18 SEP 2026 — Data centres took four per cent of Peninsular Malaysia's electricity last year and six per cent in the first half of this one. Moody's Ratings expects thirty-one per cent by 2035.
Meeting that means adding 14 to 17 gigawatts of generation capacity over the decade, at a cost the agency puts between RM80 billion and RM95 billion.
What the forecast actually says
The headline figure is a share, not a total, and the share moves because the denominator is also growing. Moody's expects electricity demand in Peninsular Malaysia to rise at a compound annual rate of about 5.8 per cent over the decade, against 2.5 per cent over the ten years to 2025.
Demand roughly doubles its growth rate, and data centres take a rising share of a rising total. That is a larger claim than data centres becoming a big customer.
The agency credits three things for it: competitively priced and available power, supportive government initiatives including the separate data-centre tariffs introduced in 2025, and spillover demand from Singapore, which has had its own constraints on where this industry can physically go.
The half that is retiring
The build-out is only half the story. Around 13 gigawatts of existing generation capacity expires between 2026 and 2035: roughly 7 gigawatts of coal and 6 of gas.
Set that against the 14 to 17 gigawatts being added and the arithmetic changes shape. Much of the new capacity is replacement rather than growth, and the net addition available to serve new load is smaller than the gross figure suggests.
This is a recurring problem in regional capacity announcements: a new gigawatt gets a press release, an expiring one does not.
Declared demand is not load
The most useful figure in the report is the smallest. Actual load drawn by data centres reached 1.26 gigawatts in June 2026, and Moody's notes this sits below the aggregate demand those data centres have declared.
That gap is the planning problem. Declared demand is the capacity an operator reserves; load is the electricity it actually draws. Utilities build against the first and are paid for the second, and the difference lands on their balance sheet.
We have watched the same gap open elsewhere. Texas froze a data-centre interconnection queue that had reached 474 gigawatts of requests, a figure nobody believed represented real projects. Malaysia's version is smaller and better disclosed, but it is the same measurement problem.
Where the risk lands
Moody's is a ratings agency, so its conclusions are about credit rather than kilowatts. Delays in commissioning power infrastructure could compress reserve margins and weigh on supply reliability; cost overruns or delays to generation projects could pressure the credit quality of Tenaga Nasional and the independent power producers.
Capital spending is put at RM4 billion to RM5 billion a year over the next five years. Sarawak Energy's export-related earnings are treated more cautiously, since the timing depends on commercial arrangements and cross-border transmission that does not exist yet.
None of this says the build-out is unwise. It says utilities carry the timing risk, and the customers driving demand do not.
What to watch
The gap between declared and actual load is the first thing to watch. If the 1.26 gigawatt figure converges on declared demand over the next two years, the forecast is conservative. If it does not, Malaysia will have procured capacity against reservations that never became load.
Whether the 13 gigawatts expiring is replaced on schedule matters more to reserve margins than any individual hyperscaler announcement, and it is the part least likely to be announced at a ribbon-cutting.
Johor has already had local disputes about what these sites cost the neighbourhood, and across the strait Batam is building against an island grid. The spillover Moody's credits for Malaysian demand is a planning decision made in another country, and it does not appear in any Malaysian forecast as a variable that could reverse.
