Every AI model, cloud service and streaming app runs on servers in a building somewhere. Those buildings need large plots of land, a lot of electricity, water or advanced cooling, and fast connections to the rest of the internet. In Southeast Asia, a growing share of them are going up in Johor.

This piece looks at why that happened, what is now holding it back, and how investors can gain exposure through listed property trusts.

From Singapore's pause to Johor's boom

Singapore was the region's original data centre hub, thanks to its connectivity, political stability and reliable power. But data centres use a lot of electricity, and in 2019 the government paused new approvals. When it reopened in 2022, it did so selectively, favouring energy-efficient projects.

Demand did not pause. Johor, minutes across the causeway, offered land, power connections and lower costs, and developers moved quickly. The Johor–Singapore Special Economic Zone, agreed in January 2025, made the pairing more formal. Malaysia's federal government has said it expects much of the country's data centre capacity to be concentrated in Johor by 2030.

How big it has become

The scale is striking. Knight Frank's Data Centre Atlas 2026 put Johor's incoming pipeline at about 8,542 megawatts, with colocation vacancy of just 0.7%, against 4.9% in Singapore. TechRepublic's summary of the report also notes that Malaysia approved RM34.6 billion of data centre and cloud investment across 33 projects in the first quarter of 2026 alone.

Low vacancy and a huge pipeline sound contradictory, but they describe the same situation: capacity being built is largely spoken for before it is finished, often by the large cloud and AI companies.

Power and water now set the pace

The constraint is no longer land. Since late 2025, Johor has halted new approvals for Tier 1 and Tier 2 data centres, the less resilient designs, citing water use. The state's coordinating committee now favours Tier 3 and Tier 4 projects that rely on air cooling and has set a water usage effectiveness threshold of 1.8, the same as Singapore, according to Bernama. Some developers have been directed to use recycled water from sewage treatment for cooling.

At the federal level, ministers have said approvals will depend on spare power and water capacity, with residents prioritised. Analysts have pointed to transmission and distribution infrastructure, more than generation, as the main bottleneck.

For investors, that cuts both ways. Tighter approvals slow new supply, which supports rents for existing operators. But they also cap growth for operators that cannot secure power.

Where REITs come in

Data centres are expensive to build, and once they are leased to large tenants on long contracts, they produce steady rent. That makes them a natural fit for REITs. Singapore's exchange lists specialist data centre REITs as well as diversified industrial REITs with data centre exposure, owning facilities across Asia, Europe, Australia and the United States.

The same metrics from our REIT guide apply, with a few extras specific to data centres:

  • Tenant concentration. A few hyperscale tenants can account for much of the income. Their creditworthiness is a strength, but losing one hurts.
  • Who pays for power. Leases that pass electricity costs to tenants protect the landlord from energy price swings. Others do not.
  • Technical obsolescence. AI workloads need more power per rack and often liquid cooling. Older facilities may need heavy spending to stay relevant.
  • Land tenure. Leasehold land in Singapore and Malaysia limits the asset's life, so check how much time is left.
  • Sponsor pipeline. A sponsor developing new facilities can offer the REIT acquisitions over time.

Three themes in one asset

A data centre REIT is a real estate investment, an income product and a technology bet at once. That is its attraction and the reason to keep position sizes sensible. Valuations can run ahead when AI enthusiasm is high, and interest rates still matter as they do for any REIT.

For more on the property fundamentals, see our real estate guide. For how groups organise their interests in assets like these, read how investment holding companies work. And for the wider view of AI across investment and business, start at our AI section.