Elephant in the room here -> There's no money to be made there (yet)
Check out the prices from vendors. Also, check out the profits. Also the margin! from the ISPs. Take into account the Capex too. Everything will seem odd, except the volume of data consumed per user [which is related, or better said limited to the maximum actual bandwidth]
Safaricom which is well known for its MPESA platform is the most profitable firm in sub saharan africa [not south africa or nearby] and it's all because of data.
Datacenters are difficult:
- Higher cost of electricity
- Unreliable electricity (blackouts and under-rated voltage)
- Costs of acquiring equipment/servers
- Heating/Cooling is on the harder side
- Lack of competition means pricing is farther up on the higher end
- Back down to cost of data, which is really expensive per GB or to lease the fiber required.
Another thing is there is a large latency chunk added as the vast majority of consumers use 3G which adds another chunk of latency in. This is slightly disincentivising to add in local datacenters.
It may be worth noting AWS is actually run in great part from Cape Town. Everyone uses their data centers in Virginia yet does anyone wonder why AWS doesn't offer anything in South Africa when it's being built here?
If I had a couple billion dollars to blow, I'd be spending it on that. Though, maybe that's why I don't have a couple billion to blow =)
The profits from serving ads for consumer goods will be pretty slim though.
This glosses over the fact that your service even can geo-scale well over high latency and high jitter networks.
Like any business, we'll get the low hanging fruit first, then move on to the higher, harder to reach branches. We'll improve Internet service in Africa, but not before the time is right.