MALAYSIA could be heading towards a potentially serious energy planning dilemma as it embarks on an aggressive expansion of electricity generation capacity, particularly gas-fired power plants, to meet anticipated demand from data centres (DCs).
The concern is whether the enormous amount of electricity projected for these facilities will eventually translate into actual consumption, especially in light of the US government’s tightening restrictions on advanced Nvidia chips reaching Chinese companies through third countries, including Malaysia.
If these restrictions cause some DC developers to delay, downsize or abandon their investments, Malaysia could find itself with more electricity generation capacity than it needs.
Could ordinary electricity consumers eventually be made to pay for power plants built to serve DC investments that never materialise?
The scale of Malaysia’s planned generation expansion is considerable.
Earlier this year, the government disclosed plans for up to 9,600MW of new generation capacity between 2028 and 2031, driven partly by growing electricity demand from DCs.
Under the Energy Commission’s (EC) NewGen25 programme, awards announced in February involved approximately 6,269MW of generation capacity.
However, most of this comprised extensions to existing gas-fired plants, totalling around 4,719MW. The remaining 1,550MW involved additional capacity, including a new 1,400MW combined-cycle gas turbine plant in Paka, Terengganu.
Meanwhile, the subsequent NewGen26 tender seeks further gas-fired generation capacity, with the EC identifying requirements of roughly 8,000MW between 2029 and 2031.
Some of this capacity is intended to replace ageing power stations. Nevertheless, they illustrate the substantial commitments being contemplated to ensure Malaysia has sufficient electricity to meet future requirements.
A major driver of this expansion is the government’s expectation that DCs will account for an increasingly large share of electricity consumption.
Official projections indicate that DCs could consume 73,274GWh of electricity annually by 2035, compared with 10,544GWh in 2026. Their share of Peninsular Malaysia’s electricity consumption is projected to rise from about 7% in 2026 to a staggering 31% by 2035.
Moody’s Ratings estimates that Peninsular Malaysia could require an additional 14GW to 17GW of generation capacity by 2035, involving investments of approximately RM80bil to RM95bil by Tenaga Nasional Bhd
and independent power producers.
But what happens if these demand projections prove overly optimistic?
The question has become particularly relevant following Washington’s efforts to restrict Chinese companies from accessing advanced Nvidia computing capacity through overseas facilities.
Malaysia has emerged as an attractive location for AI-focused DC operators, including companies seeking to provide computing services to customers outside the country.
Some of these investments may be vulnerable if US export controls increasingly restrict not only the physical movement of advanced chips but also the ability of Chinese entities to rent computing capacity housed in Malaysia.
Of course, not every DC investment is dependent on Chinese customers. Established hyperscalers such as Google, Microsoft and Amazon also cater to regional cloud computing and enterprise demand.
Nevertheless, restrictions on advanced AI computing could undermine the commercial viability of certain projects, particularly those involving graphics processing unit (GPU) rental businesses.
The danger is that DC developers may postpone expansion plans, reduce their computing capacity or cancel subsequent phases of development. Yet, the electricity infrastructure being planned to serve these facilities could already be under construction or committed under long-term contracts.
To be fair, Malaysia needs new power plants regardless of the DC boom.
A significant amount of existing coal and gas-fired generation capacity is scheduled for retirement towards the end of this decade.
New gas-fired plants are, therefore, necessary to replace ageing facilities, maintain system reliability and support the country’s energy transition.
But this makes it all the more important to distinguish between generation capacity needed to replace retiring plants and additional capacity justified by anticipated DC demand.
The latter presents a greater risk if the expected demand fails to materialise.
Under Malaysia’s electricity market arrangements, independent power producers generally operate under long-term power purchase agreements that provide capacity payments for making their plants available, even when they are not generating electricity.
These payments help power producers recover their fixed investment and financing costs.
Consequently, if Malaysia commissions too much generation capacity, reducing electricity production would not necessarily eliminate the financial obligations associated with those plants.
Fuel costs may fall when plants are dispatched less frequently, but fixed capacity payments could remain payable for many years.
Ultimately, these costs form part of the electricity system’s overall expenditure, which must be recovered through tariffs, government support or other contractual arrangements.
Under the Incentive-Based Regulation framework, the EC determines the regulated costs that can be recovered through electricity tariffs.
If substantial fixed generation costs have to be spread over fewer units of electricity sold than originally projected, the average cost of supplying electricity could increase.
This creates the possibility of higher tariffs for households and existing businesses, unless the government or DC developers absorb the additional costs.
This is particularly pertinent given the government’s repeated assurances that ordinary Malaysians will not subsidise the infrastructure required by DC operators.
But those assurances need to be tested against a scenario in which the anticipated demand never materialises.
Who ultimately bears the cost of a gas-fired power plant built partly in anticipation of DC demand if the intended customers subsequently scale back their investments?
The EC should reassess its generation development plans, incorporating scenarios in which DC electricity demand falls substantially below current projections. It should also examine whether some new gas-fired capacity can be deferred until there is greater certainty over actual electricity consumption.
Equally important, DC developers seeking large electricity allocations should be required to provide meaningful financial commitments, potentially through minimum payment obligations or other contractual safeguards.
A DC developer may have the commercial flexibility to delay or abandon a project when market conditions deteriorate. But Malaysia’s electricity system could remain saddled with long-term financial obligations arising from power plants built to serve that project.
The government must ensure that electricity infrastructure development remains aligned with realistic, regularly updated demand projections.
Malaysia should certainly build the electricity capacity it needs for future economic growth.
What it must avoid is committing itself to decades of costly excess capacity based on DC investments that may never fully materialise.
