The artificial intelligence revolution in Asia is colliding with the physical limits of the real world. A new analysis by Morgan Stanley’s ASEAN/India energy analyst Mayank Maheshwari, published in Nikkei Asia, warns that the region’s unprecedented compute boom is running into a massive “power wall.” The report projects that powering this AI expansion will require a $5 trillion energy investment supercycle across Asia by the end of the decade, fundamentally reshaping the region’s infrastructure priorities.
The Numbers Behind the Power Wall
The scale of the impending energy demand is difficult to overstate. Morgan Stanley estimates that Asia-Pacific data center power consumption will exceed 100 gigawatts (GW) by 2030, growing at a compound annual rate of 23%, roughly on par with the United States. Data center electricity use will quadruple from 193 terawatt-hours (TWh) in 2023 to 832 TWh by 2030, representing approximately 3% of total regional power demand and 15% of all new demand creation.
China is explicitly named as a central driver of this demand, alongside Japan, India, Malaysia, Singapore, and beyond. The country’s aggressive push to build a nationwide AI infrastructure, exemplified by RedNote’s 600 MW facility in Ulanqab and DeepSeek’s planned 1 GW data center in Inner Mongolia, places it at the epicenter of the impending power crunch. The hardware itself is exacerbating the problem: next-generation AI GPU racks draw 40–200 kW each, compared to just 5–15 kW for traditional enterprise racks, with next-generation platforms heading toward hundreds of kW per rack.
Energy Security Becomes a Strategic Priority
This surging demand arrives at a precarious moment for Asian energy security. The region consumes as much energy as the rest of the world combined but produces only about one-third of its needs domestically. The IEA notes that Asia’s energy consumption has risen 50% over the past decade, while investment in traditional energy supply chains has fallen to its lowest levels, now below data center investment. This imbalance becomes harder to sustain as compute demand rises.
Morgan Stanley expects Asia’s average annual energy investment to nearly double by 2030, spanning coal, natural gas, renewables, energy storage batteries, and power grids. Annual incremental energy storage system (ESS) deployment linked to data centers could reach 321 gigawatt-hours (GWh) by 2030, implying a doubling of current ESS demand. Southeast Asia alone is forecast to see battery energy storage deployments increase from approximately 2 GWh per year today to nearly 16 GWh per year by 2032.
The Defining Constraint of the AI Era
For China, the implications are profound. The country’s ambitious AI goals are now inextricably linked to its ability to rapidly scale and modernize its power grid. The massive investments required will strain resources and force difficult decisions regarding energy mix and environmental commitments. The “power wall” is no longer a theoretical future constraint; it is the immediate, defining challenge of the AI era. As the AI arms race continues, the victor may not be the nation with the most advanced algorithms or the fastest chips, but the one that can most reliably and sustainably power them.
Maheshwari’s analysis also highlights a structural advantage that China holds over many of its regional competitors. Unlike Southeast Asian nations that must build power infrastructure largely from scratch, China already possesses the world’s largest installed renewable energy capacity and a grid modernization program of extraordinary scale. The 117 GW of new renewable capacity added in just the first half of 2026, detailed in today’s separate story on China’s energy milestone, is a direct contributor to the country’s ability to power its AI ambitions.
The challenge is not whether China can generate enough electricity but whether it can deliver that power reliably to the specific locations where hyperscale data centers are being built. Grid connectivity, transmission infrastructure, and local substation capacity are the near-term bottlenecks, and they are already shaping where companies like DeepSeek and RedNote choose to locate their facilities.
The $5 trillion energy supercycle projected by Morgan Stanley is not a distant forecast, it is already underway. The investment decisions being made today by data center developers, power utilities, and grid operators across Asia will determine the physical infrastructure of the AI era for decades to come. For China, which is simultaneously the world’s largest AI investor, the world’s largest renewable energy deployer, and the world’s largest electricity consumer, the stakes of getting this transition right are higher than for any other nation. The power wall is real, but so is China’s capacity to build through it.