Malaysia's Data Center Capacity Set to More Than Double by End of 2026 as AI Investment Surges

Malaysia Takes the Lead in Southeast Asia's Building Boom

Malaysia's data center capacity is projected to more than double by the end of 2026, with the country accounting for more than two thirds of the capacity under construction across Southeast Asia's five largest economies. Research from real estate group Jones Lang LaSalle (JLL) places Malaysia as the fastest growing data center market in the Asia-Pacific region.

The expansion is drawing Microsoft, Google, Amazon, ByteDance, Nvidia and Alibaba into a market already hosting approximately 51 operational colocation data centers. Colocation facilities provide space, electricity, cooling and connections for computing equipment belonging to other companies. Their growth shows how quickly Malaysia is becoming a base for the infrastructure supporting cloud services and artificial intelligence.

Malaysia's data center market is valued at USD 6.14 billion in 2025 and projected to reach USD 11.40 billion by 2031, with a compound annual growth rate of 10.86%. That rate describes the steady annual increase that would take the market from its starting value to its projected total. The forecast represents an increase of USD 5.26 billion over six years.

The pace of construction is outstripping established regional hubs such as Singapore and Hong Kong. Malaysia's rise also coincides with a much larger global investment cycle: JLL expects nearly 100 gigawatts of new data center capacity worldwide between 2026 and 2030, effectively doubling global capacity. A gigawatt equals 1,000 megawatts, the unit commonly used to describe the power capacity of these facilities.

As of February 2, 2026, the opportunity comes with growing pressure on electricity, water, land and skilled workers. Malaysia is attracting the buildings and equipment needed for AI, while its government also wants domestic businesses and workers to gain a larger role in developing the technology.

Why Companies Are Choosing Johor and Cyberjaya

Singapore's three year moratorium on new data center projects helped redirect investment toward Malaysia. The city state subsequently introduced strict green specifications that limited expansion in the established regional hub. As cloud adoption accelerated during the pandemic, companies looked for nearby locations that could retain access to Singapore's financial and technology ecosystem.

Johor offered a practical alternative. The southern Malaysian state is separated from Singapore by a narrow strait and has available land and lower costs. Its location also helps companies serve Southeast Asian customers with low latency, meaning a short delay between a request and a response from a computer system.

Cyberjaya, Malaysia's designated technology city, provides another major concentration of infrastructure. It hosts more than 22 existing data centers, with nine additional facilities planned. Johor's proximity to Singapore and Cyberjaya's established base give investors two important locations within the country.

Construction costs in Malaysia range from USD 8 million to USD 10 million per megawatt, substantially below Singapore's costs. Operating expenses have also helped attract investment, although changes to electricity pricing are making some projects more expensive.

Malaysia introduced new power tariffs targeting data centers in July 2025. The increases raise energy costs by 10% to 14%, and facilities exceeding 100 megawatts fall into ultra high voltage categories that could bring additional annual costs of USD 15 million to USD 20 million per facility.

JLL forecasts that global construction costs will rise by 6% to USD 11.3 million per megawatt in 2026. Malaysia is experiencing annual cost increases of 5% to 7%, driven by inflation and regulatory changes. Its cost advantage remains an attraction, but investors must account for a more expensive operating environment.

Water and Electricity Are Becoming Constraints

The International Energy Agency (IEA) identifies Malaysia as a pressure point in global energy systems. Data centers could account for as much as one fifth of the country's electricity demand growth by 2030, placing Malaysia alongside Japan and the United States among countries where AI computing will strongly reshape energy consumption.

Water shortages in Johor and Selangor have already led state authorities to slow approvals for new data center construction. Cooling systems can consume large quantities of water, with a single facility optimized for AI potentially using up to 20 lakh liters daily, equivalent to 2 million liters.

Microsoft internally projected that its global water needs for data centers could rise by 150% by 2030. The company subsequently revised its estimates downward after introducing conservation technologies. The pressure in Malaysia connects the pace of construction with a basic local requirement: enough water must be available to keep equipment cool.

Malaysia's electricity grid relies heavily on coal and natural gas. The government plans to add six to eight gigawatts of generation powered by natural gas by 2030 specifically to address data center demand. Supplying that demand through fossil fuels creates tension with the country's commitments to reach net zero emissions and risks a lasting dependence on infrastructure producing substantial carbon emissions.

Grid connection waits now exceed four years in primary markets, encouraging operators to consider power supplied directly to their facilities and batteries installed on site. Such arrangements can reduce dependence on the timing of a public grid connection. Natural gas is expected to help relieve grid constraints, although some hyperscalers reject these options because of their sustainability goals. Hyperscalers are companies operating computing infrastructure at very large scale.

Malaysia's Green Data Center Guidelines, under the Malaysia Digital Economy Corporation (MDEC), require Environmental Impact Assessments for large developments and encourage infrastructure that uses energy efficiently. These assessments examine how a project would affect its surroundings, bringing environmental requirements into decisions about new facilities.

A Sovereign AI Cloud and the Question of Control

Prime Minister Anwar Ibrahim's government has actively sought digital investment through the 2021 Artificial Intelligence Roadmap. The roadmap treats a national AI ecosystem as essential to Malaysia's ambition to achieve developed nation status by 2030.

The 2026 budget allocates RM 2 billion, approximately USD 490 million, to building a sovereign AI cloud. The allocation forms part of a broader RM 5.9 billion commitment for AI research, development, commercialization and innovation initiatives.

A sovereign AI cloud would allow AI models to be trained, stored and deployed on Malaysian soil under domestic oversight. Hosting international companies brings infrastructure investment, while domestic control over computing resources gives Malaysia a means to pursue its own AI priorities.

The scale of Malaysia's allocation is small compared with major international commitments. OpenAI has announced intentions to invest USD 1.4 trillion in computing infrastructure over the next decade. Meta has committed more than USD 600 billion to infrastructure in the United States by 2028. Trump announced private sector AI infrastructure investments of up to USD 500 billion in January 2025.

Critics of Malaysia's approach argue that attracting foreign hyperscalers could leave the country primarily providing infrastructure for technologies developed elsewhere. The sovereign cloud initiative seeks to give Malaysia a more active role, but the balance between hosting foreign systems and building domestic capabilities remains central to its strategy.

Many existing Malaysian data center commitments focus on inference, the stage when a trained AI model responds to requests or performs tasks. Training is the earlier process in which a model learns patterns from data, requiring substantially more computing power and specialized chips.

Inference facilities generate economic value, but the concentration on deployment places Malaysia outside the group described as "compute-north" countries that shape the development of advanced models. Training facilities typically bring greater transfers of technological knowledge and more employment requiring advanced skills. Malaysia's ambitions therefore depend on what happens inside its facilities as well as how much capacity is built.

Microsoft and NTT DATA Expand Their Footprints

Microsoft plans to establish a second Malaysian cloud region, designated "Southeast Asia 3," in Johor Bahru. It will feature three availability zones and is projected to become operational within two to three years. Availability zones are distinct groups of computing infrastructure within a cloud region, allowing services to operate across separate facilities.

The expansion comes as Microsoft faces severe capacity constraints globally. Kevin Scott, Microsoft's chief technology officer, described the difficulty of building infrastructure quickly enough since ChatGPT's launch:

massive crunch is probably an understatement

Microsoft's Johor investment illustrates the pressure to secure new capacity while demand continues to grow. Across Malaysia, the availability of land and electricity is increasingly determining where large technology companies can establish or expand facilities.

NTT DATA acquired land worth USD 88.5 million from Tropicana Corporation Berhad in August 2024 for a campus containing six data center buildings in Johor. Its first facility is expected to come online by 2027.

Google, Amazon, ByteDance, Nvidia and Alibaba have also committed substantial resources to Malaysian facilities. Their presence places Malaysia within the infrastructure plans of major companies from both the United States and China, while increasing competition for the resources needed to build and operate those facilities.

Jobs, Engineering and Bridge Data Centres

The Asia-Pacific Data Center Association projects that Malaysia's AI industry will create 30,900 jobs annually by 2030. Those opportunities include network engineers, cloud infrastructure specialists and professionals responsible for information and communications technology security.

Data center employment follows a different pattern from the scale of the buildings themselves. Research indicates that these facilities create the fewest jobs per square foot of any major facility type. Construction can employ thousands of workers, while operations may require fewer than 200.

Concern is growing that foreign workers could fill research and engineering positions while Malaysians receive temporary construction jobs and lower level maintenance roles. The wider AI employment forecast and the smaller operating workforces make access to technical training a central issue for local participation.

Bridge Data Centres (BDC), backed by Bain Capital, illustrates the engineering capabilities being developed within Malaysia's data center industry. BDC operates more than 300 megawatts of live capacity in Malaysia, making it the country's largest operator.

The company has pursued vertical integration, bringing more of the work required to design and run its facilities into its own operations. It has accumulated approximately 600 patents related to data center designs and systems and maintains internal engineering capabilities instead of outsourcing operations.

Eric Fan, chief executive officer of Bridge Data Centres, described the company's identity in technology rather than property terms:

not a real estate play

Fan also described its focus as:

frontier solutions

BDC expects to almost double its Malaysian workforce in 2026 as new campuses begin operating in Johor and the Klang Valley. It has established education partnerships with institutions including the Singapore Institute of Management to develop regional talent.

Fan acknowledges that AI demand is also increasing costs. Designs supporting AI can cost 15% to 20% more than designs used before AI became a major driver of demand because they require advanced liquid cooling. That investment connects the industry's need for specialized engineering with its growing financial and resource demands.

Chip Restrictions Add Pressure to Malaysia's Strategy

Malaysia's growth as a data center hub intersects with competition between the United States and China over technology. As one of China's largest trading partners in ASEAN, Malaysia faces pressure to maintain its economic relationship with Beijing while responding to Washington's tightening controls on exports of advanced graphics processing units, or GPUs.

GPUs provide computing power used in AI workloads. Restrictions on access to advanced chips, alongside potential tariffs on data center equipment, could increase the cost of Malaysia's expansion and limit access to the hardware required for some projects.

Malaysia's neutrality has attracted companies seeking to spread supply chains across more locations. Chinese firms have also viewed Malaysia as a possible route to advanced chips through third countries, creating compliance challenges for Malaysian authorities as they balance investment opportunities and diplomatic relationships.

The recent US-Malaysia Reciprocal Trade Deal may complicate the National Semiconductor Strategy. That strategy aims to maintain Malaysia's position in midstream semiconductor production while gradually moving into more complex activities within the chip supply chain.

Compliance with US export restrictions could limit Malaysia's ability to act as a neutral intermediary, weakening the supply chain autonomy Putrajaya initially sought. The country's data center ambitions and semiconductor policy are consequently connected through access to the computing equipment that AI requires.

Security concerns also extend to the electricity systems powering these facilities. The IEA warns that cyberattacks on energy utilities have tripled over the past four years and become more sophisticated through AI applications. As data centers become critical infrastructure, their exposure to physical and cyber threats requires coordination among technology companies, electricity providers and national security agencies.

Regional Rivals Use Malaysia as a Benchmark

The Philippines is preparing its own expansion through PLDT Inc., which plans a hyperscale data center with 100 megawatts of capacity in Cavite. The project will be PLDT's 12th facility, with construction scheduled to begin in 2026 and completion expected by 2028.

Manuel V. Pangilinan, PLDT's chairman, described the company's capacity ambition by directly referring to Malaysia:

to be at least equal to the capacity of Malaysia

Indonesia, Thailand and Vietnam are also accelerating data center investment. Malaysia's early expansion and proximity to Singapore provide a head start as neighboring countries seek a larger share of the same regional demand.

Asia-Pacific is projected to account for 40% of global data center capacity by 2030. The region is expected to reach approximately 30 gigawatts of total capacity, supported by USD 800 billion in investment.

Malaysia's construction pipeline gives it a prominent position within that expansion. Its share of more than two thirds of capacity being built across Southeast Asia's five largest economies shows how heavily current regional construction is concentrated in one country, even as rivals announce new projects.

Can AI Demand Keep Up With Construction?

Data centers and GPU clusters have been built at a pace since 2023 that could exceed demand for AI in the near term. If workloads grow more slowly than the infrastructure intended to handle them, operators could face consolidation and repricing.

JLL expects a shift from AI training toward inference workloads in 2027. That change would move more demand from large centralized computing clusters to distributed regional hubs, potentially absorbing excess capacity in locations such as Malaysia that serve wider Southeast Asian markets.

Training concentrates substantial computing power on developing models. Inference brings those models into everyday use, spreading computing demand across the locations where services are delivered. Malaysia's investment in deployment infrastructure could benefit from that shift, even as it seeks a larger role in developing AI technology.

The forecasts extend over several years: Malaysia's capacity is expected to more than double by the end of 2026, NTT DATA's first Johor facility is expected in 2027, and JLL anticipates the change toward inference in 2027. Delivery depends on securing power, water, equipment and workers within those schedules.

The milestones already announced show how the investment cycle has developed and where the next major targets fall:

• 2021: Malaysia's Artificial Intelligence Roadmap sets out the ambition to build a national AI ecosystem supporting developed nation status by 2030.
• Since 2023: Rapid construction of data centers and GPU clusters creates concern that capacity could exceed demand in the near term.
• August 2024: NTT DATA acquires land worth USD 88.5 million for its Johor campus.
• January 2025: Trump announces private sector AI infrastructure investments of up to USD 500 billion.
• July 2025: Malaysia introduces new electricity tariffs targeting data centers.
• 2026: Malaysia's capacity is projected to more than double by year end, BDC expects to almost double its Malaysian workforce, and PLDT plans to begin construction in Cavite.
• 2027 to 2028: NTT DATA expects its first Johor facility in 2027; PLDT targets completion in 2028; Meta's US infrastructure commitment extends to 2028.
• 2030: Malaysia targets developed nation status, plans six to eight gigawatts of additional gas generation, and faces projected growth in AI employment and data center electricity demand.
• 2031: Malaysia's data center market is projected to reach USD 11.40 billion.

Key Points

• Malaysia's data center capacity is projected to more than double by the end of 2026, accounting for more than two thirds of regional capacity under construction.
• The market is valued at USD 6.14 billion in 2025 and forecast to reach USD 11.40 billion by 2031.
• Water shortages are slowing approvals in Johor and Selangor, while new electricity tariffs raise energy costs by 10% to 14%.
• Malaysia has allocated RM 2 billion for a sovereign AI cloud within a broader RM 5.9 billion AI commitment.
• Microsoft plans a second Malaysian cloud region in Johor Bahru; BDC operates more than 300 megawatts of live capacity.
• Access to advanced chips, skilled employment and electricity supply will shape how much Malaysia gains from the expansion.
• Regional competitors are expanding, while JLL expects demand to shift toward distributed AI inference facilities in 2027.