Morgan Stanley: Power Shortages are Becoming a Core Bottleneck for AI Infrastructure, Computing Power Expansion Enters the "Power Constraint Era"

BlockBeats News, June 15th - In its latest research, Morgan Stanley pointed out that the power shortage has escalated from a supporting issue to a core limiting factor for AI infrastructure development. The delivery time for power transformers has significantly increased from 12–16 weeks before the pandemic to 128–144 weeks. The backlog of US new energy grid connections has exceeded twice the existing installed capacity, while a shortage of 300,000 electricians and 43% of data centers being in high water-resource stress areas are jointly suppressing the expansion speed of computing power.

The expansion speed of the power system is far slower than the pace of data center construction, with the transmission network and key equipment supply chain having significantly longer cycles. The current average delivery time for power transformers has reached 128 weeks, with generator step-up transformers at around 144 weeks, compared to 12–16 weeks before the pandemic. This means that even if AI data centers have completed financing, site selection, and equipment procurement, they may still face delays in power connection and may not be able to start operations on time.

In the grid connection phase, the backlog of US new energy projects waiting to connect has exceeded twice the national installed power capacity, leading to a structural issue where “power generation completion does not equal available power.” Electricity must be connected to the grid before it can be converted into usable supply for data centers, shifting the site selection logic from “suitable for building data centers” to “areas where power can be quickly and stably accessed.”

Meanwhile, the boundary between AI infrastructure and the energy system's financing is becoming blurred, with some projects starting to adopt off-grid or semi-grid solutions, including direct power supply paths such as gas turbines, energy storage, and fuel cells. AI companies are also gradually shifting from relying on utility expansion to directly participating in power asset investments and securing power supply capacity, driving the capital market to integrate pricing for AI and energy assets.

In addition to power, labor and resources are also constraints. The US is expected to face a shortage of about 300,000 electricians over the next decade, with over 20% of the workforce being aged 55 and above; meanwhile, approximately 43% of data centers are located in high water-resource stress areas, where cooling water and alternative solutions are becoming significant limiting factors for new construction. Furthermore, several states have begun discussing or pushing for tighter restrictions and approvals for data center construction, further increasing project uncertainty.

Overall, power, grid connection, equipment, labor, water resources, and policy approvals are forming multiple overlapping constraints, potentially causing the expansion speed of computing power to lag behind demand growth. The report believes that this supply-demand mismatch will reinforce “computing power scarcity,” allowing participants with stable and deliverable computing power capabilities to have a stronger pricing power. The market is gradually shifting from “computing power expansion competition” to “competition for control of available computing power.”