by Kai-Hsin Hung
Globally, an upward of USD 5.2 trillion of capital expenditures are going towards AI infrastructure and compute investments by 2030. The usual story about AI in Africa is that the continent is a “compute desert,” holding less than 1% of the world’s computing power and being left behind. Our upcoming paper tells a different story. We built a dataset of 46 AI infrastructure investments announced across Africa between 2019 and 2025, worth $12.7 billion in total. The money is pouring into data centers and computing facilities, mostly in four countries: South Africa, Kenya, Nigeria, and Egypt. Africa is becoming a real and contested site of AI investment in giving us an indication of the changing compute landscape to come.
However, attracting investment is not the same as gaining control. We found that the catch is the “compute layer,” the cloud platforms and specialized chips that decide how AI is priced and accessed. It gets just 16% of investment, yet holds the real power, and is controlled almost entirely by a few US and Chinese firms like Microsoft, Huawei, and NVIDIA. A data center can be built and paid for inside Africa while still being owned and governed from outside. We call this “asymmetrical interdependence.” The lesson is that hosting a data center is not the same as controlling one.
What matters is not how much investment a country attracts, but whether its people have a real say in how that AI infrastructure is owned, priced, and used, a goal we call compute equity.
We are making the full dataset open so others can explore it for themselves. It covers all 46 projects, broken down by who invested systematically collected from 2019-2025, where the investment went, what was planned to be built, and which layers of the AI infrastructure value chains project largely sits in.
You can browse it, along with an interactive map (in Figure 1 below) showing where investment is concentrated across the continent, on Planetary AI and DataSpires (Market Intelligence). We hope it becomes a useful starting point for researchers, policymakers, and anyone trying to understand who is really building, and governing, AI in Africa and globally.
Figure 1: Regional AI Infrastructure Hubs, Secondary Nodes, and Regional Gateways and Connectivity Points
The link for the interactive map is given below:
Our dataset is also here:
https://docs.google.com/spreadsheets/d/1WZnxWc25O1r5-8N9OR3pktKMp4IWSjcU/edit?gid=1591105120#gid=1591105120
This paper is still a work in progress and currently under review. We welcome your comments you may have.
Authors of the paper:
Kai-Hsin Hung (corresponding author: [email protected])
Armita Sadeghian Barzoki
Sumaya Nur Adan
Krupa Suchak
Kofi Yeboah
Mohammad Amir Anwar