princessadentan/ai-market-formation-africa
Domaine:
digital infrastructureCréateur:
pri
Hôte:
Notes from moderating a panel on compute, talent, and policy as the real bottlenecks to AI market formation in Africa.
# AI Market Formation in Africa
### Owning the AI Transition: Compute, Talent & Market Formation
---
On March 28, 2026, I moderated a panel at Harvard Business School's 28th Annual Africa
Business Conference. The session was titled "Owning the AI Transition: Compute, Talent &
Market Formation in Africa." I was joined by Emmanuel Lubanzadio, Africa Lead at OpenAI;
Kennith Jackson, SVP AI Solutions at Andela; and Doha Ammour, VP International Business
Development at N+ONE Data Centers.
The conference theme was "Owning Tomorrow: Africa's Agency in a Shifting World." Last
year's panel focused on AI at the application layer. This year I wanted to ask a harder
question: what actually needs to be in place for early AI wins to translate into real
market formation and sustained economic impact? If 2025 was the year of widespread
consumer and enterprise adoption, 2026 is when the economic consequences of real scale
are becoming clear — who captures value, and which companies and countries benefit.
So we talked about the unsexy but essential enablers: compute infrastructure, talent
systems, and policy frameworks.
---
## How the session was structured
**Act 1: Ground reality.** Where is demand today and where are the friction points? Are
the use cases people cite actually in production — with budgets and business owners behind
them — or still in experimentation?
**Act 2: Structural bottlenecks.** What is preventing scale across compute, talent, and
policy? Not the high-level answer — the specific thing that breaks first.
**Act 3: Debate and future bets.** Where is the narrative wrong? What does Africa
actually need to own to matter in AI over the next decade?
---
## What the conversation revealed
I went in expecting the sharpest disagreements to be about compute and policy, but there was an interesting convergence around talent at a bottleneck.
**The bottleneck is human infrastructure, not the model.**
All three panelists, from distinct positions on the stack, …