Planned US Data Center Campuses Highlight City-Scale Power Requirements
DataNews Editorial Team
Visual Capitalist has compared the annual energy equivalents of three planned or in-development US data center campuses with electricity consumption in major US cities. Meta’s planned Hyperion campus in Richland Parish, Louisiana, is the largest in the comparison, at 43.8 TWh annually at full build-out.
- Meta’s planned Hyperion campus is assigned a 43.8 TWh annual energy equivalent at full implementation.
- AWS Susquehanna Data Campus is listed at 16.82 TWh annually, with full volume indicated by 2032.
- The Vantage / Stargate Port Washington campus is associated with 11.39 TWh of planned utility-forecast demand annually through 2030.
- The figures are continuous-operation energy equivalents at future stated power levels, not forecasts of actual consumption.
- The projects’ scale could increase the importance of utility capacity planning and grid interconnection delivery.
Visual Capitalist has published a comparison of estimated annual electricity equivalents for three future US data center developments: Meta Hyperion in Richland Parish, Louisiana; AWS Susquehanna Data Campus; and the Vantage / Stargate campus in Port Washington, Wisconsin.
The largest figure in the comparison is Meta’s planned Hyperion campus, with an annual energy equivalent of 43.8 TWh at full implementation. The source compares this with 5.6 TWh of electricity consumption for the New Orleans service territory in 2025, making Hyperion’s calculated equivalent nearly eight times larger.
AWS Susquehanna Data Campus is assigned an annual energy equivalent of 16.82 TWh, with full volume indicated by 2032. For the Vantage / Stargate campus in Port Washington, the source cites planned utility-forecast demand of 11.39 TWh annually through 2030.
These figures should not be read as forecasts of actual metered electricity consumption. They are annualized energy equivalents calculated on the assumption of continuous operation at stated future power levels. They also rest on different timeframes and methodological bases: Hyperion is measured at full implementation, AWS at full volume by 2032, and Port Washington against a utility demand forecast through 2030.
Even with those qualifications, the comparison indicates the scale of power-system exposure associated with large AI-oriented campuses. If the projects are delivered at the stated levels, developers and operators could face increasing dependence on utilities’ ability to reserve, connect and deliver very large volumes of power ahead of demand. Grid interconnection availability and delivery schedules could therefore become constraints alongside site availability and construction execution.
The comparison illustrates how individual AI-oriented data center campuses could become grid loads comparable with large urban electricity demand. If these projects reach their stated future operating levels, utility capacity planning, grid infrastructure and interconnection timing could become as consequential to project delivery as site control and construction readiness.
- — / September 24, 2026 · EnglishSource →
This article was prepared by the DataNews editorial team based on the sources listed above.