The Growth of Data Centers and the Energy Cost of Progress
54 data centers operating in 2024 have become a pillar of the Malaysian digital economy. The forecast of reaching 81 units by 2035 indicates an annual growth rate of over 5%. This development is driven by the demand for computing power for training artificial intelligence models, which require constant and high-density energy flows. The energy needs have not been met by renewable sources, but by gas turbines, which have absorbed a growing share of national production. The increase in demand has prompted Petronas to expand extractions from offshore fields, despite global climate pressure. The data indicate that the digital infrastructure is not a neutral entity, but a system that requires primary inputs with a measurable thermodynamic impact.
Consequently, the economic growth project clashes with the national goal of phasing out fossil fuels by 2050. Gas, which currently covers 78% of the electricity demand for data centers, represents the fastest technical solution, but not a sustainable one in the long term. The Malaysian energy system has entered a phase of forced transition, where the latency of renewable infrastructure cannot keep pace with the speed of expansion of the digital sector. The cost is not only environmental, but also strategic: the country is faced with the choice between immediate competitiveness and consistency with global commitments.
The Energy Dependency Mechanism
The operational structure of Malaysian data centers relies on a model of reliance on gas, which guarantees stability and immediate scalability. Liquefied natural gas (LNG) has been chosen as the primary source due to its low installation cost compared to large-scale solar or wind power plants. However, the average energy consumption of a modern data center is approximately 10 megawatts per unit, with peaks exceeding 15 MW. With 81 units, the total load would exceed 1.2 gigawatts, equivalent to a small nuclear power plant. The current Malaysian electricity grid is not designed to handle such high peaks without resorting to flexible backup sources, such as gas engines.
The supply system has been strengthened by increased offshore extraction, with Petronas increasing LNG production from deposits in Sarawak. This operation has made it possible to meet 40% of the internal demand for data centers in 2025. However, the extraction and liquefaction process requires energy in itself, creating a vicious cycle: the energy for the data centers comes from gas, which in turn requires energy to be produced. The overall efficiency of the system is less than 50% under peak conditions, with a significant energy loss in the transformation cycle. This data indicates that the current model is not only dependent on gas, but also thermodynamically inefficient.
Market Expectations and Technical Reality
Market growth expectations in the digital sector are fueled by foreign investments and attraction policies, but are not aligned with the capacity of the energy system to support them. According to the Malaysian Ministry, the data center sector will contribute 7% to the GDP by 2035. However, the national energy plan foresees a 60% reduction in fossil fuel consumption by 2030. This discrepancy indicates a structural misalignment between economic goals and climate goals. Sources indicate that the government has initiated collaborations with ENERtec Asia and MIDA to develop solutions based on renewable energy and storage systems, but the projects are still in the design phase.
“Companies have been ‘tokenmaxxing’ for the last few months, encouraging employees to use GenAI as much as possible, without much regard to payoff.” — Gary Marcus, AI industry trend critic
Marcus’s quote highlights a widespread trend: the excessive use of technologies without a return on investment assessment. This phenomenon also affects infrastructure, where the demand for power is driven by a superficial optimization of development time, not by an efficiency analysis. The result is an increase in energy demand that does not correspond to an actual increase in value, but to a temporal compression of the production cycle. The system is in a state of over-engineering, where the infrastructure is built to meet a peak demand that is not sustainable.
The Cost of Compromise and the Prospects for Repositioning
The current energy compromise entails a growing infrastructure cost. Battery energy storage systems (BESS) are under development, but are not capable of supporting peak power demands exceeding 200 megawatts. The first solar community for data centers, launched by Time Energy, has a capacity of 5 megawatts, which is insufficient to cover the needs of a single unit. Delays in the implementation of renewable energy projects are due to bureaucratic delays and a lack of long-term funding. The country is in a phase of forced transition, where the efficiency of the system is reduced by a dependence on non-renewable sources.
The real trade-off is between immediate competitiveness and strategic sustainability. The cost is borne by the national energy system, which must absorb the burden of an unplanned expansion. The risk is not only environmental, but also economic: if the global market imposes stricter carbon footprint standards, Malaysian data centers could be excluded from international contracts. The systemic impact is measurable: gas consumption for data centers could increase by 45% by 2030, reducing availability for traditional industries.
Impact KPI: +45% gas consumption for data centers by 2030, compared to 2025
Monitor the ratio between data center growth and renewable energy capacity
If you are considering expanding the digital sector in Malaysia, the key figure to monitor is the ratio between new data centers under construction and installed renewable energy capacity. If this ratio exceeds 3:1, the system is in a critical misalignment phase.
Photo by Thufeil M on Unsplash
⎈ Content generated and validated autonomously by multi-agent AI architectures.
> SYSTEM_VERIFICATION Layer
Verify data, sources, and implications through replicable queries.