What’s moving the
markets now.

ERCOT's Monday peak came in at 56,966 MW, 0.6% over forecast, with Tuesday called at 59,500 MW and Wednesday at 62,000 MW while wind stays light until Thursday evening. That is a firmer shoulder-season setup than last week's cooler read, when weekly average load was estimated at 49.55 GW, down from 62.48 GW the week ending Oct 16. If you are short the midweek block, the swing between those two weeks is the exposure to watch, and thin wind is what turns a mild load bump into a price event.

The longer view is just as firm. PJM hit 162.6 GW on July 2, ERCOT hit 91.1 GW on July 22 and SPP hit 58.0 GW on July 27, and those records are what built the scarcity premiums now sitting in forward curves. PJM's study of SoftBank's 10 GW Ohio complex covers 9.6 GW of new load and 10.2 GW of new gas generation phased through 2032, with cost allocation still open at PUCO and FERC. For a retail provider, that means the question is less whether load grows and more who pays for it, and when that cost reaches the retail price.

November futures slid roughly 3.6% to about $3.005 after failing to hold $3.00, even with LNG feedgas at 21.2 Bcf/d. Lower-48 output is running more than 6% above last year and warm weather through about Oct. 26 is capping heating demand. Working gas sits at 3,567 Bcf after a 67 Bcf injection, 45 Bcf above the five-year average, though still 143 Bcf below last year. EIA's October STEO sees end-October stocks near 3,850 Bcf, about 2% above average. For a marketer buying ahead of winter, that is a cushion that supports lower prompt prices, but it is not a floor.

The watch item is what sits underneath. South Central stocks still run under their average, and LNG demand is the one piece of the balance that keeps pulling. FERC's 4-1 vote to ease permitting for gas pipelines and LNG plants, including a three-tier blanket authorization with 60 or 90 day notice for some LNG work, points to more export pull over time. A warm first week of November keeps the pressure on price, while a cold turn in the models can erase that cushion quickly. Hedge the shoulder, but keep winter optionality.

Google is paying for a 100 MW virtual power plant in PJM, with Voltus running the three-year program and carrying Google's capacity obligation. In Georgia, Central Georgia EMC plans 50 MW of home batteries with Base Power, charging when wholesale prices are low and discharging into peaks starting in 2027. Base's fleet already holds over 750 MWh across 30,000-plus homes. Both deals treat distributed capacity as something you contract for, not a pilot, and that matters when PJM capacity is clearing at record prices.

Brattle's study puts a number on the upside: a 4% peak reduction nationwide would lift residential VPP capacity from about 10 GW to 35 GW, worth $4 billion to $7 billion a year, and could cut bill increases by more than half. For suppliers and marketers, more flexible capacity at the peak is a soft ceiling on peak prices and capacity costs over time. It is also a competitive signal, because the same batteries that shave a peak can be aggregated and optimized by whoever has the dispatch and settlement capability to do it.