Jakarta. The rapid expansion of artificial intelligence infrastructure is putting new pressure on Southeast Asia’s electricity systems, with data centers requiring substantial power and potentially straining grids if energy infrastructure fails to keep pace, the World Bank said.

The International Energy Agency (IEA) forecasts global electricity demand from data centers will double by 2030 and triple by 2035, reaching as much as 4% of total electricity production, the World Bank said.

The power requirements of AI data centers are particularly significant. A 500-megawatt AI data center, an industry benchmark for AI training, would consume nearly 15% of the average baseload power available in emerging market and developing economies in East Asia and Pacific, nearly four times the average share in advanced economies.

The IEA estimates that AI-related data centers increased their electricity demand by 17% in 2025, according to the World Bank. High baseload consumption can strain electricity grids and increase the risk of grid failures.

Energy is also a major operating cost for data centers. While power accounts for a relatively small share of their initial capital expenditure, energy represents about 65% of annualized operating expenditures, the World Bank said.

Electricity costs vary significantly across Southeast Asia, which could influence the attractiveness of different markets as AI-related infrastructure expands. Data from SEASIA Stats puts Malaysia’s electricity price at around $0.063 per kilowatt-hour, compared with $0.096/kWh in Indonesia and $0.093/kWh in Vietnam. China’s national average was around $0.078–$0.079/kWh as of mid-2026, according to GlobalPetrolPrices.com.

Prices are considerably higher in some other regional markets, at around $0.246/kWh in the Philippines, $0.178/kWh in Cambodia and $0.151/kWh in Thailand. In the Philippines, reliance on imported fuel, geographic fragmentation and infrastructure challenges contribute to higher power costs, while Cambodia has expanded its electricity network substantially but continues to have relatively high prices compared with several neighboring countries.

But rapid expansion is already creating local concerns. Johor, Malaysia, the fastest-growing data center hub in Southeast Asia, has faced community backlash over concerns that the expansion could worsen pollution and water scarcity, according to the report. The World Bank said consultation, environmental impact assessments and integrated land- and water-use planning could help address these risks.

The report also highlighted how data center operators are securing reliable power as electricity demand grows. In Vietnam, large hyperscalers are using power purchase agreements to secure electricity supplies, while Malaysia has introduced corporate renewable energy schemes to provide companies with access to renewable power.

The World Bank said scaling up electricity generation will be increasingly important as AI infrastructure expands. Renewable sources such as solar and wind are becoming more cost-effective options for adding generation capacity while producing fewer negative externalities than other energy sources.

The need for additional power is also linked to the region's vulnerability to imported fossil fuels. Recent disruptions to energy flows through the Strait of Hormuz have highlighted the risks, including shortages of key goods, slower growth, higher inflation and currency volatility.

The World Bank said countries will need to make significant investments in new generation capacity to reduce those vulnerabilities over the longer term.

“Scaling up energy generation capacity to meet the needs of growing economies and AI use can also limit the risks of depending on imported fossil fuels,” the World Bank said.

Add as a preferred source on Google +