Near Kordai in southern Kazakhstan, a modeled wind, solar and battery plant could meet 80% of its target electricity demand for $69 per megawatt-hour delivered. Raising that share to 90% lifts the average cost to $86. The extra equipment is buying something harder than more electricity: electricity available when the first configuration falls short.
Those estimates come from an Ember report published November 3, 2026, not an operating plant or a household tariff. Its useful finding is that a hybrid plant can supply a large share of demand cheaply without being the cheapest way to supply every remaining hour.
Count the Electricity That Arrives When It Is Needed
Ember tested 125 combinations of wind turbines, solar panels and batteries at each of 15 sites. Each faced the same hourly demand pattern, scaled to a peak of 100 megawatts. This kept the demand side comparable while weather changed by location.
The best configurations below $60 per megawatt-hour covered 70–90% of that demand, depending on the site. The percentage counts demand energy met after matching supply and demand hour by hour. It does not mean every customer had uninterrupted power for 70–90% of the hours, or that a national grid had reached that renewable share.
The cost also has a specific denominator: electricity actually delivered to the modeled demand, rather than everything the plant generated. As with the broader story of cheap solar power and the grids that must absorb it, making inexpensive electricity and making it useful are related but different problems.
Some Unused Power Is Part of the Design
The plants deliberately have more generating capacity than the demand can absorb in favorable hours. That gives them more output when sunshine or wind is weaker. In the sub-$60 configurations, 15–30% of annual generation goes unused, depending on the site.
This is planned surplus, not evidence that a grid operator unexpectedly refused the power. Its cost is already included in the delivered-electricity estimate. Building a plant that wastes nothing in its easiest hours can leave too little output in harder ones.
Batteries shift some surplus into later shortfalls. But a larger battery does not create the energy needed to fill it. In this study, adding more storage can leave it underused when there are too few days with enough surplus to charge it.
The Difficult Hours Raise the Average Cost
Kordai makes the trade-off tangible: $69 per megawatt-hour at 80% coverage, $86 at 90%. These are average costs for all electricity delivered by each configuration, not the price of only the extra ten percentage points.
Across the sites, the largest configuration tested combined 180 megawatts of wind, 250 megawatts of solar and 480 megawatt-hours of battery storage. It covered 91–97% of modeled demand at $87–91 per megawatt-hour delivered. Those ranges span different locations; they are not one plant moving smoothly along a single cost curve.
Much of the added equipment produces more electricity in hours already well supplied. The remaining gaps increasingly occur at awkward times, including winter evenings and the hours before dawn. A high annual energy contribution can therefore coexist with a substantial need for backup power.
A Thirty-Year Simulation Uses Three Weather Years
The study is a design exercise with important limits. It repeats weather from 2022–2024 over a 30-year project life, so it does not test extreme years outside that sample. Battery dispatch assumes advance knowledge of generation and demand over a rolling planning period, rather than real-world forecast errors.
The published assumptions also do not provide a complete independently reproducible financing and system-cost model. The plant is assumed to sit at its connection point; additional transmission, backup and the wider grid cannot simply be treated as included in its delivered price. The report was co-authored by the CEO of Converge-Hybrid, whose optimization software it uses.
These limits matter most if the result is read as a guaranteed price for reliable national supply. It is not one. The narrower comparison remains useful: requiring one plant to meet more difficult hours changes the equipment it needs and the cost of the energy it delivers.
One Plant Does Not Have to Cover the Whole Grid
The report explicitly leaves the cheapest whole-system mix outside its scope. Transmission can connect places with different weather; other generation can fill gaps; storage can serve more than one project; and flexible demand can move some consumption. The value of that last option depends on timing, as our explanation of heat pumps and winter grid capacity shows.
The lesson is neither that cheap renewables guarantee cheap round-the-clock power nor that one plant’s expensive final increments make a renewable-heavy grid uneconomic. A plant supplies energy into a system. The next question is which combination of plants, connections, storage and flexible demand fills the remaining gaps at the lowest total cost, not how to force every wind farm to do every job.
