A series of very large subsea cable projects have landed on African coasts in recent years, adding international capacity on a scale that would have been unimaginable a decade ago. The expectation was that retail bandwidth prices would fall accordingly. In many markets they have fallen far less than the capacity increase implies.
The bottleneck moved inland
A subsea cable delivers capacity to a landing station on the coast. Getting it from there to a customer in an inland city requires terrestrial fibre, and terrestrial fibre requires rights of way, trenching across other people's land, and protection from the road works and construction that routinely cut it. That middle-mile network is expensive, slow to build, and in several countries controlled by a small number of operators with little incentive to price it aggressively.
The result is a market with abundant international capacity and a constrained domestic distribution network — which prices like a constrained market, because that is what it is.
Where the traffic goes matters as much as how much
A significant share of traffic between two African countries has historically been routed via Europe, because that is where the peering relationships and the exchange points were. Every kilometre of that detour adds latency and cost. Internet exchange points and regional peering have improved this considerably where they have been established, and the improvement is one of the cheapest available: keeping local traffic local requires cooperation more than it requires capital.
What this means for anyone building
Latency to a user is decided by where your infrastructure sits and how traffic reaches it, not by the size of the cable offshore. Content delivery at the edge, regional cloud regions and local caching do more for a real user's experience than any headline capacity figure. And for products serving users on constrained connections, the discipline of keeping payloads small remains the highest-leverage engineering decision available.