Nepal’s Data Center Dream: Promise, Power, and a Reality Check
An analysis of the viability, energy demands, and environmental trade-offs of establishing data centers in Nepal.
- Data Centers
- Hydropower
- Energy Economics
- Nepal
In Nepal, every new technology arrives first as a dream. Data centers are no different. They promise jobs, investment, digital sovereignty, and the use of surplus hydropower. But behind the promise lies a difficult question: can Nepal’s electricity grid, water resources, and policy system carry the weight of this dream?
With the recent news about establishing data centers in Nepal, there has been an overwhelming wave of excitement about Nepal becoming a data hub in South Asia. Some reports have even suggested that companies like Google and Meta are entering the Nepali market, creating thousands of jobs and using Nepal’s hydropower-generated electricity, which might otherwise go to waste, to build green data centers.
As someone interested in emerging technologies, energy economics, and their environmental impact, I thought it was worth writing about the data center discussion currently creating so much buzz in Nepal, especially from the perspective of energy, economics, and sustainability.
To begin with, let’s talk about what data centers really are. A data center is essentially a large warehouse that houses large-scale computer systems, including IT equipment such as servers, storage systems, and networking devices. Its job is to store, process, and deliver data reliably at scale. Everything we do online, searching on Google, watching YouTube, or even reading this article right now, depends on a data center somewhere. But to keep these data centers up and running, equipment beyond just servers is required. Cooling infrastructure, power delivery systems, backup generators, electrical distribution, batteries, and other support systems are needed. Together, these systems consume a lot of energy.
To run an average data center, a facility with an area of around 20,000 to 100,000 square feet hosting anywhere from 500 to 2,000 servers may require 10 to 100 MW of electricity. To put that into Nepal’s context, where per capita electricity consumption is around 400 units according to NEA, a 50 MW average data center could consume electricity equivalent to roughly 182,000 households, assuming six members per household. That is close to the electricity use of an entire city like Butwal. It is also comparable to the annual production of Upper Trishuli 3A Hydropower Project, a 60 MW plant with an annual generation capacity of around 460 GWh.
There is also a hyperscale data center being built near where I live, in Saline Township, Michigan, by OpenAI, the company behind ChatGPT. The project has been reported to cost around $16 billion and could have a capacity of nearly 1 GW. To put that in perspective, supplying such a facility would require the annual output of roughly 10 to 11 Kali Gandaki-A hydropower plants.
Recently, there was news about Bichuten Data Vault opening a Tier-IV hyperscale data center in Nepal, which could consume around 5 MW by 2030. Based on its capacity, however, this is nowhere near the hyperscale data centers being built in the United States, nor is it enough by itself to make Nepal a data center hub. A hyperscale data center typically has a capacity over 100 MW or more and can consume electricity equivalent to the entire output of a large hydropower plant. The data center being built by Bichuten is better understood as an edge data center, designed to bring data processing closer to end users.
Nepal does, however, have certain natural and strategic advantages that make it attractive for data center development. The country has a large hydropower potential, which could support a cleaner electricity supply if generation and transmission infrastructure are properly expanded. Many hilly regions also have cooler climates, which can reduce the energy needed for cooling compared to hotter lowland areas. Nepal’s geographic location between India and China gives it strategic importance in a region where digital services are growing rapidly. At the same time, Nepal’s own demand for cloud services, digital payments, online government platforms, banking systems, telecom networks, and secure domestic data storage is increasing. For these reasons, data centers can play an important role in strengthening Nepal’s digital infrastructure, especially if the focus begins with domestic cloud capacity rather than unrealistic hyperscale ambitions.
Although building hyperscale data centers is still a far cry for Nepal from an energy consumption perspective, the establishment of Tier-I and Tier-II data centers is a highly positive step for the country. These facilities can help meet Nepal’s rapidly growing digital demand by supporting cloud services, banking systems, telecom networks, government platforms, and local enterprises with more reliable and secure data storage. They can also reduce dependence on foreign servers, improve data sovereignty, lower latency for internet services, and create opportunities for IT investment, skilled jobs, and the growth of Nepal’s digital economy.
Establishing a few data centers like Bichuten could help satisfy local demand, but making Nepal a global or regional data center hub is still far from reality. The investment of top global companies like Meta, Google, and Microsoft is not something that can be assumed so easily. Based on data provided by NEA, the total available energy in FY 2024/25 was 15,641 GWh. So, let’s do some math. A hyperscale data center established by large technology corporations can go up to 100 MW. A 100 MW facility running 24/7 would consume around 876 GWh per year. If we divide Nepal’s total available energy in FY 2024/25 by that number, Nepal’s total available electricity could theoretically support only about 18 hyperscale data centers of 100 MW each. The number would double for 50 MW facilities, but no country has become a data center hub by building only 18 hyperscale data centers. In the United States alone, there are more than 1,000 hyperscale data centers, accounting for more than half of all hyperscale data centers in the world.
Interactive: Grid Load Simulator
Adjust the slider to simulate adding 100 MW Hyperscale Data Centers to Nepal's grid.
All of this brings us to an important point: Nepal cannot afford to build many hyperscale data centers right now, because its total domestic electricity generation could support only a limited number of them. This may change in the coming decade, but for now, a few Tier-I and Tier-II data centers could meet domestic demand and help consume surplus electricity. If Nepal truly wants to become a data center hub, it must first draft proper laws to regulate them, build large hydropower plants, and improve transmission infrastructure, because demand will spike once these energy-guzzling facilities start entering the system. Only after those steps are taken can we realistically think about becoming a hub.
Now, let’s talk about job creation. The unemployment rate among Nepali youth aged 15 to 24 was around 22.7 percent in 2024. In a country where jobs are difficult to find, data centers could bring high-paying skilled jobs. But they are not the job-creation machines many people imagine them to be.
These facilities typically employ very few full-time IT workers. They create many temporary construction jobs during the building phase, but once operational, even a hyperscale data center may employ only around 50 full-time workers. Given their size and footprint, data centers do not create many jobs because they do not produce goods in the way factories do. Unlike manufacturing plants, which may need thousands of workers to produce physical products, data centers are largely big buildings that house computers and require relatively few maintenance and operations personnel.
On one hand, hyperscale data centers create very few full-time jobs, which is nowhere near enough to address Nepal’s unemployment challenges. On the other hand, the jobs they do create require highly skilled professionals. In a country already struggling with brain drain, finding such talent can be difficult. If Nepal wants to benefit from this industry in the future, it must also invest in strengthening its education system and preparing a capable technical workforce.
Another problem with data centers is that they consume a lot of water for cooling, especially when they use open-loop cooling systems. Tens of millions of liters of water can be evaporated into the atmosphere. For this reason, they are not ideal for cities like Kathmandu Valley or Pokhara, where water shortage is already a serious issue.
Instead, they should be built near rivers where they can use cold flowing water from the Himalayas to support cooling. Although water availability may not seem like a major issue for Nepal in general, during the winter season, when river levels decrease, evaporating millions of liters of water could also affect hydropower generation. On the other hand, data centers can be cooled by air, but air-cooled systems usually consume more energy than water-cooled systems.
If data centers are built in urban areas, they will certainly strain local water utilities and may create problems during dry or winter periods. To address this, closed-loop water cooling systems should be designed wherever possible. Since energy is already a major barrier in Nepal, water cooling may be the more viable option, but it must be planned carefully so that it does not disturb the water-energy nexus.
Another issue with data centers is noise. The noise comes from continuously running pumps, chillers, evaporators, cooling towers, backup generators, server fans, and other mechanical systems. Tier-I and Tier-II data centers may not create serious noise issues, but hyperscale data centers can produce significant sound from cooling equipment, diesel generators, and server ventilation systems. In some cases, the sound can reach harmful decibel levels and affect nearby residents’ health and sleep.
Case studies from places such as Arizona, Virginia, Arkansas, and Texas show residents reporting headaches, vertigo, hypertension, and chronic stress linked to constant low-frequency hums and high-decibel equipment. Let’s hope that will not be the case in Nepal. A densely packed city like Kathmandu cannot hold many such facilities. In fact, Kathmandu does not have enough space, water, or infrastructure to support them at scale. Remote locations in the hilly regions of the country may be more suitable, provided that transmission lines, roads, water resources, and community concerns are properly addressed.
Now, let’s talk about Ireland, which has recently become a major data center hub in Europe. Ireland has 121 operating data centers, driven by multinational technology investment from companies like Meta, Google, and Amazon. This has brought billions of dollars in foreign investment and corporate tax revenue. But data centers in Ireland now use around 20 percent of the country’s electricity, and that figure is projected to reach 33 percent in the coming years.
The electricity demand from data centers is outpacing generation capacity. Ireland once embraced data centers enthusiastically, but now they are straining the grid, raising electricity prices for residents, and forcing authorities to put temporary restrictions on new data centers until the problems are addressed. The Dublin grid is already saturated and cannot take much more load.
Interactive: Reality Check Matrix
Comparing established global data center hubs against Nepal's maximum theoretical limit.
By looking at these global examples, Nepal can still work toward becoming a data center hub, but it must do so through proper policy frameworks, infrastructure investment, workforce development, regulation, zoning, and community benefit programs. There will be problems if too many data centers are built without planning. Authorities must make sure that the cost of building these facilities is not passed on to ordinary consumers like you and me. The increasing demand for electricity should not make our bills go up.
Nepal’s electricity market is owned, operated, and regulated largely by a government-owned entity, so prices may not rise in the same way they do in countries where private utility companies dominate the market. Still, the revenue and technology that come from data centers should directly benefit the people. If public resources such as electricity, water, land, and transmission infrastructure are used to support these facilities, the public should also share in the benefits.
One final piece is the need for broader planning. Nepal must think carefully about how much total water and electricity demand from data centers the country can handle if they begin to mushroom. Building one or two is good. Building too many without proper planning is definitely not the right path for now.
Nepal should not reject data centers, but neither should it romanticize them. The question is not whether Nepal needs data centers. It does. The real question is how many, where, under what rules, and for whose benefit.