• Over 25 responses received from hyperscalers, data centre operators
• ADB-backed consultants assessing investment potential, infrastructure needs
• Initiative remains at planning and market consultation stage: Govt.
• Grid stability, electricity tariffs emerge as major investment hurdles
• Rs. 3 b allocated for AI, cloud computing, data centre infrastructure
• Govt. considers smaller data centres while pursuing hyperscale investment

'

As global tech giants race to secure the massive computational backbones required to fuel generative Artificial Intelligence (AI), Sri Lanka is poised to step onto the digital real estate stage. 

However, following a flurry of preliminary discussions with international hyperscalers and key infrastructure players, the island now faces a sobering reality check: ambition is plentiful, but foundational readiness remains tethered to a tight network of policy, power, and structural bottlenecks.

The Ministry of Digital Economy, working alongside the Board of Investment (BOI), recently floated a formal Request for Information (RFI) to test global waters. 

The objective: gauge interest from major hyperscalers and data centre operators regarding potential footprints in Sri Lanka, while mapping out their technical, land, and energy prerequisites.

While the preliminary response has surpassed internal expectations, Government officials, telecom insiders, and digital policy observers emphasise that Sri Lanka remains in an exploratory phase – one where ambitious visions must reconcile with the harsh constraints of local infrastructure.

Response to RFI 

Deputy Minister of Digital Economy Eranga Weeraratne told The Sunday Morning that the State’s preliminary market query had yielded significant international attention.

“There’s a current RFI released by the BOI in consultation with the Ministry of Digital Economy to gauge interest among hyperscalers and other data centre providers to establish data centres in Sri Lanka. We are trying to understand the capacity they are looking to invest in and their requirements in terms of commitments from the Government, electricity tariffs at specific rates, land requirements, etc.,” Weeraratne explained.

According to him, the ministry has received over 25 responses to the RFI from hyperscalers and is now heading towards analysing the responses with technical assistance from the Asian Development Bank (ADB) in order to fully understand the potential to establish AI data centres. “We want to assess the potential that exists in the region and determine what sort of data centre capacity Sri Lanka should go for,” he said.

Weeraratne added that the ADB had appointed certain consultants to carry out this process, noting that baseline strategy discussions remained active.

“Strategy consultation is currently underway to establish a baseline. The initial draft has been presented, but it will likely be improved significantly based on feedback. Once these changes are made, we will be in a position to publish the final proposal and get interested parties involved. The Government will also formulate the policies appropriate to these recommendations,” he said.

When contacted by The Sunday Morning, Media Secretary to the Deputy Minister of Digital Economy Nipuni Rajapaksha said that the country was still far from breaking ground on building data centres.

“Right now, this is only at the discussion stage. The infrastructure for data centres is yet to be developed,” she said, noting that issues posed by the construction of such infrastructure needed to be resolved before the initiative could move to the next stage.

Water usage and power grid stability

According to an article by the Environmental and Energy Study Institute (EESI) based in the US, data centre developers are increasingly tapping into freshwater resources to quench the thirst of data centres. Large data centres are reported to be capable of consuming up to five million gallons per day, equivalent to the water use of a town populated by 10,000–50,000 people.

As such, with larger and new AI-focused data centres coming into operation, water consumption is increasing alongside energy usage and carbon emissions.

Addressing these structural hurdles, Weeraratne pointed out that while bandwidth and liquid resources were manageable, power grid stability and tariff structures required significant intervention.

“In terms of electricity tariffs as well as ensuring grid redundancy, there is much that needs to be improved,” the Deputy Minister said. “As for water requirements, that is very manageable. There is nothing preventing this because Sri Lanka has substantial water resources.”

Weeraratne explained that while water was used in the maintenance of data centres, its primary function was for cooling. “The water will not get polluted since it is not like running an electricity power plant. It is more for the purpose of cooling. The newest systems in most data centres are accompanied by closed-loop water recycling facilities and therefore do not require large external supplies of water,” he said.

On the power front, hyperscalers typically require dedicated clean energy feeds, prompting Sri Lanka to consider alternative grid mechanisms such as corporate Power Purchase Agreements (PPAs) and virtual power wheeling. “Large data centres will come with their own power sources as well, including renewable power sources. These matters are yet to be discussed and finalised,” Weeraratne added. 

Recognising that hyperscale investments take time to materialise, the Government is executing a parallel strategy focused on immediate internal needs.

“The process thus far has involved discussion with all stakeholders and we will gradually reach our target. Until we establish hyperscale AI data centres in the country, the Government may consider investing in and procuring data centres of smaller capacity. This constitutes our two-pronged action plan,” Weeraratne stated.

Energy challenges  

While State officials navigate official diagnostics, local technology practitioners and AI educators have raised fundamental questions about national priorities, power economics, and the sheer hardware depreciation cycles defining the modern AI surge.

An AI trainer and content creator, speaking on the condition of anonymity, highlighted the stark contrast between traditional physical infrastructure and high-density computing assets.

“Earlier, railway tracks were built to last for about 200 years. Likewise, dial phones had a lifespan of about 20 or 30 years. Today, however, mobile phones last just five years and are then discarded. Similarly, data centres are exceedingly large, yet have short lifespans and require high maintenance.”

He pointed to global pushback against massive computing hubs, citing environmental and energy concerns across North America. “A country capable of setting up data centres is also capable of developing further. However, the problem with data centres is energy. This has led to protests in many US states, where they are effectively saying ‘not in my backyard,’ just like in Sri Lanka, where environmentalists oppose the establishment of such projects.”

Crucially, he highlighted Sri Lanka’s structural energy context, arguing that the nation could not afford to divert grid capacity to service foreign cloud workloads unless such centres were fully self-funded through independent renewable energy generation.

“Sri Lanka is a country that has already experienced an energy crisis. If we are to establish data centres, we must have energy,” he said.

He pointed out that while solar-powered data centres could be an option, this would pose a challenge since such facilities would not be able to operate at night. “Sri Lanka does not have an energy surplus, we have a deficit.”

The AI trainer further underscored that setting up hyperscale computing facilities without sovereign energy generation risked placing the national grid at the service of foreign tech firms while delivering minimal internal value. However, he noted that opposition to such data centres would be minimal if they were powered by eco-friendly or green energy. 

“If they use the electricity grid, that will be an issue since we lack that kind of energy. The basic requirement in developing AI is energy. At the foundation is energy, followed by data centres, Large Language Models (LLMs), and others. Energy is therefore the primary factor. If Sri Lanka can provide that, it could give us a major competitive edge.”

He offered a pointed critique of national investment choices, urging policymakers to prioritise domestic energy stability and skill development over premature cloud hosting for foreign entities.

“If we have energy, it would be better if we could first manage our own electricity supply instead of establishing data centres. Simply establishing a data centre does not mean that we will be able to develop LLMs in Sri Lanka. Instead, we will have to direct our energy resources to serve foreign countries.”

The current situation

The Government’s previous investment initiative to establish AI data centres sought proposals from investors, developers, and operators for large-scale, sovereign data centre and cloud infrastructure capable of supporting high-performance computing, AI workloads, and national digital services. The first RFI closed on 8 January this year.

In June, Deputy Minister Weeraratne said that procurement processes for AI data centre infrastructure were underway. According to him, the aim is to serve Sri Lankan demand and, eventually, regional customers, generating foreign exchange earnings. In May, the Government issued a market consultation for GovCloud 3.0.

At present, Sri Lanka is actively pursuing AI-ready data centre capacity, with the programme still at the planning, market consultation, and procurement design stage. The 2026 Budget also announced concessions intended to attract private data centre investors, including subsidised land, initially reduced electricity prices, and green energy-related incentives.

Accordingly, Rs. 3 billion has been allocated for next-generation digital infrastructure encompassing AI, cloud computing, and data centres, with Rs. 500 million identified for investor attraction measures.

Sri Lanka also discussed a national AI data centre with the Japan International Cooperation Agency (JICA) in August, confirming continuing political interest and possible avenues for technical cooperation. However, the ministry is yet to identify a signed construction contract, selected partner, financing package, or completion date.