Our Take
AI is forcing a capital reallocation in power infrastructure, but this is financial consolidation, not technical innovation—and it signals investors expect peak demand growth faster than grid expansion.
Why it matters
Data center operators need guaranteed power supply at scale. Utilities and energy companies are racing to acquire generation and transmission assets before capacity constraints push electricity costs prohibitively high. This M&A wave will reshape which companies own AI infrastructure.
Do this week
Infrastructure leads: audit your power supply contracts for expiry dates and capacity caps before 2025 budget cycles reset, so you can secure long-term allocations before prices spike.
Record consolidation in US power markets
Merger and acquisition activity in the US power sector reached $200 billion (per Financial Times reporting), the highest on record. The surge is driven primarily by artificial intelligence operators and data center developers seeking reliable, dedicated power supplies for training and inference workloads.
Utilities and energy producers are acquiring generation assets, transmission infrastructure, and grid operators to secure supply chains. Large tech companies and AI service providers are either buying power assets directly or signing long-term purchase agreements with consolidating utilities.
AI electricity demand is real and immediate
Data center operators have moved past theoretical capacity concerns. Demand for GPU clusters, model training, and inference serving now creates genuine electricity shortages in key regions (Northern California, Texas, the Mid-Atlantic). Utilities cannot build new generation or transmission fast enough to meet the timeline AI companies need.
M&A is the speed lever. Acquiring existing plants, distribution networks, and grid access is faster than permitting and constructing new infrastructure. Consolidation also allows utilities to integrate AI-driven load forecasting and demand-response systems that individual operators cannot afford alone.
This is not speculation about the future of AI. It is capital flowing today to lock down power supply over the next 3-5 years.
Secure power contracts now or pay premium rates later
If your organization operates data centers or plans to deploy large model inference at scale, power is no longer an afterthought. Utilities are consolidating and will have leverage to negotiate rates upward as AI demand competes with residential and industrial users.
Longer-term purchase agreements (5-10 year contracts with fixed or capped rates) are becoming the binding constraint on where AI companies can build. Spot market electricity will grow more volatile and expensive. Companies that lock contracts early will have a structural cost advantage over those betting on grid expansion or price declines.
If you are evaluating new data center locations, power cost and availability should move to the top of the due-diligence list, ahead of real estate, staffing, and network latency.