What AMACO’s $1.5bn Kenya AI project can teach Africa about energy

AMACO, a Greek multinational company, is proposing the HERCULES project in Kenya as a solution to the challenges of powering AI data centers in Africa. The project aims to integrate electricity generation, cooling, and AI data center infrastructure to reduce dependence on the national grid. AMACO's founder and CEO, Theodore Theodoropoulos, is in Nairobi to discuss the proposal with Kenyan officials and potential partners.

The HERCULES project initially relies on liquefied natural gas (LNG) but can later accommodate renewable energy and hydrogen. The design offers a highly mobile, scalable, adaptive, and energy-efficient platform that can be deployed independently of the conventional electricity grid.

Africa faces significant challenges in meeting the growing electricity demands of AI data centers. In 2024, global data center electricity consumption was about 415 terawatt-hours, and it is projected to reach 945TWh by 2030, with AI being the main driver of this increase. Large AI data centers can require 100MW or more of electricity, equivalent to the power supply for about 100,000 households.

AMACO's approach highlights five key lessons for Africa:

1. Build the power with the data center: Designing AI data centers with dedicated power generation is crucial. Large data centers require significant amounts of electricity, and integrating power generation directly with the data center can avoid reliance on national grids and allow for capacity expansion as computing demand grows.

2. Consider the grid constraints: Data centers can be operational within two or three years, while the required electricity infrastructure can take considerably longer to plan and build. Addressing grid constraints is essential to avoid delays in data center project implementations.

3. Do not assume the grid can absorb everything: Large data centers can consume a substantial portion of a country's electricity demand. For example, a 100MW AI data center in Kenya would account for about 4% of the country's current national peak demand. African countries should be cautious in treating hyperscale facilities as ordinary commercial entities and explore dedicated generation solutions.

4. Address cooling systems early: Cooling requirements increase as AI chips pack more computing power into server racks, leading to higher electricity and heat generation. AMACO proposes recovering heat and cold within the same system used to generate electricity, which can also help avoid pressure on local water resources. Cooling technology choices, such as closed-loop liquid cooling systems, can reduce direct water consumption. These factors should be established before project approval.

5. Plan for water usage: Data centers have varying water consumption depending on their cooling technology, location, and design. Establishing water consumption per kilowatt-hour of computing and identifying water sources is essential to ensure the sustainability of the project, especially in regions prone to water shortages, like Mombasa.