During the recently concluded MWC 2026 in Barcelona, Kenya’s Special Envoy for Technology to the United Nations, Philip Thigo, made a bold pitch to global investors: bring your artificial intelligence data centers to Kenya. Mr. Thigo warned that Africa risks falling behind in the next phase of the digital economy if it lacks computing infrastructure.
The premise is straightforward. Kenya is already a top destination for technology in Africa, and setting up localized AI infrastructure is seen as the ultimate silver bullet to winning the continental AI race against heavyweights like South Africa and Nigeria.
The pitch itself makes sense on paper. AI development depends heavily on massive computing power, and that computing power lives inside data centres packed with specialised hardware, storage systems, and high-performance networking. Without them, developers must rely on cloud infrastructure located overseas, which increases costs and limits local innovation.
The sentiment from Barcelona quickly resonated back home with Kiharu MP and major Kenya Power (KPLC) shareholder Ndindi Nyoro, who recently advocated for the liberalization of power production. Nyoro argued that to become a high-tech hub and accommodate the massive energy demands of data centers, Kenya must rethink its power production and strip away the entrenched patronage of Independent Power Producers (IPPs) and Power Purchase Agreements (PPAs) that have historically inflated costs.
While the government and political elite are eager to court Silicon Valley, the reception from Kenyans has been overwhelmingly skeptical. For a country still grappling with basic utility provision, the ambition to host hyperscale AI infrastructure feels wildly disconnected from reality.
The AI infrastructure reality
To understand the public’s frustration, you have to look at the sheer scale of resources required to keep the AI boom running.
Generative AI is notoriously resource-ravenous. Recent global statistics indicate that a single large-scale AI data center can consume up to 100 megawatts (MW) of power, enough to power over 100,000 homes. The servers running these operations generate immense heat, requiring aggressive cooling systems. Consequently, a hyperscale AI facility can guzzle between 1 million and 5 million gallons of fresh water every single day. To put that into perspective, an AI chat session involving just 20 to 50 prompts consumes roughly a 500ml bottle of fresh water.
Kenya has been making undeniable strides in the traditional data center space. We have seen Airtel Kenya break ground on East Africa’s largest data center at Tatu City, Konza Technopolis, pushing innovation spaces; and facilities like EcoCloud and Icolo turning to geothermal and solar energy to power their operations sustainably. But AI data centers are a different beast entirely, and the national grid is notoriously fragile.
Kenyans online have been quick to dissect the glaring holes in the government’s pitch, pointing out that jumping on the AI bandwagon without a concrete infrastructural strategy is a recipe for disaster.
The primary critique centers on the absolute basics: water and electricity. Critics were quick to point out the irony of inviting water-guzzling server farms to a country where Nairobi residents frequently endure days of water rationing, and Northern Kenya is actively battling severe droughts. The idea of millions of liters of water being diverted daily just to cool servers while the everyday mwananchi pays punitive water bills is deeply disenfranchising.
The power equation is equally contentious. The country currently suffers from the highest unit cost of electricity in East Africa. The grid is already operating near maximum capacity, plagued by nationwide blackouts and voltage fluctuations. So, if the state cannot provide cheap, reliable energy for a local welder in Githurai, hosting hyperscale data centers is nothing more than a pipe dream.
Environmental and Community nightmares
Beyond the utility bills, Kenyans must pay close attention to the global pushback against AI facilities such as the controversies surrounding Elon Musk’s xAI data center in Memphis.
There are chances of a logistical nightmare if these facilities are built near residential or agricultural areas. Because KPLC’s grid is prone to failure, a data center in Kenya would inevitably rely on massive industrial backup generators, leading to toxic diesel fumes and 24/7 deafening noise pollution. Furthermore, the loss of prime farmland to server farms, the overloading of local power grids leading to higher bills for everyone else, and even occupational hazards like deadly arc flash explosions are major concerns.
While Kenya’s positioning as a digital pioneer in Africa is commendable, the public consensus is clear. Until the country can figure out basic, affordable, and reliable electricity and water for its own citizens, the dream of hosting the next wave of AI data centers remains just that—a dream.
Still, the opportunity is real
Despite the skepticism, the broader argument for AI infrastructure isn’t without merit.
Local data centres could help reduce latency for African applications, lower cloud costs for developers, and allow organisations to process data within regional jurisdictions. They could also support the development of AI tools designed specifically for African languages, healthcare systems, agriculture, and education.
There is also the geopolitical angle.
Countries that control computing infrastructure tend to control innovation ecosystems. The global AI industry already reflects this reality, with the United States hosting nearly 40% of the world’s data centres. Kenya has only two AI-capable data centres compared to South Africa’s five and Nigeria’s one, which perhaps explains why South African cities lead the AI race in Africa.
For Kenya, attracting AI infrastructure would reinforce its ambition to remain one of Africa’s leading technology hubs. But until those basics are convincingly addressed, the idea of hosting power-hungry AI data centres will continue to trigger the same reaction.
