AI & African Enterprises: February 2024 Newsletter
Newsletter originally published February 26, 2024
🔋 Energy Implications of AI in Africa 🔋
This month, Nvidia's CEO announced a 265% increase in quarterly revenue (to $22b!) largely driven by demand from tech companies using their GPUs to run and train AI models. OpenAI's CEO, Sam Altman, approached US regulators and Middle East investors about raising trillions of dollars to build chips and generate power partly in an effort to decouple from Nvidia.
Much has been made of OpenAI's plans to manufacture chips, but less about their power generation initiative. Search driven by generative AI (like ChatGPT) uses 4-5x the energy of conventional web search. The energy implications are so profound, e.g. 2x increase in electricty demand by 2026 and increasing demand for water to cool data centres in US, that US Senator Edward Markey introduced legislation to investigate and measure the environmental impacts of AI.
The pace of AI adoption in Africa lags the States, but, as we covered Vol 2, Issue 7, increasing AI adoption in Africa could stress already delicate geopolitical dynamics and stretch thin resources.
Ethiopia presents an interesting live "case study" where the government invited Chinese bitcoin miners to establish operations in the country to attract foreign currency revenues in exchange for access to cheap electricity sourced from the recently completed Grand Renaissance Dam. The damn has been the root of contentious relations with neighbouring Egypt which relies on the River Nile for 90% of its water needs, and whose source is obstructed by the damn.
This month alone, Oracle opened their first African cloud data centre in Kenya, and Google in Johnannesburg reflecting increasing AI interest in Africa. South Africa has been experiencing a long-running energy crisis, that has resulted in scheduled "load shedding" to reduce pressure on the national electricity grid. Demand surges saw reports of blackouts and load shedding in Kenya in 2023. The infrastructure that supports running AI models poses the risk of excarberting existing issues of water scarcity and electricity access across Africa.
Takeaway: As AI adoption increases in Africa, the implications energy demands pose to a resource-constrained continent will become more pressing and require attention of public and private sector players.