Artificial intelligence (AI) is no longer a distant frontier it’s embedded in the rhythm of Kenya’s innovation ecosystem. Platforms like ChatGPT have become part of daily workflows for students, creatives, and businesses alike. However, as the adoption curve steepens, a vital question lingers in the background: what is the environmental cost of this progress?
Globally, AI systems demand massive volumes of electricity and water to function from the energy required to power data centres to the water used to keep servers cool. For a country like Kenya, where digital aspirations and climate vulnerabilities coexist, this reality cannot be ignored.
Research by the Electric Power Research Institute shows that each ChatGPT prompt consumes up to 10 times more electricity than a standard Google search.
While seemingly minor on a case-by-case basis, the cumulative effect, especially in a nation rapidly scaling its digital footprint is significant.
Kenya benefits from a cleaner energy mix more than most African peers, largely due to geothermal and hydropower.
But this doesn’t mean we’re immune. Intermittent supply and underdeveloped infrastructure still force data centres to fall back on diesel generators, undermining our net-zero ambitions and exposing communities to pollution and higher costs.
Less visible but just as critical is AI’s water footprint. A 2024 study, A Water Efficiency Dataset for African Data Centres, highlights that generating a single 10-page report using GPT-4 consumes up to 60 litres of water, depending on the system and the energy source used. Even though Kenyan locations may perform better than the global average in water efficiency, this is a wake-up call.
In drought-prone areas like eastern Kenya, every litre matters. Diverting water from essential services to cool digital infrastructure raises moral and practical questions we must address now not later.
We must choose wisely. With our renewable energy advantage, we should lead in building data centres that commit to 100 percent clean energy sourcing, paired with robust transparency on energy and water use. Public-private incentives can fast-track this transition.
As we expect manufacturers to disclose emissions, the same principle should apply to tech infrastructure. The National Environment Management Authority and technology stakeholders can pioneer environmental, social, and governance reporting protocols specific to AI and data centre operations.
The promise of AI is undeniable. However, its success will not be measured by adoption alone it will be judged by how responsibly we scale it.
The future isn’t just intelligent it must also be sustainable.
As we harness this new era of intelligence, we must also protect the resources that make life possible. Let us not only build smart systems, but build wisely with the environment, equity, and future generations in mind.
Emerging programs that can understand and generate human-like text or locally trained AI systems often have a smaller environmental footprint. Kenya should invest in “climate-fit” AI technology built with regional energy realities in mind.
Just as we’ve shifted public perception around plastics and emissions, we need a similar mindset shift on digital consumption.
Educational institutions, corporates, and government agencies can lead by adopting responsible AI use policies and raising awareness of the environmental implications.
Africa, and Kenya in particular, has the opportunity to leapfrog into a climate-smart AI future. By embedding sustainability principles into AI’s growth now, we can avoid replicating the high-emission, high-extraction patterns of more industrialised nations.
This is more than a policy conversation, it’s a moment to lead. Kenya can set an example by demonstrating that technological advancement and environmental stewardship are not mutually exclusive with the firm understanding that this is a cost-trade off approach.
The writer is an Account Director at Edelman Africa (Corporate Communications, Marketing, ESG and Climate Consultant)
