Data Center Ghost Demand: Texas Pauses Grid Approvals

Data center ghost demand has become difficult for U.S. grid operators to distinguish from credible capacity requests, creating a planning problem that could affect where and how quickly new AI infrastructure gets connected. Texas has responded by pausing approvals for new data center grid connections while projects in the Electric Reliability Council of Texas, or ERCOT, process are verified.

The term “ghost demand” describes electricity requests associated with projects that may be duplicated across utilities, lack sufficient financing, represent competing site options, or never reach construction. Reuters reported on September 1 that data centers have requested more than 700GW of electricity across the United States—more than ten times the estimated current electricity use of U.S. data centers.

For operators with credible developments, this is more than a forecasting problem. Speculative megawatts can compete for the same engineering attention, transmission planning, equipment, and grid capacity as projects that are financed and ready to build.

The Texas Data Center Queue At A Glance

Reported measure	Approximate scale	Why it matters
U.S. data center electricity requests	More than 700GW	Includes requests at different maturity levels and may contain duplication
ERCOT prospective large-load requests	About 474GW	More than five times Texas’ record ERCOT peak demand
Share linked to data centers	Approximately 90%	Makes data center verification central to Texas grid planning
Batch Zero projects under review	Approximately 250–300	Projects must be assessed for readiness and commercial credibility

These figures describe requests, not a reliable forecast of actual consumption. Some projects will proceed, others will shrink, move, or disappear, and some developers may have requested power for several possible sites before selecting one.

Texas Is Separating Projects From Proposals

Texas Governor Greg Abbott ordered an audit of data centers progressing through ERCOT’s interconnection process in August 2026, temporarily stopping new approvals while projects are verified. The directive followed an extraordinary increase in prospective large-load requests.

ERCOT officials previously indicated that the audit would cover approximately 250 to 300 projects in its Batch Zero process, most of them data centers, representing roughly 200GW of prospective load. The broader queue was larger still. Reporting in August placed the number of energization requests near 2,000 and prospective demand near 466GW, with data centers responsible for about 90%.

The exact totals have varied as queues have evolved, but the planning dilemma is consistent: Texas cannot reasonably assume that every requested megawatt will become operating load. The state must identify which projects are commercially viable, technically mature, and likely to meet their schedules.

Why Data Center Power Demand Can Be Overstated

Developers have strong incentives to seek electricity early because power availability increasingly determines whether a project is commercially viable. A site without a credible energization path may be impossible to finance or market to customers.

The same prospective requirement can appear in several places. A developer considering multiple markets may submit requests to several utilities before selecting a final site. Different development partners may also pursue overlapping options for the same customer requirement.

Projects can change ownership, capacity, schedules, or financing assumptions while remaining visible in planning pipelines. A preliminary 500MW concept can later become a 100MW first phase, yet the earlier request may continue influencing forecasts until records are updated.

This leaves planners asking which megawatts represent committed future consumption and which represent commercial optionality. Without consistent disclosure, maturity criteria, and rules for removing inactive projects, the two can look similar.

The Cost Of Getting The Forecast Wrong

Underestimating genuine demand risks leaving data centers without enough generation and transmission. It can delay projects, constrain regional economic development, and force operators toward expensive onsite power solutions.

Overestimating demand creates a different problem. Utilities can spend billions preparing substations, transmission lines, generation, and related infrastructure for loads that never materialize. Depending on the regulatory structure, other electricity customers may ultimately carry part of that cost.

Reuters reported that some utilities are requiring stronger financial commitments. Exelon and AEP Ohio saw projected demand fall substantially after introducing larger deposits or commitment requirements. That indicates that at least some apparent scarcity is amplified by projects that are not equally mature.

The stakes grow with project scale. An inaccurate forecast for several gigawatts can influence new transmission corridors, generation procurement, transformer orders, and utility rate planning across an entire region.

Grid Access Is Becoming A Qualification Process

Obtaining a place in an interconnection queue may no longer be enough. Grid operators and regulators increasingly want evidence that projects are credible. That can include proof of land control, financing, development milestones, equipment commitments, customer backing, engineering progress, and willingness to make non-refundable payments.

Texas’ Batch Zero framework was designed to evaluate large-load projects through study fees, financial commitments, and readiness requirements. The audit adds further scrutiny around project ownership, funding, water and energy requirements, and development status.

This changes data center site-selection strategy. Developers accustomed to maintaining several potential sites while negotiating power may face higher costs for keeping those options open. Well-capitalized hyperscalers and established operators may be better positioned to satisfy stricter rules than speculative developers.

Deposits Can Filter Demand—But Create Risk

Financial security is one of the clearest ways to test commitment. If a developer must place meaningful capital at risk, it has a stronger incentive to submit realistic capacity requirements and withdraw inactive options.

Texas projects entering Batch Zero reportedly had to post security tied to requested megawatts. Larger non-refundable portions could further discourage speculative applications, but developers argue that changing rules after deposits have been paid can create uncertainty and strand capital.

The challenge is designing requirements that filter weak proposals without penalizing credible projects whose schedules depend on decisions by utilities, regulators, or equipment suppliers. Transparent milestones and consistent treatment will be essential.

The Problem Extends Beyond Texas

Texas is the clearest current example, but the underlying issue is national. Regulators and utilities in other markets are also reconsidering how very large data center electricity requests should be handled. Pennsylvania and Ohio are among the states tightening requirements or debating cost protections for other ratepayers.

The issue will become more important as AI factory campuses move from tens of megawatts toward hundreds of megawatts and, in some proposals, gigawatt-scale developments. At that scale, a duplicate or immature request is not a rounding error.

Regional differences will remain. Some utilities operate in vertically integrated markets, while others sit inside regional transmission organizations with different planning and cost-allocation rules. A national standard may be difficult, but common definitions of project readiness could improve the quality of demand forecasts.

What Credible Developers Should Prepare

• Document site control and the exact development entity requesting service.
• Provide realistic phased load profiles rather than only ultimate campus capacity.
• Disclose duplicate or alternative requests submitted in other locations.
• Demonstrate financing, customer support, and procurement milestones.
• Align energization dates with transformer, switchgear, construction, and commissioning schedules.
• Show how onsite generation, storage, or demand flexibility affects grid requirements.
• Model transmission and substation costs and identify who will bear them.
• Maintain evidence that water, permits, fiber, and land are progressing alongside power.

These steps will not guarantee an interconnection, but they can make a project easier to distinguish from speculative demand. They also improve internal investment decisions by exposing mismatches between commercial schedules and physical infrastructure.

What Data Center Leaders Should Watch Next

The industry should expect more milestone requirements, proof of site control, customer commitments, and stricter rules around duplicate requests. Utilities may also require phased studies so that capacity allocations better reflect what a campus can realistically consume during its first years.

For credible developers, stronger verification could ultimately be beneficial if it removes inactive demand from congested queues. A clearer pipeline can help utilities prioritize engineering work and give serious projects more dependable timelines.

There is a trade-off. Requirements intended to filter unrealistic proposals can increase upfront capital and make early-stage site development more expensive. Smaller developers may struggle to compete even when their projects are sound.

Operators should also watch how utilities allocate network-upgrade costs and whether new tariffs protect residential and small-business customers. The policy response will shape project economics as much as the technical availability of power.

Frequently Asked Questions

What Is Data Center Ghost Demand?

Ghost demand refers to electricity requests that may be duplicated, speculative, insufficiently financed, overstated, or unlikely to become operating load. It makes grid planning harder because immature proposals can resemble committed projects.

Why Did Texas Pause Data Center Grid Approvals?

Texas paused new approvals while ERCOT and state regulators audit projects in the interconnection process. The goal is to verify ownership, financing, infrastructure requirements, and readiness before treating requested megawatts as credible future demand.

Does The Pause Mean Texas Lacks Power?

Texas still faces genuine growth in electricity demand and infrastructure requirements. The pause addresses the quality of the queue as well as the quantity of power available. Both constraints can exist simultaneously.

Conclusion

The data center industry has spent years asking whether grids can provide enough electricity for AI. Texas is now forcing a second question: how much of the requested electricity will actually be needed?

Power scarcity remains real, but a grid cannot be planned efficiently when speculative requests and committed developments look identical. Stronger deposits, transparent milestones, and clearer readiness standards may remove ghost demand and improve planning.

For operators, the next phase of the power race will involve more than securing megawatts. Developers may increasingly have to prove that they are financed, mature, and committed enough to receive them.