A peer-reviewed modelling framework published in Progress in Energy finds that the composition of future electricity demand on the Texas grid — not just its magnitude — determines what generation, storage and transmission infrastructure ERCOT will need through 2050. Drew A. Kassel and colleagues report that electric vehicles drive the sharpest increases in peak demand, while data centres and other large industrial loads dominate growth in total consumption.

The study, published as Kassel et al. 2026, Prog. Energy 8 035001, develops a method to assess the energy impacts of meeting ERCOT's uncertain future electricity demand. Texas offers a distinctive test case: ERCOT manages most of the state's electricity system on a grid that operates largely independently from the rest of the United States, and demand patterns there are shifting rapidly under population growth, electric vehicle adoption, AI data centre expansion and industrial electrification.

The researchers modelled five demand drivers separately through 2050: population growth, electrified heating, electric vehicles, large industrial loads such as data centres, and electrification of oil and gas operations. By isolating each source, the framework quantifies how differently shaped load growth stresses the grid in different ways.

Three results stand out. First, electric vehicles have the greatest impact on peak demand, because charging tends to cluster at similar times of day and produces sharp, synchronised ramps. Second, large industrial loads — data centres, cryptocurrency mining, hydrogen production and manufacturing facilities — exert the greatest impact on overall electricity consumption, since these facilities often run continuously around the clock. Third, electrified heating produces a smaller net effect than either: heat pumps raise winter electricity demand but simultaneously reduce summer demand through more efficient cooling, partially offsetting themselves across the seasons.

Different demand, different investments

The practical consequence for planners and investors is that each demand category calls for a distinct procurement strategy. Battery storage is particularly effective at managing the sharp peaks associated with EV charging, where the problem is temporal concentration rather than raw energy volume. Growth from data centres and other continuously operating industrial loads is most economically met through additional natural gas, wind and solar generation, which adds energy rather than shifting it in time.

The study also quantifies how uncertainty compounds these decisions. Depending on how future demand evolves, grid expansion requirements by 2050 could differ dramatically — meaning that portfolio decisions made now, under one demand assumption, could be significantly oversized or undersized if a different mix of demand growth materialises.

Why composition matters as much as size

The central contribution for R&D and infrastructure planning is methodological: the source of future electricity demand can matter as much as its size, because different demand drivers require different investments in generation, storage and network infrastructure. A gigawatt of EV-driven peak growth and a gigawatt of data-centre-driven energy growth are not equivalent planning problems, and treating aggregate demand forecasts as a single number obscures which assets the grid actually needs.

The framework is positioned as a decision-support tool for planners, investors and policymakers evaluating future electricity systems, and it is transferable in principle to other grids facing similarly uncertain electrification pathways.

Readers should note the usual caveats that apply to scenario work of this kind: the results are model outputs conditional on assumptions about adoption rates, behaviour and technology costs, not measured outcomes, and the underlying uncertainties are large by construction. The value of the framework lies in making those uncertainties explicit and attributable to specific demand drivers rather than in predicting a single trajectory.

The authors suggest the next step for the research community is applying the framework to other systems and refining how correlated, uncertain demand drivers interact — work that would tighten the link between demand-side forecasting and capital allocation on the supply side.