ERCOT recently took a step much further than other system operators have been willing to take; it sorted the interconnection queue between load that is real and load that is unlikely.
ERCOT sorted the queue into tiers and shared numbers, all true, that large load requests totaled about 474 GW (with around 90% of the requests from data centers, PDF Link). ERCOT placed 205 GW of that load into the batch zero study, classifying 66.4 GW, or 204 the projects as base load. Of that, 158 projects, 127.9 GW, met the qualification to be included as a studied load. And only about 3.9 GW of that load counted as large load peak consumption.
Late in 2025 Exelon applied a filter to its own queue of data center load requests with firm commitments, and found the portion of high-probability dropped to 40%. Reviewing 18 GW of power need at the end of 2025, Exelon Chief Financial Officer Jeanne Jones and her team saw the demand drop to 11 GW. The filter was a requirement to sign transmission security agreements backed by collateral and committed revenue and shortfall payments to be considered real.
That gap, the one ERCOT revealed with batch zero and the one Exelon calculated, is what we in the industry now call bragawatts. That is, announced capacity that may never get built. While the word captures the imagination (it is quite catchy), it isn’t a true reflection of what is actually going on. Bragawatts are not lies. Bragawatts are a predictable result of a rush to get an option on scarce grid capacity. A queue position is not a lie.
