrastructure. The software is free and proven. Development banks have money committed and unspent. So why isn't it being built — and what is the missing service?
Why it matters Because the answer is not a technology answer, and the people best placed to supply it are not currently in the room.
You are the expected reader. Read §1 Background next. It defines the three terms the whole field runs on, gives you an analogy from enterprise networking, shows how the world got from a UN panel in 2018 to a ministerial meeting this month, and traces how money actually travels from a donor to a contractor. About ten minutes, and everything after it depends on it.
Already know what DPI is? Skip to §2 The finding.
That gap is §5. Its commercial consequences are §6.
Every factual claim carries an anchor — the small blue codes like . Click one and it opens the source, the source's own as-of date, and an evidence grade.
A primary institutional source, dated B reputable secondary or trade source C the author's analysis, not a sourced fact ⚠︎ flagged: inconsistent, stale or contested
Grade C material is labelled wherever it appears. Nearly all of §6 is grade C — it is an argument built on sourced facts, not itself a sourced fact. Treat it accordingly.
What this document is not. It is not a market sizing. Only one hard systems-integration contract value in this market is publicly anchored , and inventing the rest would have been easy and dishonest. Where a number does not exist, §9 records it as an open question instead of filling the gap. It is not investment, legal or procurement advice.
§1 · Background — what this world is
Three terms, one analogy, a short history, and a map of where the money comes from. If you have never encountered DPG, DPI or SDG before, this section is sufficient to follow the rest of the document and to hold your own in a room.
Sustainable Development Goals
Seventeen goals adopted by all UN member states in 2015, targeting 2030.
What they are, functionally: the budget codes of the development world. Donors, development banks and UN agencies allocate against SDG targets. Technology that cannot be tagged to an SDG cannot be funded from those pools. You do not need to believe anything about them. You need to know that they determine eligibility.
Hover for the full story
Digital Public Good
An open-source asset certified against a nine-point standard and listed in a public registry — roughly 254 are listed.
The definition comes from the UN. In June 2020 the Secretary-General's Roadmap for Digital Cooperation defined DPGs as "open source software, open data, open AI models, open standards and open content that adhere to privacy and other applicable international and domestic laws, standards and best practices, and do no harm" . Functionally, a DPG is a procurement passport — it converts "some open-source project" into "an approved component". Certification expires annually, so it is a recurring compliance cycle, not a permanent badge .
Hover for the full story
Digital Public Infrastructure
What a country actually gets when it deploys those components as national infrastructure — not the code, but the running system plus the law, institutions, operations and budget around it.
UNDP's definition names three pillars : digital identity (prove who you are, remotely and verifiably), digital payments (move money between any two parties, instantly, across providers), and data exchange (let institutions share data without N² point-to-point integrations).
Hover for the full story
A DPG is portable. A DPI is sovereign. MOSIP the software is a global good. Ethiopia's Fayda ID is Ethiopian infrastructure under Ethiopian law, operated by Ethiopian institutions, and Ethiopia's problem when it breaks.
— Field Guide, §0.1 · The distinction that matters
Context DPGs are certified against the DPG Standard; DPIs are safeguarded under the Universal DPI Safeguards Framework — 13 lifecycle risks, 300+ recommendations . Different instruments, different bodies, different failure modes. Why it matters Get this wrong in a room and you will be corrected.
DPI is to a country what the core is to an enterprise network. The pattern has been built before under different names — and the commercial logic is the one already known from enterprise open source:
| Enterprise pattern | National equivalent | Reference implementations |
|---|---|---|
| AAA / directory / IdP / federation | Digital identity | MOSIP, eSignet, Inji; Aadhaar (IN), Fayda (ET), Ghana Card, NIN (NG) |
| Transaction bus / clearing / settlement | Instant payments | Mojaloop; UPI (IN), Pix (BR), TIPS (TZ), PAPSS (continental) |
| ESB / service mesh / API gateway / iPaaS | Data exchange | X-Road; infoHighway (MU), GSB (ZM), Irembo (RW) |
| Open-source reference stack (Linux, FRR, K8s) | The DPGs themselves | DHIS2, OpenG2P, DIVOC, the other ~250 |
The code is free. Running it at carrier grade is not.
— Field Guide, §0.2 · The commercial logic
Context Nigeria issued an $83 million tender to modernise its national ID using free MOSIP code . Nobody is paying for licences. Why it matters They are paying for integration, migration, hardening, accreditation, hosting, training and 24/7 operations. That is the entire business; everything after this is detail.
Authority and money are in different hands. UNDP and DPGA set direction but do not write the big cheques. The World Bank and AfDB hold the capital but need someone to hand them a project they can underwrite. That gap between the two clusters is where this report's thesis lives. Positions on the chart below are the author's assessment (grade C), not a sourced ranking.
Development banks have money committed and unspent. Integrators have capacity. What is scarce is the artefact that connects them: a national programme specified well enough that a bank can underwrite it and a contractor can price it. Read the constriction — flow widths are illustrative of relative scale (grade C), not measured values.
Nothing in this field is old. The entire governance apparatus was built inside eight years, and the piece that matters most to this research is four months old.
Two things to take from the shape of that. The vocabulary is younger than most enterprise software contracts — "digital public good" only acquired an official definition in June 2020 . And the African programme is brand new: AA4DPI is four months old at the time of writing , which is why so much of this document rests on a single June 2026 source rather than on a settled literature.
§2 · The finding
Read §1 first if the terms are new. This is the headline; §4 is the detail behind it. Five findings, each carried by the dataset underneath it — every figure drills down to a dated source.
One correction worth being careful about. The $450bn AfCFTA figure that circulates as "the DPI opportunity" is actually the World Bank's estimate of Africa's real income gain by 2035 if AfCFTA is fully implemented . It is not a DPI market size. Using it loosely is a tell.
Only two anchored observations exist — 25 million mobile subscriptions in 2001 and roughly 1.13 billion in 2023 . This edition draws them as two points with a connecting indicator; no intermediate values are invented.
Identity is the pillar that gets the attention and the controversy. It is also the least adopted. Data exchange — the quiet one — is furthest along.
The narrative says Africa is racing ahead on DPI. The data says the continent is at the beginning. For an incumbent, that is a disappointing market. For a new entrant, it means the land grab has not happened yet.
— Field Guide, §2 · The most commercially important fact in the document
Context Thirty-two of fifty-five countries have no active DPI pillar at all, and not one country has all three .
Eight forces act on this market in 2026. Four push down, three push up, and one is widely misquoted. Each drills down to its anchor.
§3 · The market, honestly sized
If the code is free, where is the profit pool? Governments still pay for everything around the code.
The stack below shows where the money goes in a national DPI programme. The percentage bands are the author's estimate from deal-shape reasoning (grade C) and must be replaced with sourced figures before external use — see open question Q3 in the research backlog.
The column that should change a strategy is the last one. Local-content requirements are the structural disadvantage for a US firm. Pools 3, 5 and 6 are where that disadvantage is weakest — because scarcity, product, or pre-procurement positioning outweigh nationality. Pool definitions derive from report analysis; the exposure column is the author's assessment (grade C).
Only one hard SI deal value is publicly anchored. The $83M Nigeria tender is the single best-documented ticket size in this market. Everything else is either portfolio-level MDB commitment or undisclosed. This is a genuine research gap, not an omission — see open question Q1. Do not present this table as a market sizing.
Presence confirmed from cited sources; relative position is the author's read (grade B/C). Six groups hold the field — the pricing problem is the Indian IT majors and MOSIP-native SIs; the template worth copying is the HISP network.
Three observations, all grade C — these are arguments, not facts:
§4 · Africa now — the live board
The headline is in §2. This is the detail underneath it — the 30 states the source individually names, what is demonstrably working, and the programme now driving the agenda.
§4 · Africa now
This is the single most important dataset in the document, and it is considerably more sobering than the press releases.
Discrepancy flagged. The source states n=8 active payment systems but lists only seven countries. Additionally, the appendix states coverage of "the 46 countries covered by the UNDP Regional Bureau for Africa" while the body text reports across 55 AU member states. Both inconsistencies are in the published source . Do not cite the payments count without resolving this — see open question Q2.
Thirty countries carry at least one active or pilot DPI-like system. Absence means no DPI-like system recorded, not absence of any digital system. Filter by pillar or status; hover any cell for the record.
Eight forces act on this market in 2026. Four push down, three push up, and one is widely misquoted. Each drills down to its anchor.
§4 · Africa now
This is the single most important dataset in the document, and it is considerably more sobering than the press releases.
§4 · Africa now
This is the single most important dataset in the document, and it is considerably more sobering than the press releases.
Discrepancy flagged. The source states n=8 active payment systems but lists only seven countries. Additionally, the appendix states coverage of "the 46 countries covered by the UNDP Regional Bureau for Africa" while the body text reports across 55 AU member states. Both inconsistencies are in the published source . Do not cite the payments count without resolving this — see open question Q2.
Thirty countries carry at least one active or pilot DPI-like system. Absence means no DPI-like system recorded, not absence of any digital system. Filter by pillar or status; hover any cell for the record.
Eight forces act on this market in 2026. Four push down, three push up, and one is widely misquoted. Each drills down to its anchor.
The squeeze, stated plainly. Aid is falling by a quarter. The need is unchanged. Domestic budgets are constrained by debt. This is precisely why the financing conversation has shifted from grants to bankable projects that can absorb sovereign lending. Understand this and you understand why AA4DPI exists. UNDP also names climate and energy resilience as a standing reason to build DPI now, alongside the fiscal and inclusion arguments .
Pattern worth noticing: the strongest results are in data exchange and payments, not identity. Identity gets the attention and the controversy; the quieter interoperability layer is where countries are actually succeeding. That has implications for where a new entrant should aim.
Launched 26 June 2026, New York, during UN Open Source Week. Jointly led by the UNDP Regional Bureau for Africa and the UNDP Digital, AI and Innovation Hub . Stage 1 is the bankability stage: donor-funded, pre-procurement, and therefore the one place a foreign firm's nationality is not a structural disadvantage.
Those five mission countries are the warmest leads on the continent. Ethiopia, Malawi, Rwanda, Sierra Leone and Zambia — plus African Union representatives — spent two weeks with UIDAI, NPCI, Indian ministries, the Government of Karnataka and the Bengaluru open-source community, concluding 7 September 2026 . They have just been shown what good looks like, under a programme whose stated purpose is to get them to bankable projects.
The policy paper was produced with research support from Artha Global, funded by a Co-Develop grant, with substantive inputs from CDPI, C4EC, the African Development Bank, GIZ, AFCEN and the Gates Foundation . Treat the acknowledgements as the addressable network — these are the people orchestrating the market.
The paper names five shifts required to accelerate DPI . The UNDP column is sourced; the commercial reading is the author's (grade C).
The most commercially mature economy on the continent. Data exchange and payments are active ; identity is not on the active or pilot DPI-like ID list — a notable gap for a country of its sophistication. SARS eFiling is cited by UNDP as a global example of DPI where adoption followed demonstrated public value ; the SASSA COVID-19 SRD grant reached 10M+ beneficiaries — and exposed documentation, connectivity and system-error exclusions . South Africa has joined the DPGA and the 50-in-5 campaign and is named in UNDP's Preface among countries "demonstrating what DPI can deliver" . It is also one of five countries Nigeria convened — with Ghana, Egypt, Kenya and Rwanda — to coordinate implementation of the AfCFTA Digital Trade Protocol, a peer-led cluster model the wider DPI agenda could copy .
Why it matters for entry: South Africa has the deepest banking, fintech and professional-services base on the continent, a real regulatory apparatus, and — critically — an identity gap in an otherwise advanced stack. It is also a credible base from which to serve SADC. ⚠︎ Needs verification: current status of the digital ID programme and Home Affairs modernisation — the source data shows the gap but not the plan (Q4).
The honest read: regional bodies "cannot compel national uptake. Their role is to lower coordination costs" . Do not build a strategy that depends on continental mandates. Build it on national budgets and MDB lending.
§5 · The bankability gap
Africa does not have a technology problem — the code is free, proven at the scale of billions, and certified. It does not have a political-will problem — thirty-nine countries have signed onto 50-in-5. It does not have a capital problem — the World Bank, AfDB, IsDB, EBRD and bilateral partners all have DPI windows open. What is missing is a national programme specified well enough that a development bank can underwrite it and a contractor can price it.
The biggest barrier to scaling Digital Public Infrastructure across Africa is not political will. It is the lack of structural financing and investment-ready national projects. AA4DPI is designed to help close that gap.
— Ahunna Eziakonwa, UN Assistant Secretary-General; UNDP Assistant Administrator and Director, Regional Bureau for Africa, at the AA4DPI launch
Echoed "The challenge is less the availability of capital than the readiness of bankable projects." — 5Cs framework, Capital And again "The binding constraint is the readiness of bankable proposals, not the availability of capital… unlocking existing financing through stronger project preparation."
The DPI landscape in Africa is crowded with well-intentioned pilots that have not delivered systemic change. What is needed is a shift from proof-of-concept to structured, sustained programmes that are bankable, country-owned and designed for scale.
— UNDP Africa DPI policy paper, June 2026
Why it matters This is the diagnosis of why the artefact never gets produced — pilots are fundable, programmes are underwritable, and the two are different documents.
The structural reason is a professional divide. The people who write the strategy do not build. The people who build are never in the room when the strategy is written. A ministry says "we need a national digital ID." That is a wish, not a project. Somebody has to convert it into a scope, a sequence, a schedule, a cost model, a risk register, an operating model and a procurement package. That is a distinct profession and it is undersupplied.
Nine sections. Most of them are engineering and delivery disciplines, not policy. That is why consultancies produce thin versions of this document and why banks keep sending them back. This construction is the author's (grade C), assembled from MDB appraisal norms and the UNDP 5Cs — validate against AfDB/World Bank appraisal criteria before selling (Q5).
| Payer | Instrument | Why they fund it | Typical scale |
|---|---|---|---|
| UNDP / AA4DPI | Programme TA budget | Their stated mandate: readiness assessments, governance design, institutional arrangements | Unknown — research gap |
| African Development Bank | Project preparation facility | Needs a bankable pipeline to lend against | Unknown |
| World Bank | Project preparation / ID4D | Same | Unknown |
| Gates Foundation / Co-Develop | Grant | Ecosystem building; funded the policy paper's research | Unknown |
| Government | Own budget or loan proceeds | Wants the loan approved | Unknown |
⚠︎ Honest gap. Day rates and contract values for project-preparation work in this market cannot yet be anchored. This is the highest-priority research gap in the document — see open question Q6. Do not build a financial model on assumed rates.
§6 · The opportunity
The commercial analysis — entry models, partner mapping and sequencing — is not published.
§5 establishes that the binding constraint on African DPI is the absence of investment-ready national programmes, not capital, technology or political will. §6 works out what follows commercially: who could supply that missing capability, how they would enter, what it is worth, and in what order.
That section is held back. Its absence does not affect anything else here — §§0–5 and §§7–9 stand on their own evidence, and every anchor cited across the document remains in the register at the end.
Numbering is unchanged so this edition stays aligned with the underlying research file.
§7 · Risk register
Twelve failure modes, plotted by likelihood and impact. Scoring is the author's assessment (grade C); the underlying failure modes are anchored where cited. Almost none are technology risks — they are institutional, political, financial and inclusion risks.
The pattern across R1–R12: that is precisely why a firm that can write a credible risk register is valuable — and precisely why development consultancies, who are good at policy but cannot price delivery, produce registers that banks do not believe. Click any node for the detail and the mitigation.
§8 · The live moment · UNGA 81
High-Level Ministerial and Ecosystem Dialogue — "The Africa We Build: DPI as the Engine of Agenda 2063's Digital Economy" — convened under AA4DPI, Thursday 24 September 2026, 10:00–12:00, Solana Skyline, 6th Floor, 141 E. Houston St., New York .
If one thing survives the room: the constraint being discussed is not technology and not politics. It is project preparation — turning national ambitions into something a development bank can underwrite.
Genuinely useful to anyone in the room, not leading questions:
| When | What | Where |
|---|---|---|
| Wed 23 Sep, 13:00–14:30 | Bridging the Gaps: Digital Financial Inclusion to Accelerate SDGs in Ethiopia | UNOSSC, 11th Floor, Doha Room, 304 E 45th St |
| Thu 24 Sep, 15:30–17:00 | Development Data Partnership: Private Sector Intelligence for Better Development Decisions | LinkedIn office, 350 Fifth Ave (Empire State Building) |
Ethiopia is one of the five mission countries. The second event is hosted in a Microsoft-owned building — worth noting who is convening the private-sector-data conversation.
One observation about the venue. Every other UNDP event on the UNGA 81 list is at One UN Plaza, the AU Permanent Mission, the Marriott Marquis, or UN HQ. This one is at a Solana-branded venue on the Lower East Side. That is an inference, not a fact from the source — but it suggests something about which part of the private sector is being courted into the DPI financing conversation, and it is worth confirming by simply looking at who is in the room.
§9 · What we do not know yet
Eleven open questions. Q1, Q6 and Q2 gate the business case; the rest gate precision.