Some benign yet significant developments have happened in the AI ecosystem in the last 30 days. This note is an effort to put a pin in time. This draws on learning from reading some interesting pieces in the past week. If you are wondering where things are going within the AI ecosystem then do read on. I include some synthesis of where things are and why it is important today and now.
The executive brief first
The most valuable new work in AI in Africa is moving the African AI debate beyond ethical principles and technical fairness toward questions of political economy, institutional power and material infrastructure:
- Who owns the compute and cloud layer?
- Who governs the data?
- Who bears the energy, water and environmental costs?
- Who can investigate or challenge automated public decisions?
- Can African governments act collectively rather than negotiating separately with global technology powers?
This is highly aligned with some of my more recent pieces and thinking. Africa’s AI challenge is not merely adopting responsibly, but rather building the institutional and infrastructural capacity to choose, govern and negotiate.
Now for a synthesis of what I have read this week
1. AI Futures: An African Agenda for the Global AI Governance Dialogue
Research ICT Africa, 6 July 2026. This provides a good snapshot of how African scholars and policy actors are attempting to influence the new UN Global Dialogue on AI Governance.
The agenda explicitly rejects Africa arriving merely to respond to rules established elsewhere. Instead, it frames African participation around justice, sovereignty, structural exclusion, linguistic inclusion and the distribution of economic value. Its underlying argument is that transparency, accountability and technical fairness are insufficient unless governance also addresses who owns infrastructure, controls data and benefits economically.
The learning: The African contribution to global AI governance is becoming more coherent and affirmative. It is no longer only a critique of exclusion; it is advancing a distinct governance framework.
Why this matters today: It is a strong reinforcement of the fact that indeed Africa does not need a diluted version of European or American AI governance. It needs policy grounded in its own institutional constraints, public-service priorities and position within the global digital economy.
However, Research ICT Africa research ICT Africa’s warning is also a reminder to be intentional. It did this in the explanation of the UN Dialogue, particularly its warning that international norms will mean little unless African states establish independent domestic oversight, enforcement and redress institutions.
2. The UN Global Dialogue has created a forum—but not yet governance
The inaugural UN Global Dialogue on AI Governance took place in Geneva on 6–7 July. It is the first UN platform intended to bring all 193 member states together with industry, researchers and civil society specifically around AI governance. Note though, that even though it can influence norms and future resolutions, it is not a regulator and cannot produce binding rules on its own.
The Dialogue’s mandate includes addressing AI capacity gaps and strengthening access to high-performance computing and relevant skills in developing countries. That is important because compute inequality has entered the formal global-governance conversation rather than being treated only as a commercial or innovation issue. Re: the take from Research ICT Africa.
The learning: AI governance is becoming more institutionalised within the UN system. The conversation is also broadening from risk and safety to access, infrastructure and concentration of power.
Why this matters today: Universal representation is not the same as equal influence. African governments may have seats, but the technical knowledge, negotiating capacity and implementation finance remain uneven. As such the strategic implication is that the real measure of African participation should not be how many delegates attend, but whether participation produces:
- common African negotiating positions;
- financing for regulatory and technical capacity;
- access to compute and scientific evidence;
- enforceable domestic accountability;
- and influence over standards before they become fixed.
For me the question keeps coming to mind now …. Africa Has a Seat at the AI Governance Table. Does it have the negotiating power?
3. Sovereignty is being reframed as bargaining power, not merely localisation
A useful new African Business analysis argues that digital sovereignty should not be reduced to keeping data physically inside national borders. Localisation may change where data is stored without changing who owns the cloud, controls the technical architecture, sets contractual conditions or captures the value. It proposes understanding sovereignty as the ability to increase African bargaining power in negotiations over infrastructure, standards and data governance.
The learning: This distinction is one to not put aside. You see, a data centre located in Lagos, Nairobi or Johannesburg may still rely on foreign capital, imported hardware, proprietary cloud systems and legal agreements governed elsewhere. Physical location does not automatically create public control.
Why it matters today: Many African AI strategies use the language of sovereignty without specifying its operational meaning. It is imperative now to actually do the hard work of distinguishing several layers:
- Data residency: Where information is physically stored.
- Legal sovereignty: Which laws and courts govern it.
- Technical sovereignty: Who can inspect, alter or migrate the system.
- Economic sovereignty: Who captures the resulting value.
- Institutional sovereignty: Whether public institutions have the capability to negotiate, regulate and enforce.
This can happen if Africa’s collective policies and actors can see now that a data center is not the same as Digital Sovereignty.
4. The environmental cost of “sovereign AI” is now a governance issue
A newly published paper from 15 July examines the water, energy and carbon implications of sovereign AI infrastructure in the Global South, including a Kenya-focused African case. It identifies a “sovereignty–sustainability trilemma”: governments may struggle to maximise national control, keep infrastructure affordable and minimise environmental harm simultaneously.
The authors recommend mandatory reporting of water-use efficiency, climate-sensitive data-centre siting and greater use of smaller, computationally efficient models rather than assuming every country should pursue frontier-scale model training.
(Note) This is still a preprint and we need to treat its quantitative modeling cautiously. Its conceptual contribution, however, is significant.
The learning: AI infrastructure policy is beginning to confront its relationship with energy systems, water security and climate vulnerability.
Why it matters for African governance today: The choice is not simply between foreign dependence and locally hosted infrastructure. Local infrastructure can itself impose public costs. A data centre may create investment and capacity while competing with communities and industries for power or water.
The question that comes to mind now is ….. what if the infrastructure intended to reduce Africa’s technological dependency deepens its ecological and energy dependency instead?
5. Infrastructure is already Africa’s dominant technology-policy concern—but delivery lags ambition
Carnegie’s Africa Technology Policy Tracker now contains more than 1,000 policy and legal data points. Its analysis found that digital infrastructure appears in 90 per cent of the documents tracked, while 57 per cent focus exclusively on infrastructure. The problem, therefore, is not that policymakers have failed to recognise infrastructure. It is that each new technological ambition is being added on top of unresolved connectivity, electricity, affordability and institutional-capacity deficits.
A separate June paper mapping the African AI divide similarly identifies infrastructure, accessibility and human capacity as interconnected constraints, including high connectivity costs, limited computing capacity, gendered access gaps and insufficient language representation.
The learning: As I have alluded in previous articles, the funding problem is not only that infrastructure receives insufficient attention. It is that policies accumulate faster than delivery capacity. Each new strategy creates additional requirements for the same overstretched energy systems, regulators, universities, procurement units and public budgets.
Question that comes to mind now is what if in fact Africa does not have a shortage of digital ambition, but rather a sequencing problem?
A strategic synthesis
Africa’s AI governance challenge is becoming a question of institutional and negotiating capacity: the ability to decide what infrastructure is built, under whose terms, at whose cost and with what mechanisms for public accountability.
Taken together, these new reading materials suggests that African AI governance must operate across three connected layers:
- Global negotiating power
Africa must influence global rules, compute-access arrangements, standards and infrastructure partnerships before they are determined by dominant states and companies.
- Continental and regional coordination
Fifty-five separate national approaches will have limited bargaining power against global cloud, semiconductor and model providers. Regional procurement, shared research infrastructure, common evaluation standards and coordinated data-governance positions may be more viable.
- Domestic institutional accountability
Continental strategies mean little without funded regulators, procurement expertise, impact assessments, independent research and mechanisms through which citizens can challenge automated decisions.
The risk is that global and continental strategies become highly sophisticated while domestic institutions remain unable to enforce them.
Taken together, these developments suggest that Africa’s AI future will not be determined simply by how quickly we adopt new technologies. It will be determined by whether we invest in the institutions, infrastructure and negotiating capacity needed to shape those technologies on our own terms.
The conversation is shifting. AI governance is no longer just about ethics or innovation. It is increasingly about political economy, public institutions, regional cooperation and long-term strategic investment. Those who recognise this shift early will be better positioned to influence not only how AI is used across the continent, but who benefits from it and who is ultimately accountable.
For me, that is perhaps the most important takeaway from this week’s reading. The question is no longer whether Africa should participate in the global AI conversation. It is whether we are building the capability to influence it.
These are exactly the kinds of questions we explore through Project by Projects (PxP).
We work with governments, philanthropies, public-interest organisations and mission-driven institutions to move beyond reacting to emerging trends and towards building strategies that are grounded in systems thinking, institutional capacity and long-term public value.
If your organisation is navigating AI governance, digital transformation, policy design or complex systems change, I’d love to explore how we can think through those challenges together.
Because good strategy doesn’t begin with the technology.
It begins with asking better questions.
