AI and the Environment: Kenya pushes global accountability for data centre emissions
The country’s resolution calls for global standards to account for the carbon and water footprint of AI infrastructure, which is increasingly borne by African nations.
At the seventh session of the United Nations Environment Assembly, Kenya submitted a resolution calling for the world to account for the environmental costs of artificial intelligence infrastructure.
The move addresses what many observers view as a critical gap: while AI systems power global industries, the carbon and water footprints of data centres are increasingly borne by countries in Africa.
The alarming consumption
Training a single large AI model can consume as much electricity as thousands of homes use in a year. A single hyperscale data centre can consume hundreds of millions of litres of water annually.
"If AI is developed for global markets but the carbon and water footprint is left in the Global South, that is an environmental justice issue," said Environment Cabinet Secretary Deborah Barasa.
Kenya's resolution advocates for environmentally sustainable AI infrastructure across Africa, emphasising integrated renewable energy, reduced water consumption, and adherence to recognised environmental and social governance practices," she said.
"This resolution does not mean that Kenya is positioning itself as hostile to AI development. The country is actively attracting investment and building infrastructure as an emerging hub for artificial intelligence and cloud computing in Africa," she added.
"Kenya is proving to be one of Africa's next big AI and cloud computing hubs. We have a vibrant startup ecosystem and major tech companies from USA investing heavily in Kenya," said Mark Ouma, a Kenyan tech entrepreneur based in the United States and founder of Navismat AI.
The Konza Technopolis and other tech clusters are emerging as centres for data centres, tech startups, and AI research.
Kenya's renewable energy profile distinguishes it from many countries that are expanding data centre infrastructure elsewhere. More than 80 per cent of the country's electricity is sourced from geothermal, wind, solar, and hydro power. Ouma noted that this foundation provides an opportunity for a different model.
"Kenya's AI and tech growth is being driven by investment from both local and international players, drawn by our stable regulatory environment and a vibrant, young, tech-savvy population," Ouma said. "With over 80% of our electricity coming from geothermal, wind, solar, and hydro, Kenya has a strong foundation to expand AI with a low carbon footprint."
At the same time, Kenya faces a central challenge: scaling data centres sustainably without straining the environment.
Data centres account for approximately 2 per cent to 3 per cent of global electricity consumption. A single AI grid can draw approximately 1,050 megawatts. As Kenya's tech sector expands, these demands will intensify.
The environmental impacts are multifaceted. High energy demand places pressure on local power grids. When that energy is sourced from fossil fuels, it generates carbon emissions and drives climate change.
Water scarcity is another major concern. Data centres require large volumes of water for cooling.
"This puts additional stress on already resource-limited regions such as arid areas, intensifying competition for scarce water resources," Ouma said.
Kenya's green advantage
The expansion of AI infrastructure also generates other environmental consequences. The rapid pace of hardware upgrades creates significant electronic waste, which, if not properly managed, can contaminate soil and water.
Large-scale facility construction can disrupt local ecosystems and contribute to habitat loss. The extraction of rare earth metals and minerals required for AI hardware carries ecological and social consequences.
Ouma outlined a structured approach to addressing these challenges. "Any AI expansion in Kenya must be environmentally sustainable, ensuring the technology is not just green in practice but 'green by design.'"
The first phase, he says, needs to focus on establishing clear regulatory standards aimed at preventing inefficient or unsustainable infrastructure from entering the market. These regulations would set standards for energy use, cooling efficiency, and water management for all new AI and data centre projects.
"For existing data centres, I estimate it could take two to five years to upgrade operations and meet the required sustainability standards," Ouma said. This timeline would give operators time to upgrade systems, adopt renewable energy, and improve operational efficiency without abrupt compliance pressures.
The second phase would involve progressively ambitious renewable energy targets, with regular intervals, potentially every 18 months, used to monitor progress toward fully clean operations.
"This phased approach balances the need for immediate action with the practical realities of infrastructure upgrades," Ouma explained. "It allows Kenya to gradually transition its AI ecosystem to a low-carbon, energy-efficient model without disrupting growth or technological advancement."
While large technology companies possess the financial and technical resources to invest in green infrastructure, smaller startups face different constraints. The upfront costs of adopting renewable energy, upgrading cloud systems, or installing energy-efficient hardware can be substantial.
"Whether a company is big or small, access to affordable, renewable energy and technical support is essential," Ouma says. "This ensures that no one is left behind in the green AI transition."
The Ministry of Environment and Climate Change emphasised that Kenya's initiative targets environmental accountability rather than broad AI regulation. "What we are doing is putting the environmental impacts of AI on the global agenda, so that future AI governance does not ignore water, energy and climate," the Ministry said.
The resolution aims to ensure that environmental considerations are embedded into how AI infrastructure is built before international governance frameworks are finalised.
"Our resolution is not about regulating AI in general," the Ministry clarified. "Technical standards are already being negotiated elsewhere in the UN system."
The Ministry called for science-based transparency and rigorous environmental assessments to guide decisions about the location and development of AI facilities. "We need science-based transparency, environmental assessment and international cooperation now," the Ministry said, "before AI infrastructure locks us into unsustainable patterns for decades to come."
The government is urging that environmental metrics be integrated into the work of established technical bodies such as the International Telecommunication Union (ITU) and International Organization for Standardization (ISO). Policy discussions are considering how new facilities could run on clean energy, improve efficiency, recycle water, and reuse waste heat. At the moment, Kenya is engaged in bilateral discussions to sway the other delegates to adopt this resolution.
Follow our WhatsApp channel for breaking news updates and more stories like this.
