Hans van Linschoten: 6G is a 2029 standard and a 2026 decision
Why Africa’s federated grid is its best preparation for the next network — (Part 3 of 4)
Africa’s mobile networks are one of the continent’s great success stories. They are also a lesson in dependency. By one widely cited estimate from 2019, Huawei built around 70 per cent of Africa’s 4G networks. The cloud layer arrived the same way from the other direction: through US hyperscalers whose African cloud-region footprint remains limited. Each wave came as a finished package, designed and financed elsewhere, and adopted because it was good, available and affordable. That was a rational choice at the time. It also left African operators and governments with very little say over how their own digital infrastructure evolves.
6G threatens to repeat that pattern, with higher stakes. For the first time, the network and the computer are being designed as one increasingly integrated system. If Africa imports both as one package, the dependency doubles. The good news is that the timetable gives the continent a window — and that the federated grid I described in the first two parts of this series is exactly the asset that window rewards.
What 6G actually is
Strip away the marketing and 6G is three things. It is more spectrum and faster radio, as every generation has been. It is new capabilities, such as using the network itself to sense its environment. And — the part that matters here — it is designed to be AI-native: artificial intelligence built into how the network runs and into the services it offers, rather than bolted on afterwards. The ITU’s framework for the next generation, IMT-2030, explicitly lists “AI and communication” among its core usage scenarios.
The timeline is now concrete. 3GPP, the body that writes the global mobile standards, has approved the schedule for Release 21, which will contain the first 6G specifications. The system architecture is due to be finalised by June 2028; the detailed protocols by December 2028; and the final ASN.1/OpenAPI freeze in March 2029. In parallel, the World Radiocommunication Conference in October–November 2027 will consider spectrum decisions. Its agenda includes candidate frequency bands for mobile use at 4.4–4.8 GHz, 7.125–8.4 GHz and 14.8–15.35 GHz, with regional differences in the bands under consideration.
It is worth being honest about what this means on the ground. In most African markets, 5G is still a small share of connections, and 4G remains the workhorse. Commercial 6G on this continent is a story for the 2030s. That is precisely the point: the decisions that determine who controls 6G in Africa will be taken long before the first 6G handset is sold here.
The network becomes a computer
AI-native has a practical meaning. It means that inference — running trained models — happens inside and right next to the network: to steer radio resources, cut energy use, predict faults, and offer intelligent services to applications at very low latency. All of that needs compute capacity close to users, distributed across many sites, run with the reliability of a network rather than a conventional data centre.
Someone will own that compute. There are three main candidates. It can be integrated into the network vendor’s equipment, so that the AI layer comes in the same box as the radio. It can be provided by hyperscalers, who are already extending their cloud infrastructure into operators’ networks as “edge zones”. Or it can be owned and run by African operators, enterprises and governments themselves, on an open platform they control.
The first two options are convenient. They also risk repeating how 4G and cloud dependency were created. The third option requires something that has to exist before 6G arrives: a distributed, locally owned compute layer that is already earning its keep.
Three dependencies, three exits
It helps to separate the layers, because the exits are different for each.
The radio. Mobile radio equipment will continue to come from a handful of global vendors, whether Chinese, European or Korean. Africa will not build its own base stations at scale this decade, and pretending otherwise is not a strategy. The realistic goal is optionality: open interfaces between radio and the rest of the network, so that an operator can mix vendors and switch suppliers without replacing everything.
The telco cloud. Modern networks increasingly run as software on general-purpose servers. Here open source has become a serious option. The Sylva project, launched under Linux Foundation Europe in 2022 by major European operators and network vendors, is building an open cloud stack for exactly this purpose. Operators that run their network functions and their AI workloads on open, cloud-native platforms have greater freedom to choose what runs on top, provided interoperability and operational independence are maintained.
The AI layer. As Part 1 of this series argued, open-weight models now trail the frontier by months, not years. An operator that runs its own models — for network optimisation, for customer service in local languages, for enterprise services — on its own compute does not need to rent intelligence from a foreign platform.
None of this delivers autarky, and it should not be sold as such. Chips will still be imported. European vendors and European open-source projects are not African either; replacing dependence on Washington and Beijing with dependence on Brussels would be a modest improvement at best. The real objective is choice: the ability to change suppliers, to run critical functions under African jurisdiction, and to negotiate from strength.
What preparing looks like, 2026 to 2029
For regulators, the immediate lever is spectrum.
Africa’s positions for the 2027 conference are being prepared through the African Telecommunications Union, with regional blocs such as ECOWAS coordinating their positions and the ATU having convened its third African preparatory meeting in Tunisia in September 2026.
A coordinated African voice on which bands are harmonised, and on what conditions, is the cheapest sovereignty Africa can buy. The less obvious lever is to treat spectrum and compute policy together: licence conditions and national digital strategies can favour open interfaces and shared, locally owned edge infrastructure, rather than leaving the architecture to whichever vendor package is cheapest on the day.
For operators, the move is to build edge compute now — not as a 6G science project, but as federated cloud locations that earn revenue today from enterprise, government and AI inference workloads. That is the grid from Part 2. When 6G arrives, the same distributed footprint can become the network’s compute layer. Operators should also write open interfaces into their 5G-Advanced procurement now, because today’s contracts set tomorrow’s upgrade path.
For governments, procurement is the instrument. Insist on open standards, data residency under national law, and real transfer of capability — engineers trained, source code accessible where applicable, the right to operate independently. Be most cautious with turnkey offers that bundle network, cloud and AI in one vendor-financed package: they are the fastest route to deploying 6G, and potentially the surest route to repeating 4G.
For universities and AI labs, the opportunity is the people. Every one of these choices depends on African engineers who can run, adapt and extend open platforms. The countries that train them in the next three years will have options in 2030. The ones that do not will have suppliers.
The no-regret move
What makes this strategy robust is that it does not depend on 6G arriving on schedule, or at all. A federated grid of operator- and enterprise-owned compute can pay for itself on today’s demand: sovereign cloud, private AI, backup and recovery, local-language services. If 6G arrives as planned, that grid can provide the AI-native layer the network needs, already in African hands. If 6G is delayed or fragmented between rival technology blocs, the grid still serves its customers. Either way, the investment need not be stranded.
That is the same logic as Part 1 and Part 2: build modestly, build distributed, build in a way that wins in more than one future. 6G simply raises the stakes and sets the deadline.
In the final article of this series, I will look at the newest users of such a grid — not people, but AI agents — and at what it takes to let software build and deploy software on sovereign infrastructure, safely, inside the border and within the rules.
Hans van Linschoten is the CEO of whitesky.cloud, a European sovereign cloud platform, and Co-founder of AfriQloud, a pan-African sovereign cloud and edge AI infrastructure initiative.
Follow whitesky.cloud on LinkedIn
