IMAGE: SHUTTERSTOCK
ON PAPER Nepal’s pitch to the world’s data-centre operators is compelling. The country generates some 4,400 megawatts of electricity, almost all of it from hydropower. During the monsoon, when rivers run high, it wastes up to 1,000MW because domestic demand is too low and the transmission network cannot move the electricity to where it might be sold. That surplus is cheap: Rs 4.80 per kilowatt-hour, or about $0.035, during the wet season. By comparison, electricity for a data centre in Singapore costs roughly $0.17 per kilowatt-hour; in Germany, it is above $0.30. The maths makes a data-centre operator take notice.
The climate helps. Cooling accounts for 30-40% of the energy consumption of a large data centre. Nepal’s mountainous terrain, with sites at 1,400 metres in the Kathmandu valley and higher still, reduces the need for mechanical cooling. A hyperscale facility in Arizona or Singapore must spend heavily on chillers and air-handling equipment; a facility in Nepal could redirect much of that capital expenditure elsewhere.
The country’s data-centre market was valued at about $407m in 2025 and is projected to grow to more than $1.1bn by 2035, a compound annual rate of 10.4%. Existing operators are expanding. WorldLink, through its subsidiary Data World, runs a Tier-3 data centre. The government’s budget for 2026-27 includes plans for a “Sovereign AI Compute Center” in Syuchatar, Kathmandu, which would use hydropower to run thousands of AI processors and provide subsidised high-tier computing to domestic startups. Bichuten Data Vault has announced a Tier-4 hyperscale facility that it says could consume 100 to 500MW on its own. That is a striking figure: WorldLink’s existing facility has a capacity of 3.5MW.
The difference between Nepal’s ambitions and its current reality makes the data-centre story more complicated. The government has announced intentions and allocated money. It has not established the zoning rules, environmental standards, data-sovereignty laws or grid-connection requirements that a large international operator would need before committing capital. A Kathmandu Post headline from April captured the problem like this: “Nepal wants to become a data centre hub. It has no rules for how.”
Consider the Ncell Data Center in Nakhhu, Lalitpur, the country’s most advanced facility. Residents have complained about nighttime noise, black smoke, dust and traffic. The centre runs diesel generators to ensure consistent power because the grid connection is not reliable enough for critical infrastructure. A facility that burns diesel for reliability, generates neighbourhood complaints and operates at a tiny fraction of the scale the government is promising to attract is not a compelling advertisement for foreign investment.
Kenya offers a cautionary parallel. The East African country has also sought to position itself as a regional data hub, with Chinese investment playing a significant role in its digital infrastructure. That raises a question for Nepal that its own data-sovereignty framework does not yet answer. Much of Nepal’s hydropower capacity has been built with Chinese investment and technology. If a large American technology company were to build a data centre in Nepal, what legal regime would govern the data it stores? How would disputes be adjudicated? What would happen to customer data if the Nepali government demanded access? No clear answers exist.
The most credible near-term path is not the hyperscale facility but the smaller, more forgiving market for disaster recovery and edge computing. These facilities are less sensitive to latency, less demanding of perfect grid reliability and more appropriate for a country that is still building its regulatory and technical credibility. Nepal could start by demonstrating that it can host a 10MW facility without community complaints and with reliable uptime. It could develop its regulatory framework in parallel. A track record of smaller successes would do more to attract the next tier of investment than a budget announcement for a 500MW facility that lacks the foundations to support it. The opportunity is real. The conditions for seizing it are not yet in place. ■
