Summary
- UbuntuNet is a member-governed non-profit registered in Malawi, supporting national research and education networks (NRENs) in Eastern and Southern Africa. Both "association" and "Trust" are used in the documents.
- Through a mix of carrier contracts, PoPs, member routes, and the AfricaConnect programme, it has built regional capacity that includes 10 Gbps core links and larger national connections.
- Current operations extend beyond transport into identity federation, cloud and storage, open science, cybersecurity, and NREN development, though maturity varies.
- Aggregating demand and expertise creates value, but weak campus connections, unpaid fees, supplier power, conflict, grant cycles, and reporting gaps can prevent the benefits from reaching researchers.
Network 3 lifted the core to 10 Gbps class
The 2023 report describes upgrading key core links from about 2.5 Gbps to 10 Gbps and also mentions at least 20 Gbps total core capacity with redundancy. These refer to different layers and do not imply a uniform 20 Gbps across every segment.
The most natural interpretation is that there are redundant 10 Gbps links on the relevant core segments. Actual speeds vary by route, carrier, and national connection.
Even so, it marked a substantial change in pricing and operational headroom. It is true that this remains smaller than hundreds of Gbps to Tbps‑class R&E networks in Europe and North America, but comparisons should account for regional procurement economics and starting points, not just speed rankings.
Lower unit costs were an infrastructure outcome, not a side‑effect
The 2023 report showed unit-cost drops of 50–71.5% compared with 2022. Some NRENs that previously had about 130 Mbps–4 Gbps moved to 500 Mbps–8 Gbps depending on conditions.
Collective procurement, long‑term contracts, and core efficiency drove prices down. The achievement is not merely announcing “big bandwidth” but delivering more capacity to members for the same budget.
However, passing savings on to members reduces the Alliance’s financial margin. There is less room to absorb currency swings, carrier invoices, and payment delays. Low prices are sustainable only when ongoing costs are understood and paid on time.
BotsREN and ZIMREN show how regional upgrades reach national NRENs
BotsREN was connected at 10 Gbps in 2023 and upgraded to 20 Gbps in January 2026. ZIMREN was connected at 10 Gbps in June 2025. These are access speeds for the member NRENs, not the entire network.
Still, it shows that core investment translates into explicit national services. The ultimate effect depends on how many institutions are connected behind each NREN, as well as domestic circuits, campus equipment, and operational capacity.
Even a 20 Gbps national entry point can be constrained by a university’s 1 Gbps link, power outages, or an overloaded firewall. Connection announcements should be evaluated alongside utilisation, loss, availability, and institutional coverage.
The regional layer begins beyond each NREN’s boundary
Students and researchers rarely see the name UbuntuNet Alliance. They first encounter their university or research institute’s campus network. Beyond that sit national NRENs such as KENET, RENU, TENET, ZAMREN, TERNET, and MoRENet. UbuntuNet becomes critical when a national NREN exchanges traffic with another country’s research network, reaches international scientific facilities, connects to GÉANT, or uses common services that are hard to sustain alone.
This positioning also explains why the Alliance is important yet does not look like a consumer ISP. It does not sell residential broadband, does not directly connect every university, and does not manage users’ Wi‑Fi. It places a layer of regional routing and common services on top of legally and operationally independent national networks. Other regional RRENs, commercial carriers, clouds, IXPs, and destination institutions complete the end‑to‑end path.
To the user it appears as one session, but authority is split. UbuntuNet can change its own routes, augment core links, monitor regional services, and coordinate outages. It cannot restore a university’s power, fix an out‑of‑date identity directory, force a national carrier to repair fibre, or guarantee the capacity of a remote application. The challenge is not to create the illusion of a single owner but to orchestrate independent administrative domains with clear boundaries of responsibility.
The federation is not a stopgap; it is the operational design
UbuntuNet Alliance is a federation of NRENs, not a holding company. Each member organisation retains its own legal personality, staff, budget, domestic circuits, university contracts, government relationships, and charging models. The regional layer exists because members chose to jointly procure certain capacity, interconnect at common points, operate shared services, and represent collective interests internationally.
In a region where telecom markets, currencies, higher‑education systems, procurement rules, languages, and technical maturity differ widely, this model is realistic. Centrally owning all members’ academic networks would require political authority that UbuntuNet does not possess – and should not possess. The federated form reconciles national autonomy with regional scale, and allows mature NREN experience to transfer to emerging ones.
Complexity does not disappear; it collects at the boundaries. A slow data transfer can involve a campus firewall, a national circuit, a regional PoP, an international carrier, and a destination system. Each operator sees only partial evidence. Therefore, service boundaries, shared telemetry, escalation contacts, change procedures, contribution rules, and workable exit paths become part of the infrastructure.
The legal framework is operational, but public terminology is not uniform
Current documents describe UbuntuNet as a non‑profit association registered in Malawi. Meanwhile, the 2023 annual report and transition papers also describe it as a non‑profit Trust governed by members and Trustees. The latest statutory instruments are not part of the research materials, so neither term should be treated as exclusively correct.
The actual governance structure is relatively clear. Instead of shareholders there are member organisations, with a Members Assembly, a Board of Trustees, and a Secretariat headed by a CEO. It can employ staff, receive grants, enter into carrier contracts, hold or lease assets, and operate services. There is no evidence of distributable shares or a conventional corporate valuation.
Calling it a “private telecoms company” would wrongly imply investor control and profit distribution. Calling it merely a “voluntary community” would understate its contractual responsibilities and fiduciary duties. The most defensible description is a Malawi‑registered non‑profit NREN membership organisation whose official documents use both association and Trust.
The founding problem was not a lack of internet, but cost and routing
Before UbuntuNet, universities in the region already had internet access. What was missing was a sustainable mechanism to aggregate academic demand, exchange research traffic within the region, and connect to global R&E networks on terms suitable for education and science. Institutions bought small amounts of international capacity at high unit prices, and traffic between African countries sometimes detoured through Europe.
Commercial circuits offer general internet reachability but do not automatically create direct peering with neighbouring NRENs, predictable capacity for data‑intensive research, federated identity, joint security, or collective bargaining power. Small university communities had little leverage over pricing, and new NRENs struggled to demonstrate value before building a critical mass of member institutions and capacity.
The Alliance changed the unit of demand. Campus demand is aggregated by national NRENs, which are then aggregated by the regional organisation. This community could plan PoPs, negotiate long‑term contracts, staff an operations team, and engage international partners. It did not eliminate commercial markets; it entered them as a collective buyer with a public purpose.
Five NREN concepts formed the first federation
UbuntuNet Alliance was designed in late 2005 by NREN concepts from Kenya, Malawi, Mozambique, Rwanda and South Africa. The founding base is linked to what were then KENET, MAREN, MoRENet, RwEdNet and TENET, although their maturity was not uniform.
TENET and KENET had operational experience and a university membership base. MAREN provided an institutional foothold in the country where the Secretariat would later be based. MoRENet represented a government‑backed national model. RwEdNet later underwent de‑commissioning and rebuilding, demonstrating that a regional backbone cannot substitute for a sustainable national organisation.
The mix mattered. A club of only mature networks might have remained a limited interconnection circle. A collection of only fledgling concepts would have lacked traffic, skills, people and trust. The federation allowed real services and national capacity building to proceed in parallel.
Francis Tusubira helped turn a network proposal into an institution
Francis “Tusu” Tusubira is closely associated with the founding period and the African NREN movement. What was needed was not merely router selection. Universities, governments, regulators, carriers, funders and engineers had to be persuaded to see a regional research network as a durable institution rather than a short‑term project.
That required legal incorporation, membership rules, accounting, contracting capacity, regular meetings, and a shared narrative that an NREN is not simply a cut‑price ISP. Tusubira’s role should be weighed against this collective institutional formation, without erasing the contributions of the founding NRENs, Boards, technical communities and international partners.
The durable outcome was an organisation capable of receiving programme funding, contracting for long‑term capacity, and representing a regional community. That institutional capacity became the implementation platform for AfricaConnect.
Amsterdam provided an early legal vehicle; Malawi became the regional home
In 2006 registration papers were lodged with the Amsterdam Chamber of Commerce. For a forming organisation, a Dutch legal entity provided a contractual and project framework familiar to European R&E partners, enabling international cooperation before a regional legal base was in place.
As the organisation matured, keeping a formal legal seat in Europe lost alignment with its regional mission. A new constitution was adopted in Malawi in 2012, registered in 2013, and the transfer of assets and liabilities was documented. The Secretariat now sits in Lilongwe.
This migration brought governance closer to its membership region, but it left the terminology split between association and Trust. The correct chronology distinguishes the 2005 concept, the 2006 initial legalisation in the Netherlands, and the 2012‑2013 Malawi transfer.
Early GÉANT transit demonstrated utility before a full regional backbone
Historical records indicate that R&E transit towards GÉANT was implemented in 2008. Even at limited scale, it gave member NRENs a path to European and global research networks and showed practical value without waiting for a completed regional backbone.
The GÉANT relationship brought not only bandwidth but also procurement experience, project management, identity, security, operational know‑how and access to EU development funding. It accelerated construction but also tied expansion to European programme cycles.
The reality sits between “everything was imported from Europe” and “wholly independent.” UbuntuNet used international collaboration to build capacity that its membership region governs. The key question is whether that capacity can transition to normal operations once grant support ends.
Long‑term carrier deals turned collective demand into infrastructure
Fifteen‑year contracts with WIOCC (2013) and SEACOM (2016) gave a planning horizon longer than short‑term small‑capacity agreements. Long deals lowered unit prices, secured specific routes, and made it possible to plan redundancy and upgrades.
They also create long‑term obligations. Risks around carrier quality, technology change, maintenance, foreign currency and continuous payment remain. International carrier invoices are still due even when member NRENs are late.
These agreements illustrate how a federation becomes infrastructure: it bundles demand, contracts on behalf of several members, houses equipment, and coordinates restoration. It does not, however, own all the fibre, landing stations, or transmission systems on the routes.
Going live in 2014 turned the Alliance from a concept into a production operator
In July 2014 UbuntuNet Alliance and DANTE announced a 622 Mbps high‑speed regional link. Modest by today’s numbers, it was a substantial leap for networks that were then purchasing dozens or a few hundred Mbps at high cost.
Going live brought production traffic, faults, configuration, on‑call rotas, and member expectations. Routing mistakes or outages were no longer internal project issues; they became operational problems affecting universities and research institutes.
The first production network built the PoP ecosystem, NOC processes, carrier relationships and charging mechanisms that later phases could scale up.
AfricaConnect2 widened scope and built eastern path resilience
AfricaConnect2 was announced in 2015 as a €26.6 million programme covering multiple regions and organisations including UbuntuNet, WACREN, ASREN and GÉANT. It was not UbuntuNet‑only revenue.
For UbuntuNet it widened connectivity and membership reach and underpinned the long‑term SEACOM route from Kampala to Dar es Salaam to Amsterdam. It added eastern path diversity and reduced dependence on a single corridor.
At the same time it reinforced the federated, pan‑African structure in which separate RRENs develop their own networks inside a common programme. Results are uneven because national markets and organisational maturity differ.
AfricaConnect3 made the service stack as important as the backbone
With a total budget of €37.5 million, AfricaConnect3 supported networking, pricing, training, identity, cloud and open science until April 2025. The Alliance began to be measured not just by the number of links but by whether members could actually use the common infrastructure.
Alongside Network 3, new PoPs, faster national connections and lower unit costs, it expanded federated identity, repositories, Utafiti Africa, AfricArXiv and cybersecurity.
This mission expansion is logical: without storage, identity and people, fibre is underused. But increasing services also increases software, data, security responsibilities and ongoing costs.
AfricaConnect4 brings fresh investment and a sustainability clock
AfricaConnect4 began in 2026 with an additional €40 million of EU support over four years, covering sub‑Saharan regional R&E networks. The figure is not an UbuntuNet allocation alone.
It includes national and campus layers, cybersecurity, climate observatories, EUMETCast, federated computing, GPUs and open science. It recognises that a fast regional core alone will not produce scientific output unless universities can use it.
When initial funding ends around 2030, observatories, repositories, identity, cloud and compute could remain. Each service needs an early decision on a permanent operator, budget, charging, staffing, and conditions for decommissioning or migration.
AS36944 defines UbuntuNet’s public routing boundary
UbuntuNet uses AS36944. PeeringDB and BGP observation confirm the network identity, several peers, prefixes and observed RPKI validity. This is evidence of a public boundary, not a ledger of every private link.
PeeringDB reports a self‑declared traffic range of 5–10 Gbps and about 1,200 IPv4 and 150 IPv6 prefixes. bgp.tools sees a smaller number of direct origins. The difference lies between announced customer routes, direct origins, and observation points.
None of these numbers substitutes for core capacity or member connection speeds. The ASN confirms participation in BGP operations; it does not show availability, peak traffic or physical diversity.
PoPs and peering create options, but the map is not an asset register
The 2023 report lists Cape Town, Mtunzini, Maputo, Dar es Salaam, Nairobi, Amsterdam, Gaborone and Johannesburg, with Djibouti and Mombasa under development. Older pages also mention London.
Differences can be explained by points in time, classification as core PoP versus handoff, and routing changes. A point on a map does not tell you who owns the router, the fibre, the rack, the port, or the contract.
The Johannesburg PoP and NAPAfrica peering broadened southern choices, but real redundancy depends on independent ducts, carriers, power, data‑centre facilities and control planes. Two map paths can share the same physical failure point.
Lagos–Cape Town materialises a pan‑African federation
In a 2025‑2026 effort, WACREN connected into the South African environment through ZAOXI, SANReN and TENET, enabling interoperability with UbuntuNet. Stakeholders described it as the first effective direct African RREN‑to‑RREN connection realised within the continent, avoiding European transit.
The claim should be sourced and scoped to RRENs. It does not assert that no earlier African north‑south commercial or private link existed. The novelty lies in the institutional and operational relationship between regional research networks.
It can keep more research traffic within the continent, reduce tromboning, and support cross‑regional projects. However, it still depends on carriers, IXPs, BGP policy and national partners; no single organisation controls the entire path.
Membership numbers reflect different classifications and points in time
The 2023 report cites 15 active members, 13 member countries, more than 3.5 million users and over 1,000 institutions. A March 2026 audit RfP lists NRENs from 18 countries; the Home page displayed 16 profiles. Legal membership, active status, connected status, and page‑update cycles can all differ.
Lesotho joined in November 2025, with a Trustee election in 2026. Every figure should carry its date and definition.
Membership is not proof of a live connection, and a connected country does not mean every university receives equal service. In conflict zones, legal membership and operational reality can diverge.
Campus infrastructure determines whether regional investment works
A fast core does not fix a thin campus circuit, old Wi‑Fi, unstable power, a saturated firewall, or understaffing. User experience is set by the weakest segment.
This is why AfricaConnect4 includes national and campus layers. Without storage, servers, transfer tools and a proper LAN, regional capacity goes unused.
Responsibility is shared. UbuntuNet coordinates the regional layer, NRENs build domestic networks, universities run their campuses, and governments, carriers and funders shape the environment. Measurement should follow the path beyond the national connection point to the researcher.
Lilongwe carries governance; Kampala concentrates network operations
The Secretariat is in Lilongwe, Malawi, while organisation‑ and recruitment‑related materials place the NOC in Kampala, Uganda. This is a natural distributed arrangement for a regional organisation.
The Secretariat handles membership, contracts, projects and finance; the NOC monitors links, opens tickets, coordinates carriers and supports NRENs. Separation can improve resilience, provided that secure access, clear authority and business‑continuity plans are in place.
Documentation, on‑call rotas, credential safekeeping and succession are essential. In a small organisation, critical knowledge concentrated in one person becomes a single point of failure itself.
Joint operations should strengthen, not replace, national capability
Not every NREN can run a 24‑hour NOC. UbuntuNet can provide monitoring, preventive maintenance, and first‑ and second‑line support. The national NREN retains responsibility for customers, policy, carriers and the domestic network.
Shared operations work when they compensate for gaps and transfer knowledge. If members lose the ability to understand their own configuration and cannot exit the service, a new dependency forms. Operating authority, change windows, logs, escalation, costs, and migration must be explicit.
A regional NOC can diagnose and coordinate but cannot repair domestic fibre without a carrier. Its strength lies in correlating evidence from multiple sources and in shared operational memory.
Federated identity is a second network, built from institutional trust
Where the physical network carries packets, an identity federation lets a service trust a user authenticated by another organisation. A university manages accounts; a national federation sets rules and metadata; regional and global layers bridge trust.
UbuntuNet does not hold a centralised regional password store. It runs or supports proxies, federations, metadata and eduGAIN connectivity. Quality depends on local directories, account de‑provisioning, multi‑factor authentication, certificates and attribute management.
Home‑institution credentials unlock library resources, cloud, repositories and compute. Conversely, a compromised identity provider can spread false trust to multiple services. Federated trust requires federated auditing and revocation procedures.
eduroam turns campus identity into a mobility service
eduroam lets a user authenticate with home‑institution credentials while visiting. The request may travel through campus, national, regional and global RADIUS proxies back to the home institution.
The simple user experience rests on certificates, realms, secure Wi‑Fi and accurate accounts. UbuntuNet can operate the regional tier; it cannot fix an expired account or a misconfigured access point.
Success should be measured by participating institutions, successful authentications, failures and resolution time, not just by the number of servers deployed.
eduID.africa and eduGAIN create paths to a wider trust circle
eduID.africa, operated jointly with WACREN and ASREN, acts as both a catch‑all federation and a model for NRENs that do not yet have a mature national federation. It allows participation without waiting for every country to complete a full stand‑alone setup.
MAREN and TERNET joined eduGAIN in 2025, connecting Malawi and Tanzania to the global interfederation. Joining is evidence of institutional interoperability, not evidence that every campus has reached the same maturity.
The regional layer should accelerate national capability and enable a clean migration path. If it becomes a permanent opaque substitute, it erodes autonomy.
Research4Life turns federated identity into real access to knowledge
A 2025 pilot enabled institutions in Kenya, Malawi, Uganda and Zambia to access Research4Life remotely using their home credentials. A follow‑up report showed 66 institutions in 15 countries on‑boarded.
Shared passwords are hard to expire, audit, or trace to an individual. A federated approach ties entitlement to person and affiliation, reducing the need for separate accounts for each resource.
Numbers are timestamped; they do not imply universal availability across all member institutions. Each institution manages its identity; Research4Life decides eligibility. Still, it is evidence of concrete value that researchers can see.
Cloud services mix regional infrastructure with commercial procurement
On one side are community‑run or community‑hosted platforms such as the UbuntuNet Open Science Cloud, file transfer, Drive, hosting, collaboration environments and experimental compute. On the other are commercial cloud procurement paths such as the 2024 Redington/AWS agreement.
Regional cloud can keep knowledge and control close to members. Hyperscalers offer a product breadth that a small RREN cannot replicate. UbuntuNet can aggregate demand and support, but it does not own AWS and does not set pricing, availability, or roadmap.
The 2023 report documents a Lusaka installation and hardware for Kigali, but it does not disclose current aggregate capacity, utilisation, tenant numbers or a common SLA. Hardware installation alone is not a mature multi‑region cloud. Orchestration, identity, backup, security, support and a financial model are needed.
Research support services address everyday friction, not only big science
Utafiti Africa aggregates research funding and training opportunities, reporting over 1,100 active users in 2024. It does not disburse grants; it lowers the cost of discovering opportunities.
File transfer, Drive, Meet, Learn and hosting reduce duplicate procurement and ease collaboration. Maturity varies, however; older service pages mix live and planned offerings.
Sending a large file once is different from sustaining continuous data flows for telescopes, climate, or genomics. The latter requires tuned endpoints, storage, monitoring, and sometimes dedicated circuits. Being clear about each service’s scope matters.
Open science has become a second infrastructure layer
Connectivity alone does not make papers or data discoverable, citable, or preserved. Repositories, metadata, persistent identifiers and people depend on hosting, storage, identity, certificates, backup and a sustainable organisation.
UbuntuNet and Access 2 Perspectives have co‑hosted the AfricArXiv platform since late 2023. AfricArXiv is community‑led, not solely owned by the Alliance. UbuntuNet was also selected as the Southern Node of the Africa Open Science Platform.
Repository responsibilities are long. Software, records, rights, security and preservation must be maintained, and data must be exportable if the relationship changes. Damage from an abandoned scholarly record can be harder to reverse than a swapped circuit.
Repository operations and persistent identifiers demand continuous institutional labour
The Research Repository Stewardship Programme trains librarians and data stewards. In July 2026 it ran its first Portuguese‑language workshop in Mozambique, showing that localisation is an operational need, not just publicity.
Institutions need deposit policies, metadata standards, rights handling, user support, backup, upgrades, preservation and a budget. A multi‑tenant platform reduces duplication but creates a shared failure domain. The Network Adoption Fund aims to grow NRENs into sustainable service providers.
DataCite, DOIs, ORCID and the 2026 DOCiD partnership with TCC Africa add identifiers to people, institutions and research outputs. Identifiers do not fix bad metadata. UbuntuNet’s role is to coordinate procurement and integration while keeping global governance and local responsibility visible.
Cybersecurity capability is distributed across people and institutions
UbuntuNet runs regional boot camps, national workshops, identity and systems training, and participates in international R&E security communities. There is no evidence of a single SOC monitoring all members; the model is to strengthen national and institutional teams, creating liaison relationships and common procedures.
Legal, data and incident authority remain national or institutional. The regional organisation can share tools and knowledge, but cannot automatically remediate a compromised endpoint on a campus.
In 2026 it trained finance officers from 15 countries. If they cannot handle circuit contracts, foreign currency, grant compliance and member contributions, technology is unsustainable. Administrative capacity is also infrastructure.
Governance continuity is visible even where public information has gaps
The Members Assembly is the highest body, the Board of Trustees provides oversight, and the Secretariat executes. Prof. Madara Ogot was appointed CEO for a four‑year term from 1 February 2022, and 2026 materials still refer to him as CEO, though no separate reappointment announcement has been found. Prof. Hellicy Ng’ambi became the fourth Chairperson on 1 January 2024.
Miriam Chahuruva and Sabelo Dlamini were elected Trustees on 29 May 2026. No fully reconciled current Board list after the election, or a public document listing every departure, has been located.
This is not evidence of a governance vacuum, but it limits what can be safely stated and highlights a need to synchronise public directories.
2023 finances show a large mission and extremely thin margin
2023 income was US$2,404,779, expenditure US$2,393,911, leaving a surplus of US$10,868. Connectivity services income was US$1,902,318 (79.1%); network operations costs were US$928,032.
A surplus of that size is small against regional‑scale responsibilities. Currency swings, carrier invoices and payment delays can easily absorb it. Passing procurement savings to members fits the mission but shrinks organisational reserves.
The CEO reported that O&M contribution arrears from member NRENs were severely hampering the Alliance’s ability to meet obligations. Full audited financials beyond 2023 have not been found; AfricaConnect figures should not be used to estimate current cash or debt capacity.
Funders, carriers, currencies and conflict sit inside the operating model
AfricaConnect provides the financial envelope for large expansion, commercial carriers supply the physical foundation, and clouds manage the platforms. Contracts are often denominated in euros or US dollars, while member NRENs earn in local currencies. Conflict can constrain a national network even while the regional core remains live.
The de‑commissioning of RwEdNet in 2018 and its rebuilding from 2024 illustrate the institutional fragility of national NRENs. A regional circuit could not prevent organisational collapse; recovery required rebuilding a domestic institution. Membership in a conflict‑affected country also does not prove continuous service.
These dependencies do not make the Alliance meaningless; they define its work. It must diversify supply where possible, maintain reserves, strengthen national capacity, document responsibilities, and avoid mistaking sovereignty for isolation.
Public evidence is strong on programmes and thinner on current operations
UbuntuNet publishes ample details of contracts, connections, elections, training and service launches, allowing its history and accountability to be reconstructed. For current operations, the public record lacks a per‑route asset register, multi‑year availability data, a full traffic time series, link utilisation, a public incident database, a current staff headcount, and audited financials after 2023.
PeeringDB traffic ranges, bgp.tools origins, the “20 Gbps core” expression, and 10/20 Gbps member connections are different metrics. The word “PoP” can refer to a router, a carrier handoff, or an exchange location.
Absence of public outages is not evidence of no outages; lack of metrics is not evidence of low quality. Greater visibility of availability, restoration, utilisation, payments and service adoption would make it easier to judge whether project outcomes have become steady‑state operations.
Regional autonomy comes from options, not from eliminating external dependencies
UbuntuNet increases regional control through collective procurement, intra‑Africa routing, skills development, and operating identity and repository services. It cannot erase global standards, commercial cable systems, equipment supply chains, hyperscalers, or international research partners.
Relationships with WACREN, ASREN, SANReN, TENET and GÉANT illustrate a pan‑African R&E federation that has no single owner. Lagos‑Cape Town reduces some European tromboning, but it still relies on carriers, IXPs and coordinated routing.
Every choice involves trade‑offs. An AWS framework improves procurement and increases single‑vendor dependence. A regional repository strengthens governance and demands software and staff. Long‑term contracts cut unit prices and raise supplier and currency risk.
The strategic test is changeability. Can a member NREN switch carriers, export its data, move workloads, and keep its identity? Autonomy grows when exit is practically available.
UbuntuNet turns a distributed network into a usable research environment
The central achievement is institutional as much as technical. Independent national NRENs aggregate demand, share operations, connect globally, and co‑build services that would be costly or impossible for one country alone. Backbone, identity, repositories, cloud and training are separate products, but they answer the same problem of regional fragmentation.
The organisation is not a finished product. Capacity is still small for data‑intensive sciences such as climate, astronomy and AI. Public information is stronger on milestones than on performance. Member finances are uneven. Dependencies on suppliers, funders, and services of varying maturity remain.
Yet the value is clear. UbuntuNet is not a “separate African internet” isolated from the world, nor a completed continental platform. It is a cooperative regional layer that enables a larger research environment. Long‑term success should be measured by whether members can reliably use the services, understand the costs, keep local authority, move data and identities, and exit if they must.