The Batch Zero Numbers
CenterPoint Energy told investors on July 28, 2026 that it is raising its 10-year capital investment plan from $65.5 billion to $66.7 billion, a $1.2 billion increase driven almost entirely by one thing: the amount of large electrical load lining up to connect to its Houston-area grid. The company said it had submitted more than 17 gigawatts of large-load projects through ERCOT's Batch Zero process, the first cohort to run under Texas's redesigned large-load interconnection rules, and that roughly 14 gigawatts of that total is on track to qualify as base load or studied load.
Fourteen gigawatts is not a small number next to Houston Electric's current peak system demand of 21 gigawatts. CenterPoint's own framing, a more than 65 percent increase in system peak, is the kind of number normally reserved for multi-decade grid buildouts, not a single earnings call. Management paired it with a specific promise: over the next 10 years, the new connections are forecast to reduce residential and commercial delivery charges by at least $5 billion, because large loads absorb a share of fixed grid costs that would otherwise fall on smaller customers. Nearly all of the 14 gigawatts is projected to be energized by 2031.
Why This Is Not the Usual Phantom-Demand Story
Utility claims about data-center demand have earned skepticism for a reason. Generation and load interconnection queues across the United States are littered with megawatts that were reserved, counted in a press release, and then quietly withdrawn. Lawrence Berkeley National Laboratory's Queued Up 2026 report, published in June 2026 and covering roughly 98 percent of US installed generating capacity across seven ISOs and 50 non-ISO balancing areas, found that of the capacity that requested interconnection between 2000 and 2020, only 13 percent had actually reached commercial operation by the end of 2025. Seventy-five percent had been withdrawn.
ERCOT's new Large Load Interconnection Process, formalized through Nodal Protocol Revision Request 1234 and Planning Guide Revision Request 115 and tightened further by Texas Senate Bill 6, was built specifically to break that pattern for large loads. Before ERCOT even starts studying a project, the customer must show site control and development readiness, then post financial security of $50,000 per MW of requested peak demand. If the project is later delayed, downsized, or withdrawn, the customer gets back only 20 percent of that posted collateral; the remaining 80 percent flows to the transmission provider's rate base. Any payment the customer made toward the direct cost of its own interconnection is not refundable at all, under any circumstances, including a voluntary withdrawal. That is a real exit cost, not a press-release commitment.
The Tranche the Headline Number Hides
CenterPoint's own numbers already show that not all 14 gigawatts sit at the same stage of certainty. Of the total, roughly 10 gigawatts has cleared both of ERCOT's required interconnection studies and qualifies for base-load designation, the strongest available status under the new process. The remaining roughly 4 gigawatts qualifies only as studied load, meaning it has cleared one of the two required studies but not the other. Both tranches count toward the 14 gigawatts CenterPoint cites in its 65 percent peak-demand comparison and its $5 billion savings forecast. Neither the July 28 release nor CenterPoint's public commentary breaks out how much of the projected $5 billion in savings depends on the fully studied 10 gigawatts versus the partially studied 4 gigawatts.
That gap matters because the forfeiture mechanics apply the same way regardless of tranche, but the probability of a project actually reaching energization is not the same for a fully studied load as for one still missing a study. A number that blends a near-certain tranche with a less-certain one, without separating them, is not dishonest, but it is not the whole picture either.
What an Owner Should Actually Watch
For an industrial or commercial ratepayer inside ERCOT, or for anyone weighing a Texas site-selection decision against CenterPoint's forecast, the practical read is to treat the 65 percent peak-demand figure as design capacity CenterPoint is building toward, not as a guarantee that $5 billion in savings lands on schedule. The collateral and non-refundable construction-contribution rules genuinely reduce the odds of a repeat of the national 75 percent withdrawal pattern documented by Berkeley Lab, and that is real, verifiable progress over the old speculative-queue model. But it reduces the odds; it does not eliminate them, especially for the roughly 4 gigawatts that has not yet cleared both studies.
The transferable question for any utility citing similar large-load savings math, in Texas or elsewhere, is simple: does the number reflect a full accounting of the interconnection process's exit-cost design, including what happens to forfeited collateral and non-refundable payments if load does not fully materialize, or is it a gross projection that assumes 100 percent of interconnection-eligible capacity converts to billed load on schedule. CenterPoint's Batch Zero filing is a genuinely stronger data point than the industry's history of phantom demand. It is still a forecast, built partly on a tranche that has not finished its own qualification process.
Read next: FERC's 60-Day Grid Deadline Lands This Week | PJM Opens a 10-Project Fast Lane Inside a 220GW Queue



