In May this year the President William Ruto announced that the billion dollar data centre planned by Microsoft and G42 at Olkaria, Naivasha couldn’t proceed as designed, because turning it on would have meant switching off power to a large part of the country.
The first phase alone was to draw 100 megawatts from a grid whose peak demand already presses against its installed capacity.
Whatever one thinks of the decision, it told us something uncomfortable. The investment was announced, celebrated and timetabled before the most basic questions about what it would take from Kenya, and what Kenya would get in return, had been answered.
Technology is not new to us. It has been reorganising our economies, our work and our social lives for decades, from the mobile money revolution onwards. What is new is its ubiquity. Kenyans now talk to chatbots, borrow from apps whose algorithms score their creditworthiness in seconds, and file taxes through systems that decide, invisibly, who gets flagged for audit.
is a risk. All of this runs on physical infrastructure, on data centres and cloud services that consume land, water and electricity, and almost none of that infrastructure is ours.
Our data crosses borders to be stored and processed elsewhere, then sold back to us as services. If the coming decade of African life will be computed, the question of who owns and hosts the computation is a question of economic sovereignty, as consequential as the ownership of railways and ports was a century ago.
This is why I resist the notion of a foreign technology company arriving, building the infrastructure, collecting the incentives and owning the asset.
My research team has spent this year, under our Project TERRA work on technology, equality and regulatory risk assessment, mapping out data centres, their ownership, energy and water demands, and the tax treatment they enjoy to establish what has been given away, and to whom.
The pattern that emerges is familiar from the extractive industries. Host countries provide the land, the water, the power and the tax holidays. The value created from the data processed in these facilities is booked elsewhere.
A data centre can be an extraction site wearing the costume of development, and unless the fiscal terms are negotiated with open eyes, we will repeat with our data the history we lived with our minerals.
However, the answer is not to refuse the technology. It is to govern it before it hardens into infrastructure, and here Kenya already possesses an instrument it underuses, the regulatory sandbox. The Capital Markets Authority has operated one for financial products since 2019, and the Communications Authority published a framework for emerging technologies in 2023.
A sandbox allows the regulator and the innovator to test a product, and crucially to test the rules themselves, in a controlled environment before anything is rolled out at national scale. We should be doing precisely this for data infrastructure and for algorithmic systems. Before another data centre incentive is signed, its revenue cost, environmental burden and community benefit should be modelled and tested against evidence.
Before an algorithm is deployed in tax administration, credit scoring or social protection, it should be audited for bias inside a sandbox, with disclosure standards and a route to redress agreed in advance. This is how a country writes its own protocols for engaging technology rather than inheriting protocols written in California or Abu Dhabi.
Our legal architecture is not empty. The constitution demands public participation in policy, protects privacy, guarantees equality and requires openness and accountability in public finance. The Data Protection Act has been in force since 2019.
The National Artificial Intelligence Strategy 2025 to 2030 was launched last year, and the Artificial Intelligence Bill 2026 now before the Senate proposes a commissioner and a risk based classification of artificial intelligence systems.
These are necessary. None of them is sufficient, because they largely regulate the technology while remaining silent on the political economy beneath it, on who finances the infrastructure, who forgoes the revenue, who bears the environmental cost and whose interests the algorithms encode.
A commissioner who can audit an algorithm but cannot question the tax holiday granted to the data centre it runs on is governing half the problem.
Technology governance, properly understood, is fiscal governance. It asks the oldest questions of public finance, who pays, who benefits and who decides, about the newest machinery of our lives.
Kenya has the constitutional values, the regulatory precedents and the intellectual capacity to answer those questions for itself. What it mustn’t do is answer them after the concrete has been poured.
The Olkaria pause was an accident of power supply. The next pause should be deliberate, a country taking the time to write its own rules.
Dr Latif is a lecturer and-co-founder and co-chair of the Committee on Fiscal Studies at the University of Nairobi
Follow the story
