A sudden, city-sized loss of electricity demand could push the Texas grid toward outages this summer after several planned data centre and crypto mining connections failed voltage ride-through tests. The Electric Reliability Council of Texas reported on 21 May that simulations showed groups of large users disconnected when voltage disturbances were applied, and that some could trip more than 5,000 megawatts each. ERCOT said it reviewed roughly 20 gigawatts of prospective large customers and highlighted eight projects totalling about 3.9 gigawatts that had aimed to begin operations before 1 July. The operator is reviewing the failed tests and drawing up protective plans that could delay start-ups or require stricter interconnection conditions.
The immediate consequence would be a sharp surplus of generation relative to load that can force power plants offline, ERCOT warned, and the trigger this time is failures in voltage ride-through testing by new data processing and crypto facilities.
In a report dated 21 May, ERCOT said it examined roughly 20 gigawatts of large customers seeking to connect to the grid and found reliability problems that directly affect whether new projects can come online later this month. The operator identified four groups of large electricity users, including data centres and crypto mining sites, that abruptly disconnected during simulated voltage disturbances. Simulations showed each of those groups could cause more than 5,000 megawatts of demand tripping under certain fault conditions, an amount ERCOT described as equivalent to the electricity consumption of a large city such as Boston.
What failed the tests
The failures concern voltage ride-through capability, a specification that facilities must meet so they remain connected during brief disturbances in voltage and frequency rather than disconnecting. Unlike traditional industrial loads that draw power steadily, many data centres and crypto miners are engineered to cut their connection at the first sign of trouble to protect equipment. That design choice makes them less predictable and potentially destabilising when disturbances occur.
ERCOT said eight projects totalling roughly 3.9 gigawatts were aiming to start up before 1 July. Whether those projects can proceed on that timeline depends on the outcome of the ongoing review and any enforcement of tightened interconnection requirements. The operator emphasised that the problem isn't merely theoretical: voltage ride-through failures have been recorded in practice. Since 2023, ERCOT has logged at least 26 events in which data centres or crypto mining facilities abruptly disconnected because they couldn't handle disturbances in voltage or frequency.
ERCOT pointed to a notable incident in December 2022 in west Texas when a failed transformer at a substation precipitated a mass unplugging. Nearly 400 facilities, including crypto miners and data centres, disconnected without warning. That episode produced a surplus of roughly 1,700 megawatts of electricity, about 5 percent of the grid's total demand at the time, and forced 112 megawatts of power generation to shut down. The historic example underlines how an abrupt drop in load can propagate through a system designed to balance generation and demand every second.
How ERCOT is responding
Regulators have been tightening interconnection and performance requirements for large customers seeking to join the grid, and ERCOT said its board has made voltage ride-through performance a top priority as more data processing and crypto mining capacity seeks to connect. The operator is reviewing the recent test failures and preparing protective plans. Those could include additional conditions for interconnection, enhanced testing, or delays to planned start-ups until compliance is demonstrated.
ERCOT's review is focused on ensuring that new entrants can remain on the network through typical disturbances rather than disconnecting in a way that creates a rapid imbalance. That approach aims to prevent the chain reaction that can follow a sudden fall in demand: generators see a surplus and some can trip offline, which in turn can create frequency deviations and wider instability. The risk is particularly acute in peak seasons when the margin between available generation and load is thinner, which is why the potential for trouble this summer has prompted urgent attention.
The mix of customers seeking to connect has shifted in recent years, with growth in facilities that process large volumes of data and in energy-intensive crypto mining operations. ERCOT's review of roughly 20 gigawatts of prospective large customers reflects that shift. For grid planners, the challenge is twofold: to accommodate economic activity that consumes significant electricity and to do so without creating new failure modes for system reliability.
Practical remedies under consideration include more rigorous pre-commissioning tests, staged or monitored initial connection procedures, and contractual obligations that require facilities to demonstrate sustained ride-through performance. ERCOT made clear that enforcement is a real possibility: projects that can't demonstrate compliance may face additional conditions or be delayed. That stance signals a tighter regulatory posture for interconnections that present systemic risk.
The immediate calendar milestone is 1 July, when several of the projects identified in the report had aimed to begin operations. ERCOT's review and any subsequent enforcement will determine whether those start-up plans remain intact, and whether operators of new facilities must alter designs or control systems to meet the stricter standards.
Related Articles
- Sriram Krishnan to leave White House AI adviser role in June
- Fortnite offline for hours as Epic deploys Chapter 7 update
- Chesky backs Airbnb AI lab despite doubts over LLMs
By 1 July ERCOT will decide whether roughly 3.9 gigawatts of projects can proceed under the updated interconnection requirements, or whether further delays and conditions are needed to protect grid stability. Originally reported by Reuters.
This article was created with AI assistance.