Singapore’s Data Centre Policy: Why a Booming Industry Was Told to Stop

Singapore is the rare government that told a booming industry to stop building. Data centres brought investment, jobs and strategic infrastructure, and they were also consuming a large and rising share of a national electricity supply that is entirely imported. The response, a moratorium followed by a conditional allocation system, is one of the clearest examples anywhere of a country pricing a scarce resource rather than rationing it politically. This case study is part of the telecom, media and data infrastructure pillar of the Singapore Company Stories hub.

Why did Singapore become a data centre hub?

Political stability, legal certainty, excellent submarine cable connectivity, low seismic and flood risk, reliable power supply and proximity to the fast-growing Southeast Asian digital economy made Singapore the region’s default location for critical data infrastructure.

For a decade the sector grew rapidly, attracting hyperscale cloud providers, colocation operators and enterprise facilities, and Singapore became one of the largest data centre markets in Asia relative to its size.

The connectivity advantage was decisive and is examined further in the submarine cable case study. A data centre is only useful if data can reach it, and Singapore is where the region’s cables land.

What was the problem with unconstrained growth?

Data centres consume electricity continuously at high density, and the sector’s share of national electricity consumption rose to a level that policymakers judged incompatible with climate commitments and energy security in a country importing nearly all its fuel.

Singapore has no domestic oil, gas, coal, hydro or meaningful renewable potential. Every additional megawatt of demand must be met by imported fuel or, increasingly, imported electricity, and both carry cost and security implications.

Land was the secondary constraint. Data centres occupy substantial space in a country where land allocation is a zero-sum exercise between housing, industry, defence, transport and green space.

How does the allocation system work?

Rather than approving applications first come first served, authorities ran a call for applications and awarded limited capacity to operators scoring highest on energy efficiency, use of low-carbon energy, economic contribution and alignment with national digital priorities.

This converts a scarce resource into a competitive allocation. Operators must demonstrate superior efficiency to win capacity, which raises the standard of the entire sector rather than simply limiting its size.

The approach is philosophically consistent with how Singapore manages vehicle ownership, land and water: scarce resources are allocated through explicit mechanisms rather than through queuing or political discretion.

What does the green data centre roadmap require?

The roadmap committed to enabling additional capacity conditional on improved energy efficiency across the sector and on the availability of low-carbon electricity, with further capacity contingent on green energy supply materialising.

Practical measures include raising operating temperatures in server halls, which reduces cooling load substantially, adopting more efficient cooling technologies including liquid cooling, and improving utilisation of existing capacity.

The temperature measure is deceptively significant. Data centres were historically operated far cooler than modern equipment requires, and raising the setpoint across an entire national sector releases capacity without building anything.

What has happened in the region as a result?

Johor in southern Malaysia and the Riau Islands in Indonesia have attracted very large data centre investment, positioned close enough to Singapore to serve overlapping markets while offering land, power and water at a fraction of the cost.

This has produced a functional regional cluster: latency-sensitive and sovereignty-sensitive workloads remain in Singapore, while bulk compute and storage move across the border, connected by high-capacity links.

The arrangement mirrors the manufacturing division of labour described in the precision manufacturing case study, where Singapore retains the activities that require its specific advantages and neighbours take the rest.

How does artificial intelligence demand change the picture?

Artificial intelligence training and inference workloads consume far more power per rack than traditional computing, requiring liquid cooling, higher-density electrical infrastructure and substantially more energy per square metre.

That intensifies every constraint Singapore already faced. A facility supporting artificial intelligence workloads may consume several times the power of a conventional data centre of the same size.

The policy response has been to prioritise capacity for workloads with the strongest economic and strategic justification, and to invest in regional electricity import arrangements that could relieve the underlying energy constraint over time.

How is Singapore addressing the underlying energy constraint?

The country is pursuing regional electricity imports through undersea interconnection with neighbouring countries, alongside solar deployment on rooftops and reservoirs, and exploration of low-carbon alternatives.

Regional interconnection is the most consequential of these. Importing renewable electricity from neighbours with land and resources Singapore lacks would relieve the constraint that limits every energy-intensive industry.

The obstacles are commercial and political: long-term contracts, cross-border regulatory alignment, transmission investment and the willingness of exporting countries to commit generation to a neighbour rather than their own growth.

What efficiency standards apply?

Singapore has developed data centre energy efficiency standards and certification, addressing cooling efficiency, operating temperature, equipment specification and monitoring.

Raising server hall operating temperatures in a tropical climate is the highest-impact intervention, because cooling represents a very large share of a data centre’s non-computing energy use.

The standards approach also creates export potential, since a tropical data centre efficiency standard is directly relevant to the many countries facing the same climate conditions.

What does this mean for cloud customers?

Enterprises should expect capacity in Singapore to remain constrained and priced accordingly, with cloud providers steering non-sensitive workloads toward regional alternatives.

For workloads with data residency requirements, Singapore capacity remains available but at a premium, and architecture decisions should distinguish clearly between what genuinely must stay and what does not.

The practical planning implication is to classify workloads by residency requirement and latency sensitivity early, because retrofitting that distinction into an existing architecture is expensive.

What is the sustainability standard for tropical data centres?

Singapore has developed and promoted standards specific to tropical operating conditions, addressing the fact that most international data centre efficiency guidance assumes temperate climates.

Cooling in a hot, humid climate is inherently more energy-intensive, so efficiency benchmarks designed for northern Europe or North America are not directly applicable and can be misleading.

Establishing a tropical standard also positions Singapore to influence practice across Southeast Asia, South Asia and other tropical regions building data centre capacity rapidly.

How do data centres affect the electricity market?

Large continuous loads change grid planning requirements, affect generation capacity margins and influence long-term power purchase arrangements, particularly when a single facility consumes tens of megawatts.

Data centre operators increasingly seek long-term renewable power agreements, which in Singapore is constrained by the limited domestic renewable resource and therefore depends on imports or certificates.

The interaction between industrial policy, energy policy and climate commitments in this sector is unusually direct, which is why capacity decisions are made at national policy level rather than by planning officials.

What does this mean for the region’s digital economy?

The regional outcome is a distributed model where Singapore hosts latency-sensitive, regulated and network-critical infrastructure while neighbouring markets host bulk capacity.

That arrangement works only if cross-border connectivity is excellent and regulatory frameworks on data transfer are compatible, both of which require ongoing intergovernmental coordination.

For businesses it means designing architecture around a regional footprint rather than a single country, which is now standard practice for any serious digital operation in Southeast Asia.

What alternatives to conventional cooling are emerging?

Liquid cooling, immersion cooling and heat reuse are being deployed to handle higher rack densities while reducing the energy consumed by air conditioning systems.

Liquid cooling is particularly relevant for artificial intelligence workloads, whose power density exceeds what air cooling can practically handle at scale.

Heat reuse is harder in a tropical climate, since there is limited demand for low-grade waste heat in a country that never needs building heating, which removes an option available in colder markets.

How should companies plan capacity in this environment?

Plan on the assumption that Singapore capacity is scarce, expensive and allocated on efficiency grounds, and design workloads so that only what genuinely needs to be there is placed there.

Building a regional architecture with clear residency and latency classifications is far cheaper done at design time than retrofitted once capacity constraints bite.

Long-term contracts have become more valuable in a constrained market, and organisations that secured capacity early are in a materially better position than those procuring now.

What is the broader policy lesson?

The lesson is that a government willing to constrain a profitable growing industry on resource grounds, and then to reopen it under conditions, can raise the standard of an entire sector rather than simply limiting its size.

Most jurisdictions either permit unconstrained growth until a crisis forces intervention, or impose blunt limits without a mechanism for allocating what capacity exists.

The conditional allocation approach is more administratively demanding and produces better outcomes, which is the same pattern visible in Singapore’s vehicle, land and water policies.

How do operators compete for allocated capacity?

Operators compete on demonstrated energy efficiency, commitment to low-carbon energy sourcing, economic contribution including employment and investment, and alignment with national digital priorities.

This turns a permitting process into a competitive tender, rewarding operators who have invested in efficiency rather than those who applied first.

It also gives authorities visibility into what the sector is capable of, since applicants must document performance levels they might otherwise never disclose.

How does this affect Singapore’s climate targets?

Data centre electricity consumption is a material component of national emissions, so sector growth and decarbonisation commitments are directly in tension until low-carbon supply expands.

The policy answer has been to link additional capacity to green energy availability, effectively making sector growth contingent on progress in energy supply rather than treating them separately.

That coupling is analytically honest and administratively difficult, and it is why the roadmap describes conditional rather than committed capacity.

Frequently Asked Questions