AI Infrastructure Capex Race Drives Power and Nuclear Demand

The AI infrastructure buildout is creating a power supply crunch that threatens to outpace utility and generation capacity. US electricity prices surged 61% faster than headline inflationThe rate at which prices rise across an economy. in April, driven by booming demand from AI datacenters competing for limited power supplies. American Electric Power, one of the largest US utilities, announced a 2.6 billion dollar share sale to raise capital for AI-driven demand, signaling that the current grid infrastructure is inadequate for the coming surge.

Nuclear and renewable energy providers are stepping up. BWX Technologies, a nuclear reactor manufacturer, is being pushed by activist investor Ananym Capital to tap its commercial nuclear capabilities to serve the datacenters that cannot be serviced by traditional grids. Venture Global, a liquefied natural gas exporter, announced new supply deals and expansion plans for Louisiana export projects, reflecting corporate and global demand for stable energy to power AI infrastructure. Fervo Energy, a geothermal developer, raised 1.89 billion dollars in an IPOInitial Public Offering - a company's first public sale of stock. priced above range, signaling strong investor appetite for alternative baseload power sources capable of supporting continuous datacenters.

The grid itself is showing strain. PJM Interconnection, the largest power grid in the US, may be too large to adequately function and needs urgent reform, according to Federal Energy Regulatory Commission Chair Laura Polley. China's manufacturing heartland is experiencing power supply stress as economic activity heaps pressure on coal and oil supplies constrained by the Iran war. The convergence of AI capex and geopolitical energy disruption is creating a perfect storm: power prices are rising, grid reliability is questioned, and utilities must make massive investments to avoid brownouts.

Investor enthusiasm for clean energy and nuclear is spiking, but sceptics note that the timeline to build new capacity (nuclear plants take 10-15 years, offshore wind requires permitting) means that power prices will likely remain elevated for years. Additionally, if AI chip demand slows or capex budgets are revised lower, the urgency could evaporate. For now, the narrative is clear: AI infrastructure is not unlimited; it is constrained by power, and that constraint is driving a structural bull case for energy and utilities.