Artificial intelligence is driving one of the fastest increases in electricity demand in decades. At the same time, rising temperatures are pushing peak power usage higher across major markets. The combination is beginning to stress power grids that were built for stable demand patterns.
The Steady Surge in Power Demand
Electricity consumption from data centers has accelerated sharply in recent years. According to the International Energy Agency, global data center electricity use reached roughly 415 terawatt-hours in 2024, accounting for about 1.5% of total global demand. By 2030, consumption is projected to approach 945 terawatt-hours.
IEA (2025), Share of electricity consumption by data centre and equipment type, 2024, IEA, Paris
AI workloads are a major driver. Training and inference for large models require significantly more power than traditional computing, and AI-optimized servers operate at higher utilization levels. As a result, data center electricity demand is expected to grow several times faster than overall electricity demand through the end of the decade.
This shift is already visible in utility forecasts, particularly in markets where data center development has accelerated.
Geographic Concentration
Data center demand is highly concentrated geographically. In the United States, major hubs include Northern Virginia, Texas, Arizona, and parts of the Midwest. These regions offer access to fiber infrastructure, land availability, and favorable tax regimes, but they also create localized demand spikes.
Large AI campuses can require hundreds of megawatts of capacity, comparable to heavy industrial facilities. When multiple projects cluster in the same region, utilities must expand generation, transmission, and substations simultaneously.
This creates long lead times. Transmission infrastructure often takes five to ten years to develop, while data center construction timelines are typically much shorter.
Supply Constraints
Power supply expansion is struggling to keep pace. Utilities are increasing capital expenditure plans, but permitting, construction timelines, and interconnection queues are slowing deployment.
The supply mix is also shifting. Renewables continue to grow, but intermittent generation creates additional balancing challenges. Natural gas remains a near-term solution in many regions, while nuclear and energy storage are being reconsidered for long-term reliability.
In parallel, large technology companies are signing long-term power purchase agreements and, in some cases, exploring dedicated generation or microgrid solutions to secure capacity.
AI Meets Summer Demand
The timing of rising data center demand coincides with increasing summer temperatures. Air conditioning already drives peak electricity demand in many regions. As heat waves become more frequent and begin earlier in the year, seasonal peaks are rising.
Data centers add a constant baseline load to these peak periods. Unlike residential demand, which fluctuates throughout the day, AI infrastructure operates continuously. The overlap between sustained data center demand and seasonal cooling demand increases the risk of grid stress during extreme weather events.
This dynamic is already emerging in regions such as Texas and the U.S. Southwest, where both data center development and temperature volatility are high.
Investment Implications
Rising power demand from AI is creating a new infrastructure cycle. Utilities are expanding generation and transmission capacity, while private capital is moving into power infrastructure, grid modernization, and energy storage.
Data center developers are also beginning to prioritize locations with available grid capacity, which may shift geographic development patterns. In some cases, access to power is becoming a primary constraint for new projects.
The AI buildout is increasingly tied to physical infrastructure. As electricity demand rises and grid expansion lags, power availability is emerging as one of the defining constraints of the AI cycle.
