Innovation Infrastructure

Weak innovation infrastructure hinders scale potential 

Africa's share of world research output sits at approximately 2 percent, research spending at 1.3 percent, and patent production at 0.1 percent of global totals - figures from Brookings's science and technology (2022) analysis that reflect a structural gap which has not closed materially since that assessment was made. WIPO data from Global Innovation Index 2024 shows Africa's share of international patent filings under the Patent Cooperation Treaty has grown from effectively zero in 1997 to 0.8 percent in 2023 - meaningful directional progress, but still a fraction of Africa's 18 percent share of global population.

Historical underinvestment is a compounding structural constraint. Each year of underinvestment makes the next year's gap harder to close, because the returns to R&D investment are disproportionately captured by ecosystems that already have the research infrastructure, the commercialisation expertise, and the industry linkages to translate knowledge into value. The countries that invested heavily in R&D in the 1970s and 1980s are still harvesting that investment. The countries that did not are still paying for it. Justin Yifu Lin's New Structural Economics names the mechanism: economies catch up by building capabilities aligned with their factor endowments and accumulating those capabilities through directed investment over multi-decade horizons. Generic R&D incentives produce neither.

The WIPO Global Innovation Index 2025 tells the same story at country level. Mauritius (53rd) leads sub-Saharan Africa, followed by South Africa (61st), Botswana (87th), and Senegal (89th). Rwanda (104th) is the region's longest-standing innovation over-performer - outperforming relative to its income level for over a decade. Nigeria (105th) is one of the fastest climbers this year. Kenya ranks 102nd - a decline from 86th in 2020 - and scores stronger on innovation inputs than on outputs and diffusion. That pattern is the diagnosis in a single data point: a system that generates activity and infrastructure but cannot translate it into sustained downstream use. The Kenya pattern is one expression of a general empirical regularity. Recent cross-country work using GII panel data across 99 economies (Alqararah, forthcoming) identifies three phases: low-income economies systematically produce more knowledge than they can commercially absorb (Phase 1, capacity failure); middle-income economies see the gap close as commercial infrastructure catches up (Phase 2, commercialisation expansion); frontier economies such as Singapore and South Korea return to negative for the opposite reason - deep science investment running ahead of commercialisation capacity (Phase 3, knowledge compression). The bottleneck identified in Phase 1 is the diagnosis advanced here: credit access, buyer demand, and commercialisation infrastructure - not knowledge production.

Innovation infrastructure here means the institutional architecture that converts knowledge into commercial activity: research capacity, IP transfer frameworks, commercialisation expertise, technical-equipment access, industry linkages, sector-specific testbeds. Physical infrastructure - power, connectivity, roads - is the precondition treated across the Enabling Ecosystem chapters. The two interact, with physical infrastructure deficits raising the operational cost of innovation infrastructure, but they fail in different ways and require different remedies.

The lack of commercialisation expertise is not new - the 2022 analysis identified it. What has changed is the precision with which the failure point can be located. Good ideas generated by universities or research institutions, if not taken up by ventures with high technical capability operating in environments with strong commercialisation infrastructure, do not generate value. The problem is the near-total absence of institutional connective tissue between knowledge generation and market application.

Systemic Innovation and Sendemo's Research-to-Commercialisation in Kenya report (2026) identifies the mechanism precisely: responsibility for long-horizon formative risk - through regulatory clearance, demand commitment, and procurement integration - is structurally unowned in current R2C architectures. Kenya has explicit statutory mandates, functioning institutions, and high programme activity. Conversion outcomes remain weak. The binding constraint is the absence of any actor mandated, capitalised, and governed to carry innovations through the phases where market formation actually fails. That finding is not Kenya-specific. It describes the structural condition of every African innovation system examined.

The OECD/IDIA Innovation Ecosystem Strengthening Learning Journey series (2025–26) names the same constraint at the funder level. Benjamin Kumpf of the OECD Innovation for Development Facility identifies three conundrums across the DAC membership: insufficient funding for the foundational infrastructure investments - roads, electrification, broadband, digital - that ecosystem functioning actually depends on; a lack of imagination in building economic niches with national governments over 10–15 year horizons; and innovation programmes developed in sectoral silos, disconnected from the planning departments and economic policy units that think about growth trajectories. Programme-centric funding, however well delivered, substitutes for the long-horizon institutional function that is actually missing.

The R2C Kenya finding goes further: the system pushes formative risk downstream onto researchers, startups, and small firms - the actors least able to absorb prolonged uncertainty. When programmes end, innovations have no institution mandated or resourced to carry them through the next phase. Activity is generated. Accumulation is not.

Mariana Mazzucato's mission-oriented innovation policy work names the institutional vacuum from a different angle. The breakthrough commercial innovations that produced Silicon Valley - GPS, the internet, touchscreens, voice recognition, the algorithms that became Google - came from sustained, mission-oriented public investment with horizons measured in decades, not from venture capital or generic R&D credits. Mission-oriented investment requires a public sector that maintains directionality across political cycles and has the institutional capacity to set research agendas, sequence procurement, and absorb formative risk on behalf of the system. Most African innovation systems lack the second condition; some lack the first. The policy debate has substituted hubs and challenge programmes for the institutional architecture mission-oriented innovation actually requires.

Senegal's health innovation ecosystem demonstrates what the alternative looks like. Institut Pasteur de Dakar has manufactured WHO-prequalified yellow fever vaccine since 1937 - the only WHO-prequalified yellow fever producer in Africa. The MADIBA manufacturing facility under construction in Diamniadio, financed through Team Europe alongside IFC, the US DFC, AfDB, AFD, and the Government of Senegal, will lift annual capacity to 300 million doses across yellow fever, measles-rubella, polio, and other vaccines. The model worked because successive Senegalese governments sustained institutional continuity across political cycles, framed manufacturing as an employment and skills strategy alongside a health procurement exercise, and used blended finance over a 15-year horizon to crowd in private partners.

Institutional continuity across political cycles is the binding variable, not budget. Ha-Joon Chang's historical analysis of how today's high-income economies actually developed makes the broader case: the developmental sequences that worked were sustained over multi-decade horizons through political-economy arrangements that protected directionality from electoral turnover. The countries that maintained that continuity converged. The countries that did not, did not. That is the model the innovation infrastructure debate in Africa is mostly not having.

African innovation agencies, research institutes, and universities continue to lack the components that make knowledge commercially valuable: technical commercialisation expertise, IP transfer frameworks, industry linkages. Spinout relationships - formal contractual arrangements for the use of IP developed at universities - remain underdeveloped across the continent.

Technology transfer offices, where they exist, are measured on disclosures, patents, and licences - transactions that occur early in the commercialisation process, well before the deeper outcomes the offices ostensibly exist to produce. The metric problem is structural, not African. The AUTM Licensing Activity Survey data on Western TTOs records that even at well-resourced US research universities, only a small fraction of invention disclosures convert to active licences and a smaller fraction still to spinouts, with median licensing-revenue lifecycles running far longer than typical programme cycles. African TTOs measured against Western activity benchmarks fail in part because Western TTOs would also fail when measured against the deeper commercialisation outcomes that matter. This is Goodhart's Law operating at the system level: the measures that exist measure surface activity precisely because the deeper outcomes are slow, uncertain, contested, and difficult to attribute. Fixing the metric problem is harder than fixing the activity problem, and the donor architecture has gravitated toward the easier fix.

AfriLabs counts 514 member hubs across 53 African countries, per its 2024 Impact Report - up from 327 hubs earlier in the decade. The structural treatment of hub ecology sits in Working Ecosystem Support Infrastructure. The argument here is narrower: most African hubs function as co-working spaces rather than centres of technical excellence. A co-working space provides desk space, community, and event programming. A technical innovation centre provides specialised equipment, testbeds, research capabilities, and expert staff - the infrastructure that deep tech, manufacturing, and hardware ventures actually need to develop and test products. The distinction matters because one produces visibility and the other produces capability, and funders have systematically mistaken the former for the latter.

The deeper analytical point is capital structure. A co-working hub can run on event-fee, membership, and 3–5 year donor cycle economics. A technical innovation centre requires 15–20 year capital amortisation, sustained operational subsidy through periods when industry uptake is uncertain, and procurement relationships with both public and private sector partners. The hub-network financial architecture that exists at scale is structurally different from the technical-centre financial architecture that does not. The mismatch is not a question of donor preference; it is that no current African funding architecture supplies capital with the duration profile technical infrastructure requires.

African science and technology clusters are consistently strong in scientific publications and weak in international patenting. They are science clusters, not science-and-technology clusters. Scientific output without commercialisation infrastructure is knowledge that does not reach the market. Building from science to commercial application remains the critical and largely unsolved gap across every African innovation system examined.

The geography of innovation activity compounds the architectural problem. Continental innovation activity concentrates in city clusters - Cape Town, Lagos, Nairobi, Cairo, Casablanca, Dakar - whose internal density makes them function differently from the national systems they sit within. The substantive treatment of urban concentration sits in Spatial Scaling Dynamics; the implication for innovation infrastructure is that national-level policy frameworks calibrated to averages routinely misallocate against the city clusters where the activity actually compounds.

The OECD's STI Outlook 2025 is direct: transformative change requires greater directionality in STI systems, including in their allocation of resources. Generic hub support and undirected R&D incentives produce activity. Directed investment in specific economic niches - with the institutional continuity to carry innovations through regulation, procurement, and adoption - produces capability. The difference between a hub network and a Senegal-style sector strategy is time horizon and institutional design.

The innovation infrastructure deficit is one of the primary mechanisms through which the structural gap reproduces itself across generations. When the infrastructure for converting research into commercial ventures does not exist, the knowledge that African universities and research institutions generate does not translate into scaling ventures. When it does not translate into scaling ventures, the commercial tax revenues, employment multipliers, and ecosystem experience that would give the public sector the mandate and resources to invest further in innovation infrastructure are not produced. The loop closes on itself.

The country-level loop and the funder-level loop reinforce each other. Breaking either requires accepting time horizons that neither annual budgets nor standard programme design currently accommodate.

Until the institutional mechanism that carries innovations through formative risk is built - and until funders commit to the multi-decade horizons that commercialisation requires - the innovation infrastructure gap will continue to reproduce itself regardless of how many hubs are opened, how many TTOs are established, or how many challenge programmes are funded.