Kazakhstan’s electricity demand is growing fast, up 16% in the last five years, and the country has been a net importer of electricity since 2022. The government aims to achieve self-sufficiency by 2027 and potentially become a net exporter by 2029. Replacing ageing infrastructure adds to the challenge. Coal power supplied 62% of electricity generation in 2025, but the country’s coal units are on average over 50 years old, leading to frequent equipment failures. There were 891 incidents of various severity between January and July 2026.
Wind can help meet this challenge, but not as a standalone source. In a hybrid plant, wind and solar complement each other behind a single grid connection, while a small battery helps manage shortfalls. Deliberately oversizing the capacity means there is enough generation even on less windy and sunny days even with a small battery. The result is a plant with a capacity factor comparable to that of a baseload power station, at a lower cost than new coal. This report tests that proposition with hourly simulations at 15 sites across the country.
New onshore wind farms in Kazakhstan achieved a 39% capacity factor in 2025, close to the 41% global average for offshore wind. Average wind speeds exceed 7 m/s across two-thirds of the country, compared with less than a third of Germany’s territory. Yet Kazakhstan has only 2.1 GW of installed wind capacity as of mid-2026. Wind supplied 4.4% of generation in 2025, around half the global average of 8.6% and well below wind leaders like Germany (27%). Despite its late start, Kazakhstan could catch up fast: wind’s share of generation has quintupled in five years, from less than 1% in 2020.
At each site, the best-performing of 125 configurations supplies 70–90% of the electricity in a typical demand profile, hour by hour, over 30 years, at a delivered cost below $60/MWh. This cost represents the plant’s lifetime cost per MWh actually delivered, including the cost of curtailed output. The same plants have a generation cost of $41–49/MWh, in line with Kazakhstan’s recent wind auctions and well below the $70-170/MWh of new coal.
Winter (December to February) is the most challenging season at most sites, as solar output is low and demand is at its highest. Even so, the modelled hybrid plants meet over 70% of demand at 13 of the 15 sites. A single plant has shortfalls, mostly on winter evenings and before dawn, when it delivers less than a fifth of demand. A flexible, interconnected system can fill these gaps far more cheaply than overbuilding each plant.
Wind is already growing fast and leading progress towards Kazakhstan’s national target of 15% renewables by 2030, but without an ambitious longer-term target, progress risks stalling beyond that milestone. By 2035, wind and solar could supply 23% of generation, up from 6% in 2025, bringing total renewables to 35% of generation while maintaining reliability and affordability.
Achieving this will depend on greater flexibility and a market that pays for flexibility, not just energy. Hybrid plants are already competitive with new coal on energy revenue alone. Their surplus generation and batteries can also be a source of flexibility. If batteries can earn revenue from balancing, capacity and grid services instead of only shifting their own plant’s output, they can deliver their full value to the system and further boost projects’ economic competitiveness.
Kazakhstan has a vast untapped high-quality wind resource that with some solar and a small battery can reach similar reliability to what solar and more batteries can do in sunny places at an affordable cost. And the best part is – wind resources are great across most of its territory, so hybrid projects can be built close to demand, minimising major additional transmission lines.
Kazakhstan’s electricity demand rose by 16% over the last five years – increasing by 17 TWh from 107.3 TWh in 2020 to 124.6 TWh in 2025. That increase is comparable to Tajikistan’s or Kyrgyzstan’s total electricity demand in 2025 (19 TWh each). The additional demand was primarily met by increased gas (+11 TWh), wind (+4 TWh) and electricity imports (+2 TWh), according to Ember data, with Kazakhstan becoming a net importer in 2022. Coal generation remained the dominant source of electricity, providing 62% of electricity in 2025. However, the capacity-weighted average age of Kazakhstan’s coal fleet is over 50 years, well beyond the standard 40-year lifetime for a coal power plant. This leads to frequent outages and emergency shutdowns. Between January and July 2026, 891 incidents of varying severity were recorded.
The government is aiming to achieve self-sufficiency by 2027 and potentially become a net exporter by 2029. To replace its ageing infrastructure and keep pace with fast-growing demand, Kazakhstan needs to build more capacity quickly and ensure its availability during peak periods, including winter demand spikes. The capacity also needs to be built close to where demand is to relieve pressure on the transmission network. Kazakhstan already aims to build 6,659 kilometres of new high-voltage lines by 2035 to better interconnect its three largely isolated power zones – North, South and West. This includes connecting the West zone to the rest of the national power system and increasing the limited transmission capacity between the coal-rich North and chronically undersupplied South zone. Building new capacity closer to demand would limit how much more grid needs to be built.
Kazakhstan has some of the world’s best wind conditions for power generation, with onshore capacity factors comparable to those of offshore installations elsewhere. Although the windiest parts of the country do not have the world’s highest wind speeds , the quality and speed of this resource is exceptionally homogenous across Kazakhstan’s vast territory. This means capacity can be spread out and allocated closer to demand centres and existing power lines without sacrificing output. It also makes the country less vulnerable to concurrent weather fluctuations.
Based on IRENA data for the 30 countries with the largest additions to onshore wind capacity, the weighted-average capacity factor of new onshore wind projects in Kazakhstan was 39% in 2025. This is below the figures for the top three leading countries in the dataset: Brazil (50%), Egypt (49%) and Peru (47%). However, it was almost 20% higher than the world average (33%) for onshore wind and just slightly behind the world average for new offshore wind (41% in 2025).
Despite this strong potential, wind remains underused. Wind accounted for just 4.4% of Kazakhstan’s total generation in 2025 – around half the global average (8.6%) and behind leading countries like the United Kingdom (29%), Finland (27%) and Germany (27%), all of which have lower average capacity factors than Kazakhstan.
Kazakhstan does not have the highest wind speeds in the world. The windiest 2% of Kazakhstan’s territory has average wind speeds of 8.9 m/s at 100 m. This area covers around 54,000 km², theoretically enough space for 270–430 GW of wind capacity at 5-8 MW/km². That is equivalent to roughly one-fifth to almost one-third (31%) of the 1.4 TW capacity humanity has built as of June 2026.
Kazakhstan’s geographical advantage lies not in having the world’s highest wind speeds, but in the breadth of its wind resource. High-quality wind is available in nearly every region rather than being concentrated on a handful of corridors. So unlike most countries in the world, Kazakhstan does not have to build wind in a few remote areas with high wind speeds and then build long transmission lines to bring it to where demand is. Two-thirds of Kazakhstan’s territory has mean wind speeds above 7 m/s at 100 m – well within the 7 m/s to 10 m/s range generally considered excellent for onshore wind.
In Germany, where wind accounted for 27% of all power generation in 2025, less than a third (32%) of its territory has wind speeds above 7 m/s. This means that commercially viable wind can be built close to most demand centres, without adding to Kazakhstan’s already large need for new transmission. The country still needs a strong grid, but may be able to avoid the cost and lead time of long new lines to remote sites with strong winds and little demand.
Ember analysed 15 sites across all three power zones of Kazakhstan. The sites were selected for their proximity to a major demand centre and an existing power line, rather than because they had the country’s strongest winds (see Methodology).
At each site, Ember tested 125 combinations of wind, solar and battery capacity, producing a total of 1,875 simulations. Each configuration was simulated hour by hour over a 30-year project life using site specific weather data and a typical hourly demand profile from the Agora Energiewende PyPSA-SPICE dataset. The profile represents national demand patterns and was scaled to a peak of 100 MW. Each site was tested against the same demand profile, so differences in results reflect weather conditions rather than differences in load.
At a delivered cost below $60/MWh, the maximum share of the typical demand profile met at each site ranged from 70% at the weakest site (near Kordai in the South) to 90% (near Zhanaozen in the West). These figures are averages across the 30-year project life; individual years perform above and below that average depending on the weather and to a smaller extent, equipment ageing.
Across all 15 sites, the average delivered electricity cost is $56/MWh – calculated by dividing the plant’s full lifetime cost by the electricity actually delivered to the demand profile. This is significantly cheaper compared to new coal, which costs $70-170/MWh, depending on the region and technology.
The share of demand met can vary by up to 4.5 percentage points between windier and calmer years, with the highest variation at southern sites.
At Kordai, for example, the share rises from 70% using the calmer 2022 weather data to 74% using the windier 2023 data. Sites in the North, East and West are steadier, varying by about 1–2 percentage points, while the Dzungarian Gate (site in the South zone) barely changes at all. Equipment degradation plays a small role – excluding weather variation, the share of demand met in the final three years is around 2 percentage points lower than in the first three.
The configurations that meet the highest share of demand for less than $60/MWh deliberately build more wind and solar capacity than can be absorbed by the demand profile every hour. Across the 15 sites, an average of 23% of annual generation is curtailed (15–30% depending on the site), mainly in spring and summer when solar is strongest. The curtailment in these results is a design choice and is hence fundamentally different from grid-ordered curtailment without compensation.
Its cost is already reflected in the delivered price: average generation cost is $45/MWh, compared to the delivered cost of $56/MWh. The $11/MWh difference represents the cost of curtailment which helps make the plants flexible enough to meet a high share of demand. Curtailed electricity could also support more flexible demand, including from new sources like electric vehicles (EVs) with smart charging, or be exported to neighbouring countries.
In summer (June-August), these configurations meet 78–95% of demand. Winter (December to February) is the hardest season at most sites, as solar output is low and demand is at its highest. Even so, the plants meet 61–87% of winter demand, and over 70% at 13 of the 15 sites, thanks to strong wind in winter. The exceptions are in the South, near Kordai and Shu. At Kordai, the weakest site in winter, the plant meets 61% of winter demand, with wind directly supplying 44 percentage points, solar 15 percentage points and the battery, 2 percentage points.
Deep shortfalls, defined as hours when the plant meets less than 20% of demand, occur between 4.5% of the time in the best-performing site (Zhanaozen) and 15.6% in the worst (Kordai). They occur on 90–249 days per year but tend to be relatively short, lasting 4–7 hours in winter and 2–4 hours in summer.
Filling these hours at plant level would mean overbuilding wind, solar and storage and higher cost of delivered electricity, as discussed in the next section. Connected to a flexible grid, which Kazakhstan has yet to build, the same gaps shrink: when the 15 simulated sites are considered together, the number of hours in which less than 20% of demand is met, to about one hour a year, because calm spells rarely cover the whole country at once. Section 2.2 further sets out what this means at the system level.
Higher shares, even 100%, are technically possible with extremely high overbuild, but moving closer to 100% gets progressively more expensive. Each additional percentage point above 70% requires more and more wind, solar and storage, and much of their output is curtailed, which raises the cost of every MWh delivered.
The simulations show this clearly. At each site, the cheapest configurations cost $40–48/MWh with almost no curtailed output but meet only 50–61% of demand. The best configurations below $60/MWh meet 70–90%. Beyond that, costs climb steeply. The largest configuration tested has 180 MW of wind, 250 MW of solar and 480 MWh of battery. It meets 91–97% of demand, depending on the site but at $87–91/MWh, around 50% more than the $60/MWh configurations. Each extra MWh of electricity gained by moving from the best $60/MWh configuration to the largest configuration costs $205–430/MWh.
Coverage over 70% is possible even at Kordai, the weakest site: but it costs $69/MWh to meet 80% of demand and $86/MWh to meet 90%. At the best sites, Zhanaozen and the Dzungarian Gate, 90% is within reach at around $60/MWh.
At system level, a single project does not need to meet all demand with 100% reliability on its own. The modelled wind hybrid plants provide cheap and abundant energy that already covers 70-90% of a typical demand profile for less than $60/MWh, but in reality they do not exist as isolated demand islands. The remaining 10-30% of demand that a single hybrid plant does not deliver can be provided more efficiently by an integrated power system with diverse generation sources, storage, transmission and demand-side flexibility Currently, Kazakhstan’s power system has limited flexibility due to both institutional and structural factors. Building a more flexible power system and making flexible hybrid plants a core part of it is a key task for the coming decade.
Spreading plants across the country does not eliminate the need for back-up power generation, but it reduces the amount needed because weather fluctuations are less likely to affect all sites at once. To cover all but the worst 1% of hours (about 90 hours a year), a single plant needs back-up capacity equal to 81–91% of peak demand; all 15 plants together need 48%.
The economically optimal level of wind and solar share at a system level, accounting for all sources of flexibility including other generation sources, demand side flexibility and interconnections, requires a much broader analysis beyond the scope of this report. A recent study has shown wind and solar could reach 23% share of generation in Kazakhstan by 2035, while maintaining power system reliability and affordability.
In our simulations, the optimal battery is small and does most of its work by bridging shallower shortfalls, especially in summer evenings. The battery size is not fixed and depends on what is optimal for each site’s weather conditions. Across 14 of the 15 sites, it varies between 12% and 25% of the combined wind and solar capacity of the hybrid plant (37% at Zhanaozen). Each battery provides four hours of storage. The legal requirement for auction winners since 2025 is for a much larger 30% of capacity, but just two hours of storage, which may be suboptimal for a specific project.
Each battery is only used to store its own project’s wind and solar output for hours when generation falls short of meeting target demand. That leaves the battery underutilised. Across the 15 sites, batteries supply 3-7% of delivered electricity, cycling 120–250 times a year – similar to China’s co-located batteries, which averaged 199 cycles in 2025 and well below the general best practice of 350 cycles per year.
At the project level, adding more battery capacity is not economical, as they remain underutilised due to the large number of days without a large enough surplus to charge the battery in winter. Batteries that only shift their own plant’s output could sit idle for much of the year, so realising their full system value is key.
Batteries are a powerful multitool that can respond to a sudden loss of generation within a fraction of a second, helping maintain grid stability. The August 2026 Central Asian blackout demonstrated this in practice. The outage affected the power systems of Kazakhstan, Kyrgyzstan, Uzbekistan and Tajikistan. The outage cut power to Almaty and a large part of the South power zone in Kazakhstan. While parts of Uzbekistan were also affected, according to Uzbekistan’s Ministry of Energy, batteries helped prevent a full national blackout. Uzbekistan is a regional battery leader with over 1.5 GW of batteries installed as of 2025, or 13% of peak demand, and aims to reach 4.5GW by 2030. Meanwhile, Kazakhstan’s first battery system is expected to come online by 2029, alongside TotalEnergies’ Mirny wind farm.
While in sunny countries like India, solar and batteries can supply around the clock electricity economically, in Kazakhstan the vast untapped wind potential provides a unique opportunity.
Kazakhstan has a national target to reach 15% renewable electricity share by 2030. The target includes wind, solar, bioenergy and small hydropower, but excludes large hydropower plants. By that definition, renewables reached 7% of generation in 2025, up from 4.5% in 2022. Renewable generation rose 1.7 times over that period, beating the interim national target of 1.5 fold increase.
Wind is already leading the progress towards the target. It accounted for 4.7% of 2025 generation, according to Ember data, more than doubling from just 2% in 2022. Installed capacity has surpassed 2 GW as of Q2 2026, with over 5 GW in the pipeline, of which 4 GW is expected to come from four large wind-plus-battery projects, each with 1 GW wind and 600 MWh battery.
Costs are already competitive. Kazakhstan’s first auction for wind with storage, a 1 GW project in Kostanay region, cleared in April 2025 at 18.72 tenge/kWh ($42/MWh at the September 2026 exchange rate of about 450 tenge to 1 USD), and the negotiated dollar tariffs for the TotalEnergies and Masdar projects are $39.9/MWh (18.0 tenge/kWh) and $44.9/MWh (20.2 tenge/kWh). These prices are generation prices and comparable with the $41–49/MWh generation cost of the hybrid plants in this report.
Wind and solar are a source of abundant cheap electricity. Some curtailment is not only economically acceptable – it is the key to making hybrid power plants a source of flexibility, but only if it is by design and paid for. A fleet of hybrid plants spread across the three power zones adds flexibility of its own. When operated smartly, hybrid plants stop being a flexibility liability and become part of the solution: cheap power where demand is, with the flexibility to keep the lights on.
Interconnections extend the same flexibility benefits across borders that can create green energy corridors that unlock the renewable potential and strengthen energy security for the whole Central Asia and the Caucasus region.
Kazakhstan is already considering a law on energy storage systems. 2.4 GWh of batteries are already due to come online in the next few years at the four large wind-plus-battery projects. To ensure these batteries provide system-wide value and to incentivise future investment in smart hybrid projects, new rules need to be in place as soon as possible and ensure the possibility for revenue stacking. That means access to an ancillary-services market (frequency regulation, reserves, black start), access to the balancing market, and arbitrage under market rules rather than regulated charging and discharging tariffs. China’s regulatory journey from storage mandates to new market mechanisms that allow revenue stacking is a potential source of key learning for Kazakhstan. The mandates began in 2017 and required new wind and solar projects to add storage but co-located batteries were used little, cycling just 80 times a year in 2022. Since 2022, China has opened spot, ancillary-service and capacity payments to batteries, supporting a notable increase in utilisation and in February 2025 scrapped the mandates.
Kazakhstan does not have a dedicated renewables target beyond 2030. Instead, the 2050 target is for 50% renewable and other alternative sources, which include nuclear. The lack of a specific renewables target risks undermining progress once the 2030 target is reached. There are already valid concerns that integrating more variable renewables beyond the 15% by 2030 target will be challenging due to insufficient system flexibility. Kazakhstan’s coal plants have slow ramp rates and cannot run below about 70% of capacity, while other sources of flexibility like demand response and hydropower, remain untapped due to market rules that do not pay for flexibility. The grid operator, Kazakhstan Electricity Grid Operating Company (KEGOC), has argued for a pause on new renewable projects until flexible capacity catches up.
Increasing system flexibility while keeping up the pace of renewable deployment is the central task of the next decade. A clear and ambitious post-2030 wind and solar target would be a clear signal to investors that the pipeline does not stop at 15%.
Ember used Converge Hybrid’s optimisation software to evaluate the energy and cost performance of different hybrid power plant configurations in meeting electricity demand based on site-specific weather data, using hourly-level dispatch resolution across a 30-year project life. Each configuration combines wind, solar and battery and is simulated for every hour of the project life against an hourly demand profile, with a linear-programming (LP) dispatch model setting the battery’s hourly charge and discharge to match demand as closely as possible before charging surplus or curtailing. Two modes were used: optimisation sweeps, which run over 120 full 30-year simulations per site to identify the least-cost combination meeting a given reliability target; and individual simulations of a chosen design, which additionally produce the full energy, cost and financial analytics.
Data sources for simulations:
Selecting the 15 sites:
The simulations used wind and solar data for 15 sites across Kazakhstan’s North, South and West zones. Each site was selected based on proximity to existing demand centres and existing power infrastructure, rather than wind speeds. Then, the exact location of the site was moved to the windiest developable land within 20 km. The adjustment of the site did not take into account land use, protected areas, slope or road and crane access, so the corrected coordinates should be read as the starting point for a developer’s own siting, not as final turbine positions.
Costs are 2026 estimates adjusted for operational conditions in a cold climate.
Hourly simulations
For each site, 125 configurations were simulated (wind 36–180 MW, solar 50–250 MWac, battery 0–480 MWh in four-hour blocks). The hourly simulations were run over a thirty-year period. In each hour, available wind and solar first serve demand directly; the linear-programming dispatch then operates the battery to close the gap between generation and demand — charging when generation is in surplus and discharging when it falls short — co-optimised over a rolling multi-day horizon so that stored energy is used where it displaces the most unserved demand. Export is capped at the 100 MW grid connection, and any generation that can be neither delivered nor stored is curtailed. Dispatch assumes foresight of generation and demand over the planning horizon, so the results reflect a well-run plant rather than real-time forecast error.
Losses and degradation
Combined losses of 16.5% for wind (covering wake, availability, electrical and cold-climate effects) and 12% for solar are applied to gross output. The four-hour battery has a round-trip efficiency of 84.5% and is cycled across its full state-of-charge range.
Solar degradation is assumed at 0.5% a year, so the array produces about 85% of its initial output in year 30. The battery loses 2% of its capacity a year, ending the 30 years at roughly 70% of its rated energy, and is replaced once, in year 15.
Ember: Dave Jones, Rashmi Mishra, Nicolas Fulghum, Chelsea Bruce-Lockhart, Ufuk Alparslan, Ardhi Arsala Rahmani and Rini Sucahyo
Converge Hybrid: Charles Haworth, Mia Kovacic
We are grateful to Timur Shalabayev (Qazaq Green), Demir Kabylbayev (Eurasian Development Bank) and Tatiana Lanshina (Agora Energiewende), for reviewing an earlier draft. Their review does not imply endorsement of the findings, and any remaining errors are Ember’s own.
Wind turbines installed across a rural plain in Kazakhstan during sunset.
Credit: Ruslan Konev / Pexels
Ember is an energy think tank that aims to accelerate the clean energy transition with data and policy. Ember is the trading name of Ember Energy Research CIC, a Community Interest Company registered in England & Wales #06714443. 'Ember' is a trademark held at the United Kingdom and European Union Intellectual Property Offices. All content is released under a Creative Commons Attribution Licence (CC-BY-4.0). Website powered by 100% renewable electricity.