AI capex surge strains US power grids and pushes utility valuations

The artificial intelligence boom is colliding with America's aging electrical grid, creating an infrastructure bottleneck that utilities are scrambling to address. US consumer power prices climbed last month by the most in three years, with electricity inflationThe rate at which prices rise across an economy. running 61 percent faster than headline CPI. The culprit is clear: hyperscale data centers powering large language models and AI training pipelines are consuming megawatts of electricity at unprecedented scale, straining transmission capacity and forcing utilities to race toward new generation. American Electric Power, one of the largest US utilities, announced a $2.6 billion share offering to fund infrastructure expansion and meet soaring AI-driven demand.

The power demand inflection is structural, not cyclical. AI model training requires continuous, high-availability power to GPUs and cooling systems. Microsoft reported that hackers had compromised Mistral AI software downloads, underscoring how critical secure, reliable compute access has become to the sector. Geothermal energy developer Fervo Energy raised $1.89 billion in an oversubscribed IPOInitial Public Offering - a company's first public sale of stock., betting that next-generation thermal baseload power can fill the gap left by intermittent renewables. Utilities across the country are facing similar pressures; PJM Interconnection, the biggest power grid in the US, may have become too large to manage, according to FERC Chair Laura Fineman, requiring urgent operational reform.

The equity implications are profound. Utilities and energy infrastructure plays are receiving capital allocation upgrades as investors recognize the structural nature of the demand. Fervo's IPOInitial Public Offering - a company's first public sale of stock. pricing above range signals appetite for clean baseload solutions. Conversely, marginal power consumers in non-AI sectors face margin pressure as power costs rise; this is particularly acute for aluminum and other energy-intensive manufacturers who compete globally on cost. Small businesses are also struggling with power price pass-through, adding to cost-of-doing-business headwinds. The geopolitical lens is important: US energy independence via sustained domestic generation attracts both traditional and ESG-focused capital, creating a rare alignment of economic and political interest.

The downside risk centers on deployment delays and regulatory friction. Building new nuclear, geothermal, or large-scale solar installations takes years due to permitting and environmental reviews. If demand outpaces supply for an extended period, power prices could remain elevated and threaten AI capex returns. Some utilities are experimenting with demand-response and dynamic pricing; if these tools prove insufficient, data center operators may face forced load-shedding or geographic relocation, undermining the concentration of AI infrastructure in existing hubs like Northern California and Texas.