> ## Content Index
> Fetch the complete content index at: https://economist.kg/llms.txt
> Use this file to discover other available public pages before exploring further.

# Kilowatts and Toktogul: why Kazakhstan’s data center debate matters to Kyrgyzstan
- URL: https://economist.kg/en/kilowatts-and-toktogul-why-kazakhstans-data-center-debate-ma/
- Published: 2026-11-19T06:27:29.000Z
- Updated: 2026-11-19T06:27:29.000Z
- Description: Economist Kubatbek Rakhimov examines what Kazakhstan’s debate over cheap electricity for data centers means for Kyrgyzstan’s winter supplies, seasonal power exchanges and the operation of Toktogul Reservoir.
- Author: Economist.KG
- Tags: Мнение, Энергетика, #en

***Kubatbek Rakhimov*, *PhD in Economics, associate professor, head of the International Politics program at the Higher School of Diplomacy and Law within the Social Sciences and Humanities Institute of Kyrgyz-Russian Slavic University (KRSU); executive director of the Applikata Center for Strategic Solutions; co-chair of the Kyrgyz-Kazakh Business Council.***

Kazakhstan is debating the Data Center Valley project in Ekibastuz. In a social media post, Kazakh expert and entrepreneur **Ermek Shamuratov** calculated that selling electricity already in short supply to a foreign operator at 2.5 US cents per kilowatt-hour is **uneconomic**.

> The electricity allocated to the operator has to be replaced through imports or new generation, both of which cost several times more. By his estimate, Kazakhstan loses 18–22 tenge on every kilowatt-hour sold. His conclusion: selling electricity abroad makes sense only **at replacement cost**, around 6.5–7.5 US cents per kilowatt-hour.

For now, this is an expert’s assessment rather than the government’s position. But those of us in Kyrgyzstan should look closely at the numbers. We also receive electricity from Kazakhstan in winter. If pricing at replacement cost ever becomes the rule, **Kyrgyzstan will be affected too**. That makes it worth explaining how our energy relationship with Kazakhstan differs from ordinary electricity sales.

#### Prices: what Shamuratov is comparing

Kazakhstan’s Ministry of Artificial Intelligence and Digital Development is developing a **long-term tariff** of around 2.5 US cents per kilowatt-hour for investors in the valley. At an exchange rate of roughly 444 tenge to the dollar, that works out to about 11.1 tenge. This is comparable to the tariff ceiling at Ekibastuz GRES-1, which has stood at 10.92 tenge per kilowatt-hour excluding VAT since September 10, 2026.

> Kazakhstan’s average electricity export price in 2025 was similar: $23.7 per megawatt-hour, or around 2.4 US cents per kilowatt-hour. About 70% of those exports went to Russia. Kyrgyzstan paid more: in January–May 2025, Kazakhstan sold us 415.7 million kWh for $12.1 million, or around 2.9 US cents per kilowatt-hour. That is higher than both the average export price and the tariff being offered to the data center. At least in 2025, we were not buying at a discount. But Shamuratov’s calculation raises a more important question: **what Kazakhstan gets in return**.

#### What Kazakhstan gets from the data center

The project’s supporters say it will bring **AI computing capacity** and **large investments** to the country. In October 2026, the ministry clarified how the arrangement would work. Firebird will own the servers. KT-Telecom operates the valley and finances the **buildings** and **utility infrastructure**. Users in Kazakhstan will be able to rent computing capacity, but the prices and the amount available to them have yet to be determined. The return from the data center therefore needs to be assessed more broadly than through the electricity tariff alone.

#### Electricity exchanges with Kyrgyzstan

We need to take an equally broad view of our relationship with Kazakhstan, which involves more than electricity sales. In summer, when water is plentiful, Kyrgyzstan **supplies electricity to Kazakhstan** and **receives it back in winter**. **Taalaibek Ibrayev**, then energy minister, explained in November 2025 that this arrangement had always been in place and that winter deliveries used a return factor of 1.3: we receive 30% more electricity than we supply in summer. That year, Kyrgyzstan also returned 161.1 million kWh to Kazakhstan under a 2021 agreement.

> The second part of the relationship involves **Toktogul Reservoir**. In November 2025, energy and water authorities from the three countries agreed in Almaty that Kazakhstan and Uzbekistan would supply electricity to Kyrgyzstan in winter, allowing us to reduce winter drawdowns from Toktogul. In summer, the stored water goes to irrigate farmland in southern Kazakhstan and Uzbekistan. We do not sell this water: the Syr Darya is a transboundary river. We provide a **flow regulation service**, shifting water releases from winter to summer, even though that water could otherwise generate electricity at our own hydropower plants in winter.

Our electricity shortfall, however, is primarily structural. In 2025, Kyrgyzstan consumed 19.3 billion kWh, generated 15.4 billion and imported around 3.9 billion. Most of those imports are **ordinary commercial purchases**, unrelated to how Toktogul is operated. Only the electricity equivalent of the deferred water releases can be credited to the flow regulation service. If we fail to separate these two categories, Astana will rightly see our argument as an attempt to pass off **our own shortfall** as help for a neighbor.

#### The risk to winter supplies

For us, the Ekibastuz data center is primarily a competitor for the same kilowatt-hour. According to press reports in October 2026, the project’s first-phase capacity of **215 MW** is already fully reserved. At full load, it would consume almost **1.9 billion kWh a year** from GRES-1, roughly twice the **890 million kWh** Kazakhstan supplied to Kyrgyzstan in the whole of 2025.

> Kazakhstan sees this risk differently. KEGOC head **Nabi Aitzhanov** said on November 10, 2026 that the valley’s electricity demand was already included in supply and demand projections. Even with that demand, Kazakhstan expects an electricity surplus in 2027 and a surplus of generating capacity in 2028\. If the forecast proves correct, there will be **enough electricity for both** the data center and neighboring countries. If it does not, **winter exports** will be the first supplies cut. We should guard against that risk now.

#### Proposals

- **First**, distinguish electricity exchanges from purchases in the agreements. Exchanges should cover summer deliveries from Kyrgyzstan, which Kazakhstan already returns in winter using a multiplier, and deferred water releases that benefit our neighbors. Those deferred releases are not yet credited separately: under the current protocols, winter electricity deliveries to us are classified as exports. I propose crediting the deferred releases in kilowatt-hours as well. Purchases to cover our structural shortfall should be handled separately, at market prices. Any revision of Kazakhstan’s export prices would then affect only those purchases, leaving the exchange arrangements intact.
- **Second**, agree on the return factor and Toktogul’s operating rules for several years ahead. For now, the parties negotiate the terms season by season. They did so in September and November 2025, and again in June 2026, when they signed a protocol in Bishkek covering the growing and autumn–winter periods of 2026–2027\. A multiyear agreement on Toktogul’s operation and the electricity exchange rules would give both sides greater predictability. It would not set import volumes and would therefore be consistent with President **Sadyr Japarov’s** goal of ending electricity imports within two to two and a half years. As new generating capacity comes online, purchases will decline, while Kazakhstan will still need flow regulation for irrigation.
- **Third**, reach an agreement before the data center’s first phase starts operating. In a year, once it is running, negotiations will concern the actual supply and demand balance at GRES-1\. While the figures for 2027 remain forecasts, both sides have more room to compromise.