In a single year, Malaysia is on track to more than double its data centre capacity — from roughly a gigawatt to around 2,055 megawatts by the end of 2026, on JLL's numbers. That is close to a 70% growth rate in twelve months, with several thousand more megawatts already queued behind it. Numbers like that do not happen quietly. They happen because YTL, Gamuda, Sunway and half the KLCI decided data centres were the trade of the decade.
Most of the coverage treats this as a property and construction story. It is bigger than that. These buildings are the physical layer of the AI economy — the place your Claude query, your Copilot summary and every hyperscaler's model actually runs. Understanding who is building them, and where the value really sits, is worth doing properly.
The numbers behind the boom
The scale is genuinely large for a country of 34 million people. Government approvals ran to roughly RM144 billion across more than 140 data-centre projects between 2021 and mid-2025. Malaysia has pulled in tens of billions of US dollars of investment over four years, and analysts see the sector's value climbing steeply through 2030. Johor Bahru alone has hundreds of megawatts completed, well over a gigawatt under construction, and thousands more in the pipeline.
What changed? Three things collided: an explosion in AI compute demand, Singapore's long squeeze on its own data-centre growth, and Malaysia's combination of land, power and subsea-cable connectivity right across the causeway.
The players: operators, developers and builders
It helps to separate the roles. Some companies operate data centres (they own the facility and sell capacity). Some develop them (they assemble the land, power and design). And some build them (they win the construction contracts). Malaysia's big names sit in different spots, and a few straddle several.
YTL Power
Completed Malaysia's first Nvidia-powered AI data centre in Kulai, Johor (Oct 2025), running GB200 Grace Blackwell chips. The 600MW YTL Green Data Center Park is a ~RM10 billion partnership with Nvidia; YTL AI Cloud is now live.
Gamuda
Won a RM1.72 billion hyperscale build in Port Dickson for a US tech MNC (April 2026), on top of a portfolio spanning Cyberjaya and Elmina. Also sold a 389-acre Springhill parcel for data-centre development.
Sunway Construction
Landed a RM1.75 billion hyperscale contract in Bandar Serendah, Selangor (April 2026) and a RM570 million Johor job — part of a run of data-centre wins that now stretches to Singapore.
IJM, Mah Sing, Bridge & more
IJM is landing hyperscale build jobs; Mah Sing and others are converting land into data-centre parks; specialist operators such as Bridge Data Centres anchor Johor. The Johor-Singapore Special Economic Zone ties it together.
YTL: the one that owns the whole stack
YTL Power is the standout because it plays every role at once — land, power, development and operation. In October 2025 it completed Malaysia's first Nvidia-powered AI data centre in Kulai, Johor, running Nvidia's GB200 Grace Blackwell chips: the same tier of hardware that frontier labs in California train on. It anchors the 600MW YTL Green Data Center Park, part of a roughly RM10 billion tie-up with Nvidia, and feeds the YTL AI Cloud service.
The reason this matters more than a typical hyperscale shell is control of the full stack — including power. YTL is pairing the campus with its own large-scale solar generation plus grid backup, which is exactly the kind of energy answer the boom will live or die on.
Gamuda and Sunway: the builders cashing in
For the big contractors, data centres have become a core order book. Gamuda won a RM1.72 billion hyperscale build in Port Dickson for a US technology MNC in April 2026, adding to developments in Cyberjaya and Elmina, and separately monetised land by selling a 389-acre Springhill parcel for data-centre use. Sunway Construction landed a RM1.75 billion hyperscale contract in Bandar Serendah, Selangor, and a RM570 million Johor job in the same window — part of a streak that now reaches into Singapore.
IJM is winning similar work, and property developers such as Mah Sing are converting landbanks into data-centre parks. Specialist operators like Bridge Data Centres anchor the Johor cluster. It is, in short, an entire supply chain reorganising around one demand curve.
Why Johor became the epicentre
Johor now holds around 80% of Malaysia's operational data-centre capacity. Proximity to Singapore is the whole story: as Singapore constrained its own build-out, demand spilled directly across the strait to a neighbour with cheaper land, available utilities and the same subsea cables. The Johor-Singapore Special Economic Zone has since formalised the relationship. For a fuller read on the drivers, White & Case's analysis of what is propelling the boom is a good primer.
The catch: power, water and a moratorium
None of this is free. Data centres are enormously hungry for electricity and water — a single 100MW facility can consume millions of litres of water a day for cooling. That collides with Malaysia's net-zero-by-2050 commitment and its National Energy Transition Roadmap. By early 2026 the government had moved to restrict new approvals for non-AI data centres, precisely because of the strain on utilities. We cover that tension in depth in what the data centre boom means for your business.
The part that actually decides the payoff
Here is the uncomfortable strategic point. Building the world's data centres makes Malaysia a landlord to the AI economy. Being a landlord is fine — the rent is real. But the outsized returns go to whoever uses the compute: the companies and workers who build products, automate work and ship AI on top of it. A gigawatt of GPUs in Kulai does nothing for a firm in Petaling Jaya that has not trained anyone to use AI.
That is the gap worth closing. The infrastructure is being handled by people with balance sheets in the billions. The capability to exploit it is a training and talent question — and one every Malaysian organisation can act on now. See the careers the boom is creating, and the applied skills in our AI Engineering programme and AI Automation training, both HRD Corp SBL-KHAS claimable via the HRDC route.