Texas Has 238 Gigawatts of Requested Power and 3.9 Gigawatts Actually Drawing It
Federal forecasters cut the 2027 electricity load growth outlook for Texas to 6 percent from 14 percent in a single monthly cycle. Nothing about physical electricity consumption changed between those two publications. What changed is that the state paused new data center approvals and started asking which projects in its queue are real.
By Keith Reynolds | Publisher & Editor, ChargedUp!
A load forecast that falls by more than half in thirty days was not measuring demand. It was measuring interest. That distinction has direct consequences for property owners in three places: the utility capital spending that lands in your rate base for the next three decades, the price you pay for land marketed as power-advantaged, and the position of your own service request in a queue that may be substantially occupied by projects that will never be built.
The Funnel Nobody Publishes
The Electric Reliability Council of Texas (ERCOT) entered 2026 with more than 233 gigawatts of large load interconnection requests, roughly 77 percent of it from data centers, a figure that had grown almost 300 percent in a single year from 63 gigawatts at the end of 2024. ERCOT Vice President of System Planning Kristi Hobbs told the grid operator's board in December that rules built to evaluate 40 to 50 large loads at a time had been outgrown.
Set that against what is actually connected. As of March 2026 the large load queue stood at roughly 238.6 gigawatts. Approximately 9.0 gigawatts had received Approval to Energize. Observed peak consumption from large loads was roughly 3.9 gigawatts. The request-to-operating funnel is on the order of 1.6 percent.
For scale, ERCOT's all-time peak demand is 91.1 gigawatts, set on July 22, 2026, according to the U.S. Energy Information Administration's Hourly Electric Grid Monitor. The queue is more than two and a half times the largest amount of electricity Texas has ever used at one moment. The load actually operating from that queue is about four percent of a single summer peak.
Those queue and energization figures come from an interconnection tracker compiling ERCOT filings rather than from a single ERCOT publication, and they are a snapshot rather than an audited accounting. The order of magnitude is what matters, and it is corroborated by the grid operator's own description of a process it says it has outgrown.
Why Queues Inflate
The mechanism is well documented and unglamorous. Developers file multiple interconnection requests for the same project across different utility territories, or file early-phase requests for projects unlikely to be completed, because the cost of asking is low and the cost of missing a queue position is high. The World Resources Institute describes the result as double counting and phantom load that distorts utility resource planning, produced by requests from developers, private equity funds, land brokers and shell companies, many without site control, a construction timeline, or a signed customer.
Karl Rábago, a former Texas utility commissioner, put the economics plainly to Utility Dive: the phantom load problem arises because the cost of getting into a queue is lower than the weighted likelihood a developer will use the position, so developers buy queue positions all day long. Astrid Atkinson, a former Google engineering director now running grid software firm Camus Energy, told the same publication that interconnection requests conservatively run five to 10 times the number of data centers actually built.
Utilities have begun saying so themselves. Exelon has publicly assessed that only about 22 percent of its 65 gigawatt pipeline through 2040 is likely to materialize. CenterPoint Energy's data center interconnection requests in Houston went from 1 gigawatt to 25 gigawatts inside twelve months. Wood Mackenzie analysis reported by Bloomberg concluded that utilities nationally will realistically commit to roughly 28 percent of the more than 1,000 gigawatts of requests on file, with other analysts putting the figure lower still.
Where This Reaches Your Operating Statement
Four transmission mechanisms carry a distorted queue into property-level economics. None of them require a single data center to fail.
1. You Are Already Funding the Overbuild
Utilities justify capital spending against load forecasts, and those forecasts draw on queue data. Approved capital enters rate base and is recovered from all customers, typically over thirty to forty years. If forecast load does not appear, the transmission lines, substations and generation built to serve it do not disappear from the rate base. They are recovered from the customers who remain, which includes every commercial building in the service territory.
Florida's public counsel, Walt Trierweiler, framed the exposure at the state's August 25 hearing on large load cost allocation: the risk of devastating economic loss in subsidization is significant should any one of these data centers fail to materialize, or fail to be used throughout its life. That is the same risk in every jurisdiction where a utility is building against a queue rather than against contracts.
At an 8 percent capitalization rate, every $1,000 of durable annual electricity cost added to a building's operating statement removes roughly $12,500 of asset value. A rate increase driven by capital recovery for load that never arrives is a permanent, compounding transfer from property income to utility return on equity.
2. The Scarcity Premium on Powered Land Rests on a Queue Number
A locational premium has formed for sites with confirmed electrical service, and the premium is real where the service is real. The diligence problem is that the scarcity argument supporting the pricing is frequently sourced to the same queue figures now under review. A broker citing gigawatts of regional demand as evidence that power is running out is citing interest, not commitments.
This cuts in two directions and both are actionable. An owner paying a premium for a power-advantaged site should require evidence in the form of an executed interconnection agreement, an energization date, or documented substation headroom, rather than a regional demand statistic. An owner holding a site with genuine confirmed capacity is holding something scarcer than the queue implies, because the funnel above suggests most competing requests will not convert.
3. Your Service Request May Be Queued Behind Vapor
This is the mechanism most often read backward. A building owner seeking a service upgrade, a fleet operator planning depot charging, or a developer needing capacity for an industrial tenant sits in a study process alongside speculative requests. Study sequence, cost allocation and timing are all affected by requests ahead in line, including requests that will be withdrawn.
Queue reform is therefore not a threat to a legitimate project. It is the most plausible path to a shorter timeline. Screening that removes speculative positions moves real requests forward, which means a pause framed in the press as a slowdown may function as an acceleration for owners with committed projects and documented load.
4. The Cost of Entry Is Rising for Everyone
The policy answer to a phantom queue is always to raise the cost of occupying it. Texas Senate Bill 6 directed the Public Utility Commission of Texas to write large load interconnection standards, and the proposed rule sets a 75 megawatt threshold with financial security, site control, batch study and curtailment requirements. A separate forecasting rule that took effect March 1, 2026 reaches down to a 10 megawatt threshold. Comparable measures exist elsewhere: ComEd charges $1 million for requests of 50 megawatts or more, Ohio requires new data centers to pay for at least 85 percent of projected energy use, and Virginia has moved large load customers onto 14-year contracts.
Thresholds that begin at large loads tend to migrate downward. An owner planning a service upgrade should assume that financial assurance, site control documentation and curtailment terms will be part of the application, and should price and schedule accordingly.
What to Do Before the Next Capital Decision
1. Ask your utility for the large load queue in your service territory, broken into requested, under study, and approved to energize. Georgia Power has published quarterly Large Load Economic Development Reports disclosing size, review status, proposed service date and load ramp for projects above 115 megawatts. That disclosure format exists and can be requested elsewhere.
2. Treat the load forecast as the contestable item in any pending rate case in your territory. It is the assumption that authorizes the capital spending you will pay for through 2060, and it receives markedly less scrutiny than the return on equity.
3. When evaluating a site priced on power availability, require documentary evidence. An executed interconnection agreement, a written energization date, or a utility letter confirming substation headroom. Regional demand statistics are not evidence of site-specific capacity.
4. If you hold a pending service request, ask your utility where you sit relative to large loads and whether pending screening rules would re-sequence the queue. Ask specifically whether your request is behind positions lacking site control.
5. Budget for higher application costs on any upgrade approaching the large load thresholds now being written, including financial security, site control documentation and negotiated curtailment terms.
The Case Against Overreading This
Discipline requires stating what the revision does not prove. It does not establish that data center demand is fake. A forecast cut in response to a policy pause may reflect a forecaster's caution about timing rather than a re-scrub of the underlying requests, and delayed load is not canceled load. The revision is specific to Texas. And 6 percent annual load growth remains extraordinary against a sector that spent two decades near zero.
The defensible reading is narrower and still consequential. The sensitivity of the forecast to a single administrative action reveals how much of it rested on inputs nobody had validated. A number that moves eight percentage points because a state started asking questions was carrying an error band that was never disclosed, and utility capital plans, transmission proposals and land pricing have been built on top of it.
The Bottom Line
Announced capacity and deliverable capacity have separated, and most of the market is still pricing off the announcements. In Texas the gap between the two currently runs from roughly 238 gigawatts requested to under 4 gigawatts operating.
Owners who continue to underwrite against queue-derived demand narratives will overpay for powered land, under-contest the rate cases that set their electricity expense for a generation, and misread queue reform as a delay when it is more likely a cleanup. The ones who ask for energization evidence instead of demand statistics will be buying and building against what the grid can actually deliver.
Sources
https://www.ascendanalytics.com/blog/large-load-interconnection-queues-data-center-grid-access
https://www.eia.gov/electricity/gridmonitor/dashboard/electric_overview/US48/US48
https://www.wri.org/insights/us-data-centers-electricity-demand
Frequently Asked Questions
What exactly did the federal forecast change?
The U.S. Energy Information Administration lowered its 2027 electricity load growth forecast for Texas to 6 percent from 14 percent in the July outlook. The August Short-Term Energy Outlook was released August 11, and the agency cited the state's August 3 pause on new data center development.
How large is the gap between requested and operating load in ERCOT?
The large load interconnection queue stood at roughly 238.6 gigawatts as of March 2026. Approximately 9.0 gigawatts had Approval to Energize and observed large load peak consumption was roughly 3.9 gigawatts. ERCOT's all-time system peak is 91.1 gigawatts, set July 22, 2026.
Why do interconnection queues contain projects that will not be built?
Filing an interconnection request is inexpensive relative to the value of holding a queue position, so developers file for the same project across multiple utility territories and sites, and file early-phase requests for projects that may never be financed. Grid operators have limited means of distinguishing a financed project from a placeholder.
Does this mean data center demand is not real?
No. Load growth in Texas remains extraordinary by historical standards even at the revised figure, and delayed projects are not canceled projects. What the revision demonstrates is that the forecast was highly sensitive to an administrative action, which indicates a substantial share of it rested on unvalidated requests.
Is queue reform bad for a property owner with a pending service request?
Generally the opposite. Screening that removes speculative positions shortens the study process for requests that have site control, documented load and financing. Owners with committed projects should expect higher application costs and better sequencing.
How does phantom load reach a building's net operating income?
Utilities justify capital spending against load forecasts. Approved capital enters rate base and is recovered from all customers over roughly thirty to forty years regardless of whether the forecast load appears. At an 8 percent capitalization rate, each $1,000 of added annual electricity cost removes about $12,500 of asset value.
