What the demand surface reveals about coal-gas competition

Stockpile-driven softness in 2024, peak-day strength in 2025, and quiet acceleration in 2026

Disentangling structural from seasonal factors in power generation is both harder and more important than ever. In the first post of this series, I showed why traditional metrics don’t effectively capture these changing fundamentals, and how new methods reveal two key changes in coal-gas competition:

• Gas-fired generation is becoming less elastic at lower levels of fossil generation.

• Coal-gas competition shifted sharply between 2023 and 2025: gas ceded market share in 2024, then rebounded in 2025, but that rebound concentrated on higher-demand days.

This week, I’ll unpack the drivers of these trends, extend the analysis through early 2026, and pull out implications for 2027 and beyond.

Why is gas price elasticity limited on low-demand days?

Examining how gas prices and fossil generation jointly and non-linearly affect gas-fired generation shows limited elasticity on low-demand days, as indicated by little color change from left to right at the bottom of Figure 1. I estimate ~140 GWh (~1.1 Bcfed) of elasticity per $1/MMBtu move in gas prices on these low-demand days, but ~290 GWh (~2.2 Bcfed) on a moderate-demand day.

Figure 1 | Gas-fired generation surfaces

To evaluate whether this reduced elasticity on low-demand days is a robust result, we need to examine both fleet- and unit-level operational patterns. Hourly generation data shows how rising renewable generation has dramatically changed fossil load shapes, albeit in different ways across low-, medium-, and high-demand days.

Figure 2 | Fleet load shapes by fossil generation quartile