Microsoft and G42's stalled data-centre project near Lake Naivasha exposes a difficult question for Kenya's digital ambitions: can a country build an AI economy without sacrificing the communities and ecosystems that already depend on its water?
Kenya wants to become one of Africa's leading technology and artificial-intelligence hubs.
That ambition is understandable.
The country has a sophisticated technology sector, strong digital-payment infrastructure, abundant renewable-energy resources and a strategic position in East Africa. A major Microsoft-G42 data-centre campus at Olkaria, near Lake Naivasha, was supposed to become a flagship of that ambition.
But there is a problem that cannot be solved with venture capital, cloud computing or artificial intelligence:
water.
The proposed project has stalled since May 2026 over power-capacity concerns. Yet the pause has brought another question into sharper focus: what would a massive new industrial facility mean for an already heavily pressured water system?
That question deserves far more attention than it is currently receiving.
The $1 Billion Promise
Microsoft and Abu Dhabi-based AI company G42 announced their Kenyan partnership in 2024 as part of a roughly $1 billion digital investment programme.
The proposed data-centre campus would be located in Olkaria, an area already famous for Kenya's geothermal-energy industry.
The logic was obvious.
Data centres consume enormous quantities of electricity. Olkaria provides access to geothermal power, allowing the companies to market the facility as a renewable-energy-powered digital infrastructure project.
Kenya's own Yearbook describes the planned campus as a green data centre designed to run on geothermal energy and incorporate high-tech water-conservation technology.
On paper, it sounds almost perfect:
AI + geothermal energy + foreign investment + jobs + digital infrastructure.
But the environmental equation is considerably more complicated.
Lake Naivasha Is Not Just a Lake
For people living around Naivasha, water isn't an abstract environmental issue.
It is an economic necessity.
Lake Naivasha and its wider basin support agriculture, horticulture, tourism, fishing, livestock and domestic water supplies.
Kenya's own environmental assessment describes Lake Naivasha as an exceptionally important freshwater resource in a semi-arid region. Its waters and associated aquifers are used for domestic consumption, livestock and irrigation.
The basin also supports one of Kenya's most important agricultural industries.
Flower farms around Naivasha produce for international markets.
Tourism depends on the lake.
Fishermen depend on it.
Pastoralists depend on water for livestock.
Communities depend on groundwater and surface water.
And geothermal development already creates industrial demand.
This is therefore not a case of introducing a data centre into an empty desert.
It is introducing another major potential water user into a complex and already contested water system.
The Most Important Number Is the Number We Don't Have
Here is the problem.
The proposed Microsoft-G42 project has publicly emphasised water conservation, but the companies have not disclosed a project-specific water-consumption figure.
That makes it difficult for communities and regulators to properly evaluate the project's potential impact.
And that is unacceptable for infrastructure of this scale.
If Kenya wants to approve a major AI facility near one of its most important freshwater ecosystems, the public should know:
- How much water will the facility require?
- Where will the water come from?
- How much will be consumed rather than recycled?
- What cooling technology will be used?
- Will groundwater be extracted?
- Will lake water be used?
- Will treated wastewater be used?
- How much water will be required during drought conditions?
- What happens when water demand from households, farmers and industry conflicts?
- Who gets priority?
These are not anti-technology questions.
They are basic questions of responsible infrastructure planning.
Naivasha Is Already Under Pressure
The concern isn't hypothetical.
Kenya's Marine and Fisheries Research Institute recently assessed the Lake Naivasha basin and identified excessive water abstraction, pollution, catchment degradation and sedimentation among the threats facing the ecosystem.
Nakuru County has similarly acknowledged that unsustainable water abstraction, rapid urbanisation, agricultural expansion and climate change are putting pressure on the basin.
And the lake's history demonstrates how sensitive its water balance can be.
A Kenyan environmental assessment records that between 1990 and 2023, Lake Naivasha's recorded level ranged from approximately 1,884.5 metres above sea level in 2010 to 1,891.5 metres in 2021. The 2009–2010 decline was associated with ecological consequences including increased nutrient concentrations, algal blooms and fish deaths.
This is an ecosystem capable of changing dramatically.
Climate Change Makes the Equation Even Harder
There is an interesting paradox unfolding around Naivasha.
The lake is not simply getting "drier."
In fact, satellite observations show that Lake Naivasha expanded substantially during the recent period of unusually heavy rainfall, flooding farmland, greenhouses and infrastructure around its shores.
That is precisely why simplistic arguments about water availability can be dangerous.
A lake can experience flooding in one period and severe water stress in another.
Climate variability makes water management harder because infrastructure has to survive both extremes.
A data centre designed around today's conditions could face very different realities ten or twenty years from now.
AI Doesn't Just Consume Electricity
This is something the public debate around artificial intelligence frequently overlooks.
We talk about AI's enormous appetite for computing power.
But computing infrastructure also requires cooling.
Servers generate substantial heat.
That heat has to go somewhere.
Depending on the cooling technology, data centres can require significant quantities of water.
The industry is therefore increasingly experimenting with alternatives including air cooling, closed-loop systems, immersion cooling, reclaimed water and other approaches designed to reduce freshwater consumption.
That is why the Microsoft-G42 project must be judged not simply by whether it promises "water conservation."
Kenya needs measurable commitments.
"Renewable" Doesn't Automatically Mean "Sustainable"
This is perhaps the most important lesson.
Geothermal energy is renewable.
That is good.
But renewable electricity does not automatically make every industrial project environmentally sustainable.
You can power a facility with clean energy and still put pressure on:
- water;
- land;
- biodiversity;
- waste systems;
- local infrastructure;
- communities.
Environmental sustainability is not a single metric.
A project cannot simply say:
"We use geothermal power, therefore we are green."
The entire resource chain has to be examined.
And the Local Communities Are Asking a Simple Question
What happens to them?
Residents around Olkaria have reportedly raised concerns about existing water access.
Some communities say they already depend on water vendors, while pastoralists describe animals travelling considerable distances to find water.
Whether every reported shortage can be attributed to industrial development is a separate question.
But the perception itself matters.
If communities believe multinational corporations are receiving privileged access to scarce resources while local residents struggle to secure water, the social licence for the project could collapse.
And no amount of corporate messaging will fix that.
This Is Where Kenya Needs Strong Regulation
Kenya already has institutions responsible for water and environmental management.
The Water Resources Authority regulates water abstraction and water-use permits.
The National Environment Management Authority oversees environmental assessment and environmental regulation.
The problem isn't necessarily the absence of institutions.
It is whether those institutions have enough information, independence and enforcement capacity to properly manage a project of this magnitude.
Before construction begins, Kenya should require a comprehensive assessment covering the cumulative water impact of the project alongside existing agriculture, geothermal operations, households and future industrial development.
One project may appear manageable in isolation.
Five projects can completely change the equation.
The Data Centre Should Not Be the Scapegoat
There is also an important argument on the other side.
It would be wrong to portray Microsoft and G42 as uniquely responsible for Naivasha's water problems.
Agriculture has enormous water requirements.
Urbanisation is increasing.
Existing industry consumes water.
Climate change is altering rainfall patterns.
Pollution is degrading water quality.
And decades of development have contributed to the basin's pressures.
The data centre would be one additional demand, not the sole cause of the problem.
That distinction matters.
The debate should therefore not be:
"Data centres versus communities."
It should be:
"How can Kenya allocate a finite resource fairly among competing needs?"
Kenya Should Demand a Water Budget Before a Data Centre
If Microsoft and G42 genuinely want to demonstrate that AI infrastructure can coexist with environmental sustainability, Kenya should require them to publish a detailed water budget.
Not marketing language.
Numbers.
The public should be able to see:
Annual freshwater withdrawal.
Annual water consumption.
Percentage recycled.
Source of water.
Peak drought demand.
Cooling technology.
Wastewater treatment.
Discharge standards.
Emergency shutdown conditions.
Long-term water-efficiency targets.
And critically:
What happens when the lake or groundwater system reaches a predetermined stress threshold?
If the answer is that the data centre gets priority over local communities, Kenya has a serious problem.
If the answer is that industrial consumption is automatically reduced under specified drought conditions, that is a much more responsible model.
Kenya Could Actually Turn This Into a Global Model
There is an opportunity hidden inside the controversy.
The world is building data centres at extraordinary speed because of AI.
Countries everywhere are confronting the same question:
Where should these facilities be located?
Cheap electricity is one factor.
Land is another.
Connectivity matters.
Political stability matters.
But water is becoming increasingly important.
If Kenya develops a model in which large data centres must demonstrate:
- renewable energy;
- water neutrality or extremely low freshwater use;
- wastewater recycling;
- transparent resource accounting;
- community participation;
- environmental monitoring;
- drought-response mechanisms;
then Naivasha could become a blueprint for sustainable digital infrastructure across Africa.
That would be far more valuable than simply hosting another data centre.
The AI Revolution Cannot Become a Water Revolution Against the Poor
There is a broader philosophical issue here.
Artificial intelligence promises enormous economic benefits.
It could improve healthcare.
Agriculture.
Education.
Financial services.
Scientific research.
Government.
Manufacturing.
Kenya should absolutely pursue those opportunities.
But technological progress becomes politically dangerous when its benefits are privatised while its environmental costs are socialised.
If a multinational corporation gets the economic benefits of a data centre while local communities bear the cost of depleted groundwater or degraded ecosystems, that is not sustainable development.
It is simply another version of the old resource-extraction model—only this time the commodity isn't oil.
It is computing power.
Africa Must Learn From Its Resource History
Africa has seen this story before.
Foreign companies arrive.
They bring investment.
Governments celebrate jobs and infrastructure.
The project generates enormous economic value.
But local communities eventually discover that they have surrendered something they cannot easily replace.
Sometimes it is land.
Sometimes minerals.
Sometimes forests.
Sometimes oil.
In the digital age, the resource could increasingly be energy and water.
Africa should not repeat the mistakes of the extractive economy simply because the new industry happens to involve artificial intelligence.
So, Can Naivasha Support Kenya's AI Ambitions?
Yes—but only if Kenya establishes the rules before the servers begin running.
The question isn't whether Kenya should welcome Microsoft, G42 or other technology investors.
It should.
The question is whether investment comes with sufficient safeguards.
The stalled project provides Kenya with an opportunity to get this right.
Before construction resumes, the government should demand complete disclosure of the facility's expected water requirements, conduct a transparent cumulative impact assessment and guarantee that domestic and ecological water needs cannot simply be pushed aside when industrial demand rises.
The lake has survived dramatic changes before.
But Kenya should not gamble with an ecosystem that supports communities, agriculture, wildlife and a major part of its economy.
Kenya's AI Future Should Not Be Built at the Bottom of a Dry Well
There is nothing contradictory about wanting both technological advancement and environmental protection.
In fact, Kenya's greatest opportunity is to prove that they can reinforce each other.
Microsoft and G42 have an opportunity to demonstrate that a next-generation AI data centre can operate with radically reduced freshwater consumption.
Kenya has an opportunity to demonstrate that African governments can negotiate technology investment without sacrificing local interests.
And Naivasha's communities have a right to demand something very simple:
Before anyone asks the lake to power the next AI revolution, prove that the lake can afford it.
Because the future of artificial intelligence may be measured in gigawatts and GPUs.
But for the people living around Naivasha, the future is still measured in something much more fundamental:
**whether there is enough water tomorrow.**
Comments
Post a Comment