Barbados Policy Pulse — Part 1 of 3: The Greenland Data Centre

Barbados Wants a Data Centre. Nobody Has Said Where the Water Is Coming From.

In March 2026, Senator Jonathan Reid stood in the House of Assembly and asked Parliament for $187.8 million to build Barbados into a “world-class” digital economy by 2030. Leading the plan: a government-owned data and green power centre in Greenland, St Andrew, scaling from a 100-kilowatt pilot to a 15-megawatt facility, with the first 5 MW owned by the state and the remaining 10 MW opened to private investors (Barbados Today).

The pitch was framed almost entirely around sovereignty. Reid argued that Barbadian data sitting on foreign servers leaves the country exposed to “any other person’s or any other state’s whims and fancies,” and that digital infrastructure now belongs in the same category as roads and bridges. BIDC CEO Mark Hill added that the site was chosen for its security and existing infrastructure, and that clearing had already begun.

What neither official addressed, in that speech or in any follow-up reporting we could find, was water.

A resource-heavy industry lands in a water-stressed country

Barbados is ranked among the ten most water-stressed countries in the world by the UN Office for the Coordination of Humanitarian Affairs. More than 80 percent of the island’s freshwater comes not from rivers or reservoirs but from thin lenses of groundwater floating above saltwater in porous limestone aquifers, and hydrogeologists at the Barbados Water Authority have already documented saltwater intrusion in parts of the west coast from over-pumping (Cari-Bois Environmental News Network). The island’s total annual water resources are estimated at roughly 59 million cubic metres in an average year, dropping to about 45 million cubic metres in a one-in-fifteen-year drought.

Data centres, meanwhile, are thirsty by design. Conventional evaporative cooling systems, the cheapest and most widely used method, lose up to 85 percent of the water they draw to evaporation rather than returning it to the supply, and a single medium-sized facility can consume on the order of 110 million gallons a year for cooling alone (MOST Policy Initiative). Beyond direct cooling, the industry carries a secondary pollution risk: cooling towers are a known site for Legionella growth, and the biocides traditionally used to treat cooling water, chlorine-based compounds among them, can damage aquatic ecosystems if discharged without proper treatment.

This is arguably the single biggest technical risk in a plan the government has otherwise pitched as a national security necessity.

What other governments are asking developers to disclose

Barbados would not be inventing this wheel. Regulators elsewhere have already started requiring answers before land breaks:

• The UK’s Environment Agency and Ofwat now push developers to demonstrate water efficiency, alternative cooling sources, and alignment with regional water plans before large facilities are approved.
• The EU’s Energy Efficiency Directive requires data centres above a certain size to publicly report their potable water usage annually, with penalties left to member states.
• Singapore’s Public Utilities Board requires large industrial water users to file Water Efficiency Management Plans and meter consumption, explicitly to push operators toward non-potable cooling sources.

None of these frameworks ban data centres in water-stressed places. They just require the water math to be shown publicly before construction, not after.

Three concrete alternatives worth putting on the table

1. Closed-loop and reclaimed water cooling. Rather than drawing on potable groundwater, cooling systems can run on treated wastewater or harvested rainwater in a closed loop, cutting freshwater draw by an estimated 50 to 70 percent. Barbados is already building relevant infrastructure for this: the Green Climate Fund-backed South Coast Water Reclamation Plant is designed to route treated wastewater into irrigation and aquifer recharge. There is no public indication that this reclaimed supply has been connected, even conceptually, to the Greenland project, but the pipeline logic already exists on the island.

2. Direct-to-chip or immersion liquid cooling. Instead of cooling the room, these systems cool the processor directly, using a sealed liquid loop that barely touches freshwater supply at all; some vendors claim water savings of up to 95 percent compared to evaporative systems. It costs more upfront and requires different equipment, but for a facility being built from scratch, this is the point in the process where that choice is cheapest to make.

3. China’s underwater data centres, as a genuine (if imperfect) case study. China’s Highlander/Hailanxin has now built commercial underwater data centres off Hainan Island and, more recently, a wind-powered facility off Shanghai, using pumped seawater as a natural, continuous coolant. The company reports the approach is 40 to 60 percent more power-efficient than land-based cooling and, at full 100-module scale, projects savings of roughly 105,000 tons of freshwater and 122 million kilowatt-hours a year at the Hainan site (Merics; Data Centre Magazine).

This is not a plug-and-play answer for Barbados, and it comes with its own environmental questions worth naming honestly. Thermal discharge from these facilities does warm the surrounding water. Hailanxin reports a rise of under 2°C at the outlet from its Hainan facility, which the company calls manageable, but outside analysts have pushed back, noting that continuous heat discharge at scale (their stated 100-pod buildout would put 50 to 100 megawatts of waste heat into the sea) is a real, if localised, thermal load on marine ecosystems (The National Interest). Corrosion, sealing integrity over a 25-year design life, and maintenance access are unresolved engineering questions even for Highlander itself. Microsoft ran its own underwater pilot, Project Natick, off Scotland from 2018 to 2020, concluded the hardware reliability case was real, and still chose not to commercialise it.

For a small island state, the more relevant takeaway from China’s experiment may not be “put servers in the ocean.” It is that the industry itself now treats water as a design variable to be engineered around, not an unavoidable cost of doing business. That framing has not yet appeared in anything the Barbadian government has said publicly about Greenland.

The actual ask

Barbados does not need to copy China’s approach, or anyone else’s. What the current public record is missing is much simpler: a stated cooling method, a stated water source, and a commitment to some form of environmental assessment before the facility scales past its pilot phase. The government has been transparent about the economic case for this project. The same transparency has not yet extended to its resource footprint, in a country that can least afford to guess wrong on water.