Tasnim Kabir: Bangladesh should not treat data centres as pure technology projects. In this country, climate is not background scenery; rather it is a binding engineering constraint.

A data centre appears clean and digital from the outside but inside it is a demanding physical system that requires continuous electricity, precise cooling, humidity control, fire protection, water management, backup power, secure connectivity, trained operators, and disciplined maintenance. Failure in any of these systems can take digital services offline.

Bangladesh is entering a new phase of digital ambition. Artificial intelligence, cloud services, mobile finance, e-commerce, government platforms, health records, online education, industrial automation, and data-driven public services will all depend on stronger computing infrastructure. Data centres will join power plants, ports, roads, telecom networks, and economic zones as national infrastructure. They cannot be planned only as technology projects. They must be planned as climate-sensitive infrastructure.

Here are several issues to think before building data centres.

Servers generate heat continuously. Higher computing density, especially from AI workloads, intensifies the load. Cooling is, therefore, central to reliability, energy cost, equipment life, and environmental performance. 

In a hot and humid climate, designs proven in cooler regions often underperform. Outdoor temperature, humidity, dust, monsoon conditions, air quality, and grid stability all shape system behaviour. Bangladesh must verify local design-temperature data, humidity ranges, flood risk, and seasonal cooling loads before approving major sites in Dhaka, Chattogram, Gazipur, Narayanganj, Sylhet, Khulna, or coastal zones.

Some facilities use water directly or indirectly for cooling while others rely on air or closed-loop systems. The correct choice depends on climate, site conditions, power cost, water availability, regulation, and scale. Simplistic claims that every data centre will either consume excessive water or none at all miss the point. The real questions are engineering ones: which cooling method is proposed, how much water is required, whether recycled or wastewater can be used, whether the site is water-stressed, and what happens during drought, contamination, flooding, or utility disruption. 

A responsible national strategy should require transparent water-use assessments that examine expected consumption by technology, local availability, and competing demand from households, agriculture, industry, and public utilities.

Bangladesh’s geography makes flood risk unavoidable. A data centre cannot be sited solely on the basis of cheap land or political convenience. Elevation, drainage capacity, river proximity, urban waterlogging, cyclone exposure, emergency road access, and the reliability of surrounding utilities must all be assessed. Floodwater need not enter the server hall to cause failure. It can block fuel delivery, damage electrical rooms, close access roads, compromise fibre routes, or disable nearby substations. 

A facility designed only for normal weather is not resilient.

Data centres need stable electricity, not merely available electricity. Voltage fluctuations, frequency instability, outages, transformer failures, weak feeders, and slow restoration raise both risk and cost. Backup generators and batteries are essential protection layers; they are not substitutes for a reliable grid. When the primary power system is weak, the entire facility becomes more expensive and less trustworthy. 

Data-centre planning must, therefore, be linked to power-sector planning from the outset. Dedicated substations, redundant feeders, protection coordination, backup-fuel logistics, power-quality monitoring, and rigorous maintenance discipline are non-negotiable. Grid reliability indicators, outage frequency, restoration times, power-quality data, and available substation capacity in candidate zones should be verified before commitments are made.

Data centres are energy-intensive. Efficiency is a business requirement, not an imported slogan. Poor cooling design, overloaded electrical systems, inefficient equipment, and weak maintenance raise operating costs and undermine competitiveness. Investors judge more than land price or labour cost; they assess whether a country can deliver reliable, efficient infrastructure over time.

A data centre is operated by people. Bangladesh will need engineers and technicians skilled in electrical systems, cooling, fire protection, building-management systems, SCADA, automation, cybersecurity, safety procedures, emergency response, and reliability-centred maintenance. This capability must form part of national workforce planning. Universities and polytechnic institutes should develop practical training pathways for critical-facility operations rather than waiting until facilities are already built.

Clear standards are required for site selection, power redundancy, cooling design, fire safety, water use, environmental impact, cybersecurity, uptime expectations, maintenance documentation, and disaster recovery. Standards must be strong enough to protect the national interest without becoming so burdensome that they deter investment.

A poorly planned facility can burden the grid, create environmental problems, or become a fragile point in the digital economy. A well-planned one can support AI, cloud services, banking, exports, public services, research, and skilled employment.

The practical response is not to reject data centres because they are demanding. It is to plan them intelligently. Bangladesh should identify climate-resilient zones based on power availability, fibre connectivity, flood risk, land stability, water strategy, emergency access, and proximity to economic activity.

Feasibility studies should address heat, humidity, power quality, cooling efficiency, water risk, and disaster recovery. Renewable integration should be encouraged where technically and economically viable, without the pretence that intermittent power alone can support critical facilities. Efficient design, strong maintenance culture, and transparent operating standards should be promoted.

Digital ambition must not be confused with digital readiness. AI does not run on policy statements. Cloud infrastructure does not operate on optimism. Data centres do not remain reliable simply because they are labelled modern. They work only when the physical realities beneath the digital economy are respected.

Climate is a design input, a cost factor, a reliability risk, and a national planning issue. If Bangladesh absorbs this lesson early, it can build facilities that strengthen the economy without weakening the grid, wasting resources, or exposing digital services to preventable failure. If it learns too late, it may construct expensive assets that look advanced yet remain fragile.

The future digital economy will reward countries that understand both software and infrastructure. Bangladesh has the ambition to participate. It must now demonstrate the engineering discipline to build for its own climate, its own grid, its own water realities, and its own people.

Writer: Tasnim Kabir is a Data Centre Engineering Operations Engineer (DCEO Engineer) at Amazon Web Services (AWS)