Gloom and doom dominates the coverage of AI’s ever growing energy demands. News outlets hit us daily with alarming headlines declaring that AI’s energy demand is insatiable, out of control, pushing the world into an energy crisis, and exhausting the power grid. This anxiety is not unjustified, as AI and related technologies are estimated to account for about 2% of global electricity demand and projected to double by 2026. Tech companies are also exacerbating the situation with their usual shenanigans, such as inflating AI’s clean energy and emissions credentials using questionable carbon accounting.
So while the gloomsters may have a point, the African perspective is frequently missing. And this matters because when one looks at the potential impact of AI’s energy demand on the continent, the results are surprising, counter-intuitive and even hopeful.
I recently had the opportunity to give my perspective on the BBC World Service program, The Inquiry, as one of four ‘expert witnesses’ weighing in on the question that everyone is currently obsessed with: “Do we have enough energy to power AI?” This proved to be an excellent opportunity to highlight some intriguing and counterintuitive aspects of the energy and AI landscape in Africa and other energy-scarce regions, where the power demands of AI present unique opportunities. The recording of the program is available on BBC Sounds, as well as on Spotify and wherever you get your podcasts. Listen from minute 16:08 to hear my segment, which stands alone well, but the whole episode (only 23:30 mins in total) is worth a listen.
My key points from the segment are summarized below:
Africa needs to increase its power demand: The continent's economy is significantly underpowered, with Europe generating ten times more electricity per capita than Africa. Power-hungry data centers, which are central to AI’s energy needs, could serve as valuable anchor customers. They would be a crucial source of stable demand and revenue for struggling power utilities, driving investment in power infrastructure, and enhancing power quality across the network with their stringent performance requirements.
Africa has the potential to provide clean energy for AI: Many African countries possess significant renewable energy resources, and some, like Kenya, already have very green grids, with nearly 90% of its electricity coming from renewable sources. This will make certain African countries attractive to tech companies striving to meet their emissions and clean energy goals, especially as they face increasing scrutiny regarding the climate impact of AI.
Demand for cloud services is growing on the continent: More users in Africa are coming online and businesses are integrating digital services into both their operations and core offerings. Local data center capacity is important to improve the quality of cloud services and reduce costs for this growing user base. Additionally, rising concerns about data sovereignty and privacy are also increasing pressure on major cloud service providers to expand their physical footprint or “cloud regions” globally. In Africa, major players like Amazon Web Services, Microsoft Azure, Google Cloud and Oracle Cloud currently have cloud regions only in South Africa. This growing demand is expected to drive substantial investment in the African data center sector, with billions of dollars anticipated to flow into it over the coming years.
A great example of the convergence of these factors is the recently announced partnership between Microsoft, G42, and the Government of Kenya to establish a large-scale data center and launch Microsoft’s first cloud region in East Africa. This project, hosted by the Kenya Electricity Generating Company (KenGen), will be powered entirely by geothermal energy from the Olkaria geothermal complex in Naivasha. Although the proposed 1 GW capacity has rightfully raised eyebrows — it would surpass Africa’s total current data center capacity, estimated at around 690 MW — a new hyperscale data center, even if it ends up smaller, along with a dedicated East African cloud region, will greatly benefit users in the region. Additionally, Oracle has announced plans to open a new cloud region in Kenya, signaling further investment in the sector.
The future of AI in Africa is obviously not without challenges. There are some serious barriers and concerns, including in the power sector, which I explore in greater detail in my recent article published in the journal Science. However, it’s refreshing to focus on the opportunities rather than just the threats for a change. The global tech commentariat tends to universalize the concerns of wealthy countries, and thereby miss opportunities in the rest of the world. This is why we need to amplify more geographically diverse viewpoints, especially on issues like AI.
As always, I welcome your thoughts and feedback.
Just when I’m certain that AI only poses extractive purposes on the continent, you come in with a glass half full. Thanks for the perspective-shift.
Great article Rose! In your opinion, do you think the major players in the data center space within Africa would largely be comprised of the global companies - Amazon, Microsoft etc? And what about telecom companies?
Thanks