Summary
- ERCOT has delayed progression of data-centre large loads through its Batch Zero process pending verification.
- The grid operator said in June that 89% of more than 438GW of tracked large-load requests came from data centres.
- Britain has separately identified around 140 transmission-queue data centres representing approximately 50GW of requested capacity.
The Electric Reliability Council of Texas has delayed the next step in its Batch Zero large-load process while it verifies data centre and other major electricity-demand projects in the queue.
ERCOT said it received a direction from Governor Greg Abbott on 3 August requiring a verification process before data centre large loads could advance through Batch Zero. The grid operator subsequently delayed issuing conditional classifications that transmission and distribution providers had expected to receive while it completed data validation and due diligence.
The scale of the pipeline explains the intervention. When regulators approved Batch Zero in June, ERCOT said it was tracking more than 438GW of large-load requests, with almost 89% attributable to data centres. The framework covers qualified projects of 75MW or more and was designed to study their combined effect on the grid rather than processing applications independently.
A queue too large to treat literally
The difficulty is that a connection application is not the same thing as a data centre that will be financed, built, fitted out, and occupied. Developers may make applications at multiple potential sites or before financing and customer commitments are mature, while utilities still have to consider the possibility that the load will materialise.
Reuters reported on 1 September that US grid operators are confronting more than 700GW of data centre requests nationally, creating concern about so-called ghost demand. In Texas, the verification work is intended to distinguish projects with credible ownership, funding, development, and customer characteristics from demand that may never be built.
That distinction affects far more than an administrative queue. Network planners use expected demand to determine where substations, transmission lines, generation, and other infrastructure will be required. If the forecast is materially inflated, capital can be directed towards reinforcement for projects that disappear while viable customers wait behind them.
Texas had already been redesigning its process before the delay. Batch Zero groups large projects together so ERCOT can assess system-wide reliability effects, available capacity, and transmission requirements. The verification order moves project credibility further towards the front of that assessment.
DataCentral has also tracked large-load reforms in PJM, where rapid data centre growth has pushed the US grid operator towards new registration, curtailment, and procurement mechanisms.
Britain is confronting the same queue arithmetic
The numbers are different, but the planning problem is recognisable in Britain. The UK government’s 2026 consultation on strategic electricity demand said National Energy System Operator analysis had identified around 140 data centres in the transmission queue representing approximately 50GW.
The government described that level as far beyond market forecasts and what the electricity network could reasonably support. Its concern is that speculation and oversubscription can lead to more expensive network planning and delay connections for projects that are more likely to proceed.
British queue reform is consequently moving towards stronger financial commitments and project milestones, using evidence of maturity to determine which data centre projects retain scarce capacity.
Texas provides a more abrupt version of the same discipline. Rather than assuming every megawatt in the application stack represents future consumption, the system operator is being required to test the projects before advancing the study process.
That does not mean the underlying demand is fictitious. AI and cloud investment are creating substantial new electricity requirements in both North America and Europe. The issue is whether a queue containing several competing versions of that demand can be used directly as a forecast of what networks must build.
The outcome of the Texas verification process will therefore be important beyond the state. If it removes a large portion of requested capacity, it will provide a practical measure of how much speculative demand can accumulate when grid access itself becomes a scarce development asset.
European policymakers are already trying to solve the same problem before scarce capacity is locked behind projects that exist more convincingly in connection paperwork than they do in land, finance, customers, and construction programmes.
