Who Is Really Feeding Off Johor’s Data Center Boom?

By Wan Ge in Tokyo

The IMF’s October Regional Economic Outlook for Asia and the Pacific contains a passage worth reading closely:

Malaysia and Singapore have increased exports as assemblers and packagers of technology products, while strong investment in data centers in China, India, Japan, Indonesia and Malaysia has also supported growth. Data-center investment has generated broader demand for intermediate goods from regional suppliers, including Thailand and Vietnam.

Johor, Malaysia, is one of the clearest examples on the investment side of this chain. It has become one of the most concentrated data-center construction markets in the Asia-Pacific.

The IMF’s description captures a straightforward industrial logic: the economic impact of data-center investment does not stop where the facilities are built. Projects in Johor also generate demand for components and intermediate goods supplied elsewhere in the region.

But the IMF’s description is necessarily broad. Along this supply chain, many economies may receive orders, yet the type, value and sustainability of those orders differ significantly.

Johor: Huge Capacity on Paper, but Power Is the Bottleneck

Under the IMF’s statistical framework, Malaysia currently has more than 1,000 MW of operating data-center capacity, with over 1,500 MW under construction.

A more detailed estimate from real-estate services firm Cushman & Wakefield shows that, as of the first half of 2026, Johor had 1,110 MW of operating IT capacity, 602 MW under construction and another 2,486 MW planned. That brings the combined operating, under-construction and planned total to 4,198 MW, placing Johor first in the Asia-Pacific under this methodology.

The two datasets come from different institutions and use different time periods and scopes, so they should not be directly compared. Here, megawatts refer to the IT load power capacity of data centers, not the physical size of the facilities.

Planned capacity, however, does not translate immediately into usable computing capacity.

Johor’s colocation data-center vacancy rate is only 0.7%, compared with 4.9% in Singapore, 20.5% in Jakarta and 23.3% in Bangkok. Such a low vacancy rate means that most completed capacity is effectively taken as soon as it comes online, leaving new customers with little readily available space.

In the first quarter of 2026, Malaysia approved 33 data-center and cloud-computing projects with total investment of MYR34.6 billion, or roughly US$8.5 billion. Yet by July, among 27 projects approved under the national Data Centre Task Force framework, only one had entered operation, while the other 26 remained under construction.

It is easy to assume that land or capital is the biggest constraint on computing expansion. In Johor, however, the tighter bottleneck is electricity.

As of December 2024, 38 data-center projects in Malaysia had signed power-supply agreements with a combined maximum demand of 5,900 MW, while only 405 MW of actual load had been put into use.

This earlier dataset should not be compared directly with the 2026 capacity figures above, but it clearly illustrates the gap between contracted electricity demand and actual consumption.

Wood Mackenzie has identified the transmission and distribution grid as the primary constraint on Johor’s expansion and estimates that data centers could account for around 40% of the state’s electricity consumption by 2035.

In February 2026, the Anwar government tightened approvals, temporarily suspending ordinary new data-center projects and prioritizing facilities serving AI and other frontier technologies.

This leads to an important conclusion: Johor’s computing blueprint may be enormous on paper, but equipment purchases and component orders will not arrive all at once. Demand will be released in step with the development of power-grid infrastructure.

Thailand: A Midstream Component Supplier Tied to the Global AI Cycle

The IMF identifies Thailand as one of the economies benefiting from spillover demand.

Thailand’s place in the chain is primarily as a supplier of intermediate electronic components used in servers.

A September 2026 industry report by Kasikorn Securities described AI and data centers as key growth engines for Thailand’s electronics sector, supporting demand for power electronics, high-end printed circuit boards, PCB assemblies and semiconductor components.

Thailand’s stronger electronics exports have benefited from the global construction boom in hyperscale computing clusters.

At the company level, capacity utilization and margins at Delta Electronics Thailand (DELTA) have been recovering. HANA has entered a capacity-ramp phase for AI-related products, while KCE has benefited from stronger PCB sales and higher average selling prices.

But there is a clear limit to this benefit.

Thailand supplies general electronic components rather than equipment dedicated specifically to Johor data centers. Kasikorn Securities has also highlighted risks from weaker-than-expected recovery in non-AI end markets such as automobiles and new-energy products, while elevated costs for copper and copper-clad laminates continue to squeeze margins.

Krungsri Research has also noted that global data-center construction drove demand for hard-disk drives up by more than 40% in 2024, benefiting Thailand as one of the world’s major HDD manufacturing bases.

Yet the industry faces a long-term substitution threat from solid-state drives.

Krungsri previously forecast that the value of Thailand’s integrated-circuit exports could rise 29.5% to 30.5% in 2025, even as production volume potentially contracted.

That suggests Thailand’s electronics export growth may be driven more by product upgrading and higher prices than by simple capacity expansion. KCE’s rising average selling prices are consistent with this pattern.

In simple terms, Thailand is benefiting from higher-value component orders, but its fortunes remain tied to the global AI cycle. Johor is only one source of demand among many.

Vietnam: An Assembly Hub Where Higher Exports Also Mean Higher Imports

Vietnam occupies a different position in the supply chain.

Its core role is final assembly rather than the production of intermediate components.

In the first eight months of 2026, Vietnam exported around US$101 billion worth of computers, electronics, mobile phones and related components, up 51% year on year. Mobile-phone component exports alone approached US$12 billion.

During the same period, imports in the same broad category reached US$161 billion, up 68%, producing a trade deficit of roughly US$60 billion.

It is important to note that this category includes mobile phones and consumer electronics and is not entirely related to AI or data centers. The exact share represented by servers and other AI-related products requires further verification.

Still, the import structure is revealing. A large share consists of chips, memory, processors, display panels and circuit boards — evidence of a model based on importing components, assembling locally and re-exporting finished products.

The Vietnam Electronic Industries Association said in June 2026 that AI servers, computing equipment and high-performance components were among the fastest-growing segments, with related companies reporting order growth of 15% to 30% year on year. Many order books were already filled through the end of the third quarter.

But the association also acknowledged that low localization remains the industry’s biggest weakness, with many specialized production stages still heavily dependent on imported materials and components.

The Vietnamese government has recognized the problem.

At a September meeting, Deputy Prime Minister Pham Gia Tuc called for stronger domestic supply chains and a reduction in the electronics trade deficit, instructing the Ministry of Industry and Trade to monitor large importers and analyze the structure of imported goods.

But building a domestic supply chain takes time. In the short term, this model will be difficult to change.

That creates a revealing contrast.

In a single sentence, the IMF lists Thailand and Vietnam together as beneficiaries of intermediate-goods demand. Yet Thailand is itself an intermediate-goods supplier, exporting electronic components directly. Vietnam mainly receives assembly orders — and the more assembly orders it receives, the larger its bill for imported components becomes.

Both economies benefit from computing investment, but the quality of those benefits is very different.

The Bigger Picture: The Benefits Are Not Shared Equally

Johor has a huge pipeline of planned computing capacity, but its electricity bottleneck is stretching out the timetable for new orders. At the same time, tighter approvals and policy preference for AI projects mean future demand is likely to become increasingly concentrated in high-performance computing.

Thailand is winning component orders through its midstream manufacturing base, but its performance remains tied to the global AI cycle and faces pressure from raw-material costs and weaker traditional end markets.

Vietnam is gaining assembly orders, but its weak domestic supply chain means export growth is accompanied by a sharp rise in imports.

Johor’s 0.7% colocation vacancy rate shows that existing capacity is effectively fully booked. New capital must either wait for projects under construction to come online or look toward other Southeast Asian markets.

Large investment announcements worth tens of billions of ringgit can easily generate optimism.

But the transmission of orders through the computing supply chain is far less immediate than investment announcements make it appear.

So, Who Is Johor’s Data Center Boom Really Feeding?

The IMF’s short answer is Thailand and Vietnam.

A closer look at the industrial chain gives a more nuanced answer:

Thailand is receiving higher-value component-supply orders. Vietnam is receiving assembly orders — along with a large import bill for the components required to fulfill them. Both benefit, but the quality of the gains differs substantially.

Meanwhile, Johor’s own electricity bottleneck, saturated existing capacity and tighter approval regime will determine how quickly demand is released across the entire supply chain.

In other words, the real pace of the boom is likely to be considerably slower than the investment pipeline on paper suggests.