ERCOT
Its own interconnection, energy-only by conviction, the largest renewables build and the largest data-centre queue in the country, and Uri as the stress test everyone still argues about.
On 4 May 1976, a Texas utility closed a switch and sent power from Texas into Oklahoma for a few hours. Central and South West Corporation did it deliberately, and the intention was legal rather than operational. If Texas electricity crossed a state line, Texas electricity was interstate commerce, and interstate commerce belonged to federal regulators.
The episode is remembered as the Midnight Connection, and the litigation it started ran for years. Texas won. The courts held that a few hours of deliberate interconnection had not converted an intrastate grid into an interstate one, and the Electric Reliability Council of Texas remains outside the jurisdiction of the Federal Energy Regulatory Commission today.
Everything else in this chapter follows from that outcome. ERCOT covers roughly 90% of Texas load, synchronises with nobody, and connects to its neighbours only through a small number of direct-current ties that move power without joining the systems electrically. Its rules are written by the Public Utility Commission of Texas and the Texas legislature. It is the only large power market in the developed world that can redesign itself without asking a national regulator, and it has used that freedom repeatedly.
Three answers, all at one extreme
Part Three set out three questions. Texas answered each of them in the direction furthest from the vertically integrated model, and answered them as a set rather than one at a time.
Table 18-1: How ERCOT answered the three axes
| Axis | ERCOT | Consequence |
|---|---|---|
| Who dispatches | Independent operator, owning no generation, under state rather than federal oversight | Rule changes are a state matter, so they happen faster than anywhere else |
| How location is priced | Nodal since December 2010 | Congestion appears in prices, so the wind build followed the transmission |
| How capacity is paid | Energy-only, with an operating reserve demand curve and a 5,000 dollar cap | No capacity payment exists, so every plant lives on energy and ancillary revenue |
Chapter 15 covered the move to nodal pricing and Chapter 12 covered the scarcity machinery, so neither is repeated here. What matters is that the three choices reinforce one another. An energy-only market needs prices that reach high enough to finance plant, a nodal market delivers those prices at the specific locations where the shortage exists, and a state regulator can adjust the mechanism without a federal proceeding when the result proves unsatisfactory.
The largest build in the country
The usual expectation would be that a market with no capacity payment and no central plan would struggle to attract construction. Texas has the opposite problem.
Around 31 gigawatts of solar has connected since 2020, with a comparable quantity in development. Grid-scale batteries reached about 14 gigawatts by March 2026, close to double the 7.8 gigawatts installed in January 2025. Texas passed California in wind years ago and has been adding solar faster than any other state.
Three things explain it, and only one is about the market design. Texas has land, sun and wind in quantity, and a permitting regime that allows a project to be built. It has a competitive retail market that produces buyers for long-term contracts. And it connects generation quickly, using an approach closer to connect-and-manage than the study-everything queues that have accumulated multi-year backlogs in most other American markets. A generator in ERCOT can be told it may connect and accept that it will sometimes be curtailed, rather than waiting years for a study that allocates the cost of network reinforcement first.
The transmission followed rather than led, and it was planned rather than priced. West Texas wind was stranded behind constraints for years, which is the story Chapter 10 used to introduce congestion, until the state built the Competitive Renewable Energy Zone lines to move it east. That was a legislated infrastructure programme rather than a market outcome, which is a useful corrective to any account of Texas as a market purist.
The load arriving now
On 22 July 2026, ERCOT set an all-time peak of 91,308 megawatts, breaking the previous day's record of 87,403, which had itself broken the 85,508 megawatts set in August 2023. Two records in two days is unusual, and it is the visible surface of something much larger sitting in the queue.
ERCOT entered 2026 with more than 233 gigawatts of large-load interconnection requests, up around 269% in a year, with data centres roughly 77% of them. Later tracking put the figure above 438 gigawatts, approaching 89% data centres. Against a system whose all-time peak is around 91 gigawatts, those numbers describe requests for several times the entire existing grid.
Treat large-load queue figures with care. A developer shopping one project around several sites appears several times, and a request costs far less than a project. The queue measures interest rather than demand, which is precisely why ERCOT needed a way to tell the two apart.
That is what Batch Zero is for. Approved by the Public Utility Commission of Texas in June 2026 and applying to loads of 75 megawatts or more, it stops processing large-load requests one at a time. Qualifying projects are studied together, allocated across a six-year horizon, and folded into a single statewide transmission plan. A system that had built its reputation on connecting things quickly discovered that first-come first-served stops working when the requests exceed the grid several times over.
The bet, restated
ERCOT is the clearest working example of the argument Chapter 12 described, and it now faces the version of that argument its own success created.
The energy-only case says scarcity prices will call forth capacity when capacity is needed. Texas has demonstrated the mechanism works for generation that is cheap to build and quick to connect. Solar and batteries arrived in quantity, fast, with no capacity payment. Batteries in particular responded to exactly the signals the design intended, first chasing the ancillary service revenue described in Chapter 13, then moving to energy arbitrage as those returns compressed.
The unresolved part is whether the same signal will produce the things a winter evening needs. A four-hour battery covers a four-hour shortfall. Uri lasted days. After Uri the offer cap came down from 9,000 dollars to 5,000, which lowered the ceiling on exactly the revenue an energy-only market relies on to finance the plant that runs a handful of times a decade. The state has since added programmes to encourage dispatchable generation directly, which is a capacity mechanism in everything except the name it is given.
So the purest energy-only market in the world has spent the years since 2021 quietly adding instruments that pay for availability. That progression is worth watching rather than mocking. It is the same pressure that produced capacity mechanisms everywhere else, arriving in the one place that had most firmly decided it did not need one.
Why this is the chapter to reread
Every concept in Parts Two and Three appears in Texas in a form clear enough to see.
Negative prices from Chapter 11 occur across West Texas on windy nights. The scarcity pricing of Chapter 12 runs as an explicit formula rather than an accident. The ancillary market of Chapter 13 was competed to almost nothing by batteries inside three years. The nodal design of Chapter 15 exists because the state rebuilt its market to get it. The capacity argument of Chapter 16 plays out here as an experiment nobody else was willing to run at this scale.
Texas can do all of this because of a lawsuit it won in the 1970s about a switch closed for a few hours. The most consequential fact about the largest competitive electricity market in the United States is a question of jurisdiction.