Insight, Energy Systems
Kenya generates about 93 per cent of its electricity from renewables and has some of the most expensive power in the region. A plan to more than triple generation does not address why. Meanwhile a company selling flexibility just raised US$150 million at a billion-dollar valuation.
Kenya has outlined plans to raise generation capacity from roughly 1.5 gigawatts to about 5.5 gigawatts, with substantial additional geothermal and hydro and a proposed two-gigawatt nuclear component (as reported). The country already produces around 93 per cent of its electricity from renewable sources, and power remains expensive because of transmission losses, financing costs, ageing infrastructure and contractual problems elsewhere in the system.
That sentence should be read twice, because it retires an argument that is still made constantly. Clean generation does not produce cheap electricity. Kenya is the proof, and it has been the proof for years.
The delivered cost of electricity is a stack of bands, and generation is one of them. In Kenya the other bands are large and visible in the monthly bill.
Fuel and foreign-exchange pass-throughs, on a 93 per cent renewable system.
For June 2026 meter readings, the regulator applied a fuel-energy cost charge of 314 Kenyan cents per kilowatt-hour and a forex-fluctuation charge of about 72 cents, having applied a fuel charge of 347 cents and a forex adjustment of 123 cents in April. A system that is overwhelmingly renewable is still passing through a fuel charge, because the marginal dispatchable plant is thermal and the contracts sit in dollars. Consumers were asked to absorb about 779 million shillings of forex losses in a single month, of which independent power producers accounted for roughly 85 per cent.
Capacity paid for and not used.
Kenya Power has been contractually obliged to pay for contracted capacity it does not need, curtailing cheaper renewable generation while continuing to pay expensive independent producers under take-or-pay provisions. This is the part worth sitting with. The utility pays for output it declines to take, and it declines to take cheaper output in order to honour the payment.
System losses.
Every percentage point of loss between the plant and the meter is, on one Kenyan analyst's estimate, a revenue reduction of about 1.5 billion shillings. Losses are recovered through the tariff up to a regulated allowance, which means the consumer pays for a defined quantity of waste.
Set the expansion plan against that stack and the question answers itself. More generation reduces the first band only, and only if the new plant displaces the marginal thermal unit rather than being curtailed alongside the existing renewables. Under take-or-pay contracting, additional capacity in a system that is already long can increase the bill rather than reduce it, because capacity payments accrue whether the plant runs or not.
A two-gigawatt nuclear component deserves its own note. It is a proposal with a long lead time and it may never be built, so it should not be over-read. But it is worth observing that adding a very large inflexible block to a system whose difficulties are inflexibility, weak transmission and contracted capacity it cannot absorb is a solution pointed at the one problem the system does not have.
Four cost bands sit inside a Kenyan kilowatt-hour. Additional generation reaches only one of them, and only partially.
Three of the four bands are contractual and institutional. None of them is fixed by a power station.
The genuinely consequential Kenyan energy news this year was not a capacity target. It was the Energy (Electricity Market, Bulk Supply and Open Access) Regulations 2026, gazetted in May, which dismantle the legacy single-buyer model and establish rules for forward contracts, spot markets and non-discriminatory open access. Alongside it, a revised currency framework permits future power purchase agreements to be denominated in shillings, in foreign currency, or in a hybrid of the two.
Those two changes act on three of the four bands directly. Shilling-denominated contracts address the forex pass-through. Competitive procurement and open access address the contractual rigidity that produces curtailment alongside capacity payments. A spot market creates something the system has never had, which is a price that moves.
That last one is the interesting part, and it connects to a different story entirely.
Emerald AI has raised around US$150 million in a Series A at a valuation of roughly US$1 billion (as reported), co-led by Energize Capital and DCVC with Nvidia, Siemens, GE Vernova and RWE participating. Its software allows AI data centres to reduce or shift electricity demand when grids are constrained, rather than sitting on the network as a permanently inflexible load.
Note who is on that cap table. Two grid-equipment manufacturers and a European utility. This is not a bet on artificial intelligence. It is a bet that flexibility becomes a traded, remunerated product.
Which points at the underlying problem, and it is not a Kenyan problem. It is close to universal.
Flexibility is valuable in almost every electricity system and remunerated in very few.
Revenue in a conventional power system attaches to two things: energy delivered, and capacity made available. There is no product called "not consuming at seven in the evening" that an ordinary consumer can sell to anybody. The value of that restraint is real, it is often larger than the value of an additional megawatt, and there is usually no contract capable of carrying it.
This is a missing market in the strict sense. The good exists, its value is demonstrable, and the institutional apparatus to trade it does not. And systems build what they can bill for, which is why the cheapest resource in most grids remains unbuilt while the most expensive one gets a capacity payment.
Kenya's demand side is unusually well suited to this, which is what makes the timing worth noticing. Irrigation pumping, cold storage, tea processing, water-utility pumping and, increasingly, battery-swapping stations are all loads with genuine scheduling tolerance. A swap station is close to an ideal flexible load: it is energy storage that has been installed for another purpose and has hours of slack in when it charges. There are now thousands of such sites appearing across the region, and nobody is paying any of them to move their consumption.
Demand response needs infrastructure the utility may not have.
Interval metering, communications, settlement systems and a market operator capable of clearing them. These are not trivial and they are not free, though they are considerably cheaper than a gigawatt of anything.
A financially distressed utility sees demand response as lost revenue.
This is the deeper obstacle and it is rarely stated. A utility whose tariff application has just been withdrawn by government, having sought increases of up to 31.8 per cent, is not primarily looking for ways to sell fewer units. Until the utility's revenue is decoupled from volume, demand-side resources are working against the interests of the institution that would have to enable them. That is a regulatory-design question, not a technology question.
Flexibility is worth less when the constraint is local.
If the binding limit is a specific transmission corridor rather than system-wide capacity, then flexibility only has value if it sits on the right side of the constraint. Location matters, and system-wide programmes tend to ignore it.
When a capacity expansion lands, the useful question is not how many gigawatts or how clean.
It is: which band of the delivered cost does this reduce, and by how much. If the answer is generation cost only, in a system where generation is already the smallest and cleanest part of the problem, then the plan is about capacity rather than about price, and it should be defended on that basis rather than as a route to affordable power.
Kenya has, this year, done the harder and less visible thing. Market reform, open access and contract flexibility act on the bands that a power station cannot reach. Whether they work will be determined by implementation over several years, which is the slow part and the part that receives no announcements.
The Lab works on this in energy access and across the markets where the gap between installed capacity and affordable, reliable supply is widest.
If you are financing generation in a system whose cost problem sits somewhere else, that is worth establishing before the plant is built.
This is an independent insight piece by Transitions Lab. For the Lab's applied work, see Energy Access & Off-Grid Systems. See also The Customers Who Can Leave on what happens when a price signal drives the largest customers off the grid entirely, Stacking, Not Switching on the take-or-pay bill that outlives the demand forecast when the invisible half of the grid appears in the accounts as a surprise, Who Pays Decides What Gets Built on the customer class now setting the direction of firm clean generation, and The Load That Grows When It Is Hot on the missing flexibility market for a load that arrives on the same afternoon the rest of the system peaks. To discuss a study, see Contact.