The numbers come first. Data center power demand across Asia-Pacific is projected to rise 165% between 2023 and 2030, according to a white paper co-authored by Marex and Oxford's Smith School. Yet in 2024 the region delivered only about 38% of its announced data center capacity — one of the widest plan-to-delivery gaps of any market globally.
The arithmetic behind that gap is straightforward. The International Energy Agency's Southeast Asia Outlook puts grid and storage investment in 2025 at $13 billion. The IEA says $50 billion is needed annually through 2050. That is a $37 billion-per-year shortfall, and it is not closing on its own.
Japan bets trillions; the grid hasn't caught up
Japan recently announced a 370 trillion yen ($2.3 trillion) budget, with more than a quarter earmarked for artificial intelligence and chips over the next 15 years. The commitment is real. The transmission infrastructure to support it is not. Renewables, as Guy Wolf — global head of market analytics at Marex — notes in Fortune, are typically built far from demand centers and generate power intermittently. Without new transmission and storage, server racks struggle to operate at full capacity.
Malaysia and India illustrate the problem in concrete terms. Johor has banned construction of Tier 1 and Tier 2 data center facilities citing strain on local water infrastructure. India's ambition to double projected capacity by the end of its next financial year faces severe grid delivery lags. Much of the region's headline megawatt figures, Wolf writes, are 'bragawatts' — announcements that look impressive on paper but are far slower to turn into real energy.
Markets are already pricing the fantasy
Commodity markets are not waiting for reality to catch up. Copper prices have stayed elevated on assumptions of surging data center construction demand. Transformer costs are running at two to three times pre-2020 levels as developers lock in scarce equipment. If interconnection queues stretch the way they have in the U.S. and Europe, the mismatch between announced and delivered capacity could produce the kind of boom-bust cycle that metals markets saw in the last decade.
The U.S. offers a cautionary parallel. In the first three months of this year, 75 data center projects worth a combined $130 billion were blocked or delayed by local opposition, matching the total number blocked in all of 2025. Up to half of all planned U.S. projects may not come online this year. Asia is replicating that dynamic with the added weight of state-controlled energy infrastructure.
The structural fix: open the markets
Wolf's diagnosis points directly at governance. Most Asian electricity systems still rely on vertically integrated, state-owned utilities acting as single buyers, with retail tariffs set administratively and limited third-party trading. Investors in Europe and the U.S. take forward-contract trading layers for granted; investors in Asia do not get the pricing signals they need to commit capital.
Singapore, Malaysia, and South Korea are beginning to require battery-storage plans and grid-impact assessments from data center developers — a step toward accountability, if not yet toward price competition.
Meanwhile, the U.S. has $4 trillion in additional data center construction planned through 2028. Every quarter that Asian operators wait, Wolf argues, is another quarter of compute, talent, and capital that could be deployed elsewhere.
CEO Times take: The lesson here is one free-market readers will recognize instantly. State monopolies in electricity do not just distort prices — they destroy the forward-pricing signals that private capital needs to invest with confidence. Asia's AI ambitions are not constrained by engineering; they are constrained by bureaucracy. A more liberal approach to distributed energy generation and wholesale electricity trading would attract investment and reduce dependence on imported oil and gas. The region's governments have the masterplans. What they lack is the institutional willingness to let markets do the work.
