Half of US data center pipeline is stalled. Who gets to move the costs to your bill while they wait?

A Bloomberg analysis reported by the Economic Times found that up to 50 percent of all planned U.S. data center developments are now at risk of being delayed or canceled altogether.^[1] The culprits are familiar: aging electrical grids, multi-year backlogs for high-voltage transformers and switchgear, regional utilities drowning in interconnection requests, and growing community resistance.^[1] The story reads as an infrastructure problem. It is actually a cost-allocation trap.

Here is the mechanism: utilities filed load forecasts assuming all or most of that pipeline would build on schedule. Those forecasts justified specific capex programs for new generation and transmission. State regulators approved the capex, and utilities added the costs to rates. But the pipeline is not materializing. In Texas, ERCOT's interconnection queue holds roughly 474 GW of requests, more than five times the grid's record peak demand, with data centers accounting for about 90 percent of that total.^[4] The key question regulators are not asking: how much of that 474 GW is actually contracted and collateralized versus merely announced or speculative? The Harvard Electricity Law Initiative found that utilities routinely justify rate-base investments using optimistic data-center load forecasts, then keep those investments in the rate base even when the load does not show up. Ratepayers eat the stranded cost. Hyperscalers pay for capacity only if they use it.

The freeze itself is overdue accountability. Gov. Greg Abbott directed Texas regulators to audit every data center in the grid queue pending verification of tax breaks, power use, water impact, and ownership.^[3] He pledged to deny connection to any facility that fails the audit, citing risk to grid reliability.^[3] At least 15 states have weighed pauses on data center development, and at least 100 localities have already approved their own.^[5] But the pause does not undo the capex already approved. The question now is whether utilities will be forced to justify the new gas plants and transmission lines they built on forecast assumptions that reality has already disproven.

The load-growth revision is enormous. Lawrence Berkeley National Laboratory predicts data center demand will grow from roughly 176 terawatt-hours in 2023 (about 4.4% of total U.S. electricity consumption) to between 325 and 580 TWh (6.7 to 12.0%) by 2028.^[8] That range itself reflects deep uncertainty. But utilities and regulators have been treating the high end as a floor, approving generation and transmission as if hyperscale buildout would continue uninterrupted. Texas data center electricity demand is expected to grow from 7.7 GW in 2025 to 14.5 GW by a future year, but that projection was made before the freeze, and grid constraints are now the binding constraint, not demand.^[7] The mismatch between forecast and physics has halted projects. It should also halt any rate-case request predicated on the old forecast.

The solution is not to wait for the freeze to thaw. It is to isolate hyperscaler cost from ratepayer exposure: bring-your-own-generation, ratepayer-protection tariffs with high minimum-take ratchets and collateral, and mandatory curtailability commitments that allow new load to connect without requiring new rate-base generation. Virginia's GS-5 large-load tariff, Ohio's AEP special contract, and Oregon's Schedule 96 show that utilities can offer differentiated tariffs with 10 to 14-year terms, demand ratchets at 60 to 85 percent, collateral at roughly $1.5 million per megawatt, and 100 percent cost responsibility for dedicated upgrades. Under such a structure, the utility still profits from connection and operating fees, but hyperscalers pay for stranded capacity if their load underperforms. Ratepayers do not. Interrogate any pending data-center rate case or interconnection tariff filing: does it contain a high minimum-take ratchet, collateral, and a long term? If not, it is a bet with other people's money.