Grid stagnation: Why copper is now a national security risk for AI

The surge in AI-driven investment has exposed a critical bottleneck: electricity grids that cannot expand fast enough are creating strategic and economic vulnerabilities, making grid stagnation a national security concern.

Hyperscale data-center campuses can be built in under two years, while necessary transmission upgrades—transformers, substations and long-distance lines—often require five to ten years and extensive permitting. Morgan Stanley analysts warn that this timing mismatch could cause localized power shortages in 2027–2028 and delay many projects currently stuck in interconnection queues.

Research from the Harvard Kennedy School’s Belfer Center and Lawrence Berkeley National Laboratory shows data-center electricity demand rising from roughly 4.4% of U.S. consumption in 2023 to an estimated 6.7–12% by 2028, squeezing regional grids and forcing utilities to reassess cost allocation. Several utilities have already flagged large new load requests that exceed existing transmission capacity.

Beyond economics, there is a geopolitical dimension: countries with centralized energy planning can overbuild capacity for strategic sectors, potentially capturing AI leadership. A grid pushed near its limits is also more vulnerable to outages and targeted cyber operations, linking energy resilience to digital and defense capabilities.

Analysts expect a mix of policy and market responses—new rate structures to shift upgrade costs to large loads, investments in storage and demand-response, and accelerated permitting for transmission projects. Yet most agree that software and flexibility tools can only buy time; expanding physical copper and transmission infrastructure remains essential if the U.S. is to sustain both commercial AI development and broader grid reliability. Investors should watch utility capex plans, regulatory decisions on cost allocation, and interconnection queue backlogs as signals of where risk and opportunity will concentrate.