Hyperscaler datacenter demand has pushed US electricity growth forecasts sharply higher through 2030. The grid was not built for this load. The companies below are building the generation, transmission, on-site power, cooling, and utility exposure attached to AI infrastructure.
Generation at the apex. Utilities at the foundation. Public companies mapped by their role in supplying datacenter power.
Merchant and contracted generators with exposure to firm power demand from hyperscalers, grid operators, and large industrial loads. This layer includes power producers whose capacity, interconnection rights, and operating fleets are becoming central to datacenter site selection.
Firm generation is part of the datacenter power question, but full reactor and fuel-cycle coverage belongs in Civil Nuclear. This cross-reference layer surfaces representative nuclear names tied to long-duration supply, fuel availability, and new reactor development.
Transmission contractors, electrical equipment providers, and grid engineering names that sit between generation and load. Datacenter growth depends on substations, switchgear, transformers, transmission corridors, and utility-scale electrical work that can absorb new capacity.
Gas turbines, backup power, engine platforms, thermal management, and cooling systems that keep high-density datacenters operating when grid interconnection is constrained. This layer captures the equipment side of site-level resilience.
Regulated utilities serving regions where datacenter load, industrial electrification, and transmission planning are changing capital plans. These names carry public service obligations as well as exposure to larger load forecasts.
Microsoft, Amazon, Google, and Meta have committed to large AI infrastructure programs through 2027. PJM capacity prices rose sharply in the 2025 auction, and Constellation announced a Three Mile Island restart agreement with Microsoft. The theme tracks the public companies attached to that power constraint.
Constituent lists are reviewed quarterly against revenue mix, segment disclosures, grid exposure, regulatory filings, and stated capex commitments. Quotes are sourced from licensed market data; the Veridion Score is computed from six published factors. Inclusion is not a recommendation.