Cliff Daily — May 4.

Why it matters

This is a hard deadline that lands today and the RFI participant set is exactly the audience Cliff's PJM-side wedge would target — large-load participants and hyperscalers responding to PJM with how they want bilateral contracting structured for the September–March window. The RFI responses are not public-by-default but will surface in the FERC filing PJM makes once the design is finalized; worth watching the docket for the response set as it appears. Three structural points to pull into pjm-er26-1088-co-location.md and the broader PJM regulatory page: (1) the 14.9 GW figure is sized for the near-term subset of the 50–60 GW projected shortfall, not the full shortfall — meaning the 'backstop' framing is a deliberate hedge against the central-procurement-as-default outcome FirstEnergy is already opposing, (2) the bilateral phase being September-through-March is the same six-month window where ERCOT's Batch Zero LPC declarations are due (July 24) and the OH/PA/WV/MD local-moratorium calendar is most active — i.e., PJM's bilateral-contracting season runs straight through the worst of the local-opposition cycle, (3) the central-procurement Phase 2 in March 2027 is the residual mechanism that would aggregate any unfilled bilateral demand into a PJM-administered auction, structurally similar to ERCOT's Batch Zero allocation but with a single-shot true-up rather than rolling. Cliff's de-rate calculator and PJM-side underwriting math should treat the bilateral / central split as a regulatory bifurcation: bilateral-phase pricing is negotiated, central-phase pricing is auction-clearing — and a hyperscaler that fails to lock in a bilateral contract by March 2027 effectively pays whatever the central auction clears at, which is the Cliff-relevant tail-risk.

Why it matters

Three signals worth tracking in the regulatory-knowledge-graph and gas-feasibility page: (1) 106 GW of gas-fired in the new queue is roughly 10x the typical PJM gas-tied additions in a normal cycle and almost certainly contains a large share of co-located / data-center-tied generation rather than utility-procured capacity — meaning the new PJM queue is structurally a data-center-power buildout, not a generation-mix transition. The headline implication for Cliff is that gas-feasibility consolidation page becomes the right surface to model: which states' new-build gas-tied projects clear PJM's revised queue process versus get held up at state air permit / siting authority. (2) 18 GW of nuclear in the queue is the largest nuclear surge in the PJM queue history and almost entirely SMR / advanced-reactor proposals — a tell that the Vistra / Talen / Constellation co-location-with-existing-nuclear playbook has triggered a new-build-nuclear cohort that didn't exist 18 months ago. (3) The Expedited Interconnection Track is the structurally important piece for Cliff's developer-financing capital stack — 10 projects per year with state-siting-authority support is the explicit route by which a hyperscaler with state-political-capital can short-circuit the standard 4–6 year PJM queue. The state-siting-authority gate is exactly the political-economy chokepoint that's becoming the binding constraint as local moratoriums proliferate; an EIT-ineligible site is now a separate de-rate input. The April 29 queue reopening is the real demand-side data point that monetizes against the May 4 RFI's 14.9 GW backstop number — the queue tells you what gets built, the RFI tells you what PJM will buy if the queue under-delivers.

Why it matters

This is the cleanest 'years of soft-cost spend voided by a court ruling' data point Cliff has — Compass and QTS reportedly invested 'tens of millions of dollars' across years of approvals, and the AI-paperwork-automation market research already tracks PW Digital Gateway as $54M/yr CBA voided 2025 in the host-agreement table. Two new things to add to ai-paperwork-automation-market-research.md: (1) the Brookfield exit confirms the Compass-side losses are not litigated-recoverable — Brookfield wrote down the soft costs rather than appeal, which sets a precedent for how institutional capital marks similar entitlement risk on the rest of the NoVa pipeline. The implied per-acre soft-cost loss across the 2,100 acres is on the order of $20–40K/acre for Compass's share, a clean upper bound on hyperscale-corridor entitlement-risk exposure. (2) The QTS appeal continuing while Compass exits is the first observable bifurcation between two co-developers on the same corridor — QTS is taking a different risk view than Brookfield/Compass, which materially affects how the VA Supreme Court reads the case (one defendant, not two, with mixed standing). For the ercot-large-load-underwriting and de-rate-calculator GTM: this is the canonical example of why submission-readiness underwriting needs to model not just procedural compliance but the political-economy chokepoint in the appeals process — a 27-hour hearing voided a 2-year approval, and the appeals path took another 8 months to fully unwind. That timeline (procedural defects alleged → trial court invalidation → appellate confirmation → board stops defending → developer walks) is now the canonical worst-case for VA hyperscale entitlement, and it's directly inputtable into the de-rate calculator's regulatory-risk module.

Why it matters

This is a structurally distinct BTM template from yesterday's Joule 4 GW Utah lead and worth modeling separately in BTM-product-surface.md. Yesterday's Joule deployment is *permanent* off-grid (no interconnect, ever); Fleet's Storey filing is *transitional* off-grid (BTM gens operate 2–3 years until utility delivers permanent service). The transitional model is materially more capital-efficient per MW because (a) air-permit envelope is smaller for a 2–3 year operating life — RICE NESHAP / Title V applicability turn on hours-of-operation thresholds, and a 24-month window can fit under the synthetic-minor or short-term construction-permit envelope rather than a full Title V major-source filing, (b) Cat / Cummins genset financing is markedly cheaper for 2–3 year contracts than for 20-year permanent BTM, (c) the eventual switchover to grid service captures the utility-tariff economics once the queue clears. Tract's existing wiki entry (m-and-a-map.md, landgate.md, byog-byong-state-patchwork.md) already flags Tract as the sandbox-developer model; Fleet's 350 MW Nevada filing is the first observable Tract-led BTM template at scale and the first one tied to a named hyperscaler tenant (Nvidia per Bloomberg). The 'over-the-counter permits' inside TRIC's core area is also the structural answer to the Camden GA / Sedgwick KS local-moratorium pattern from yesterday — when the master-developer agreement pre-grants permitting authority below the county-commissioner level, the moratorium chokepoint disappears entirely. Worth opening a new strategy/tract-fleet-storey-transitional-btm.md page tracking the filing through the Nevada Public Utilities Commission and the air-permit module of the de-rate calculator should now have transitional-BTM as a discrete deployment mode alongside permanent-BTM and grid-tied. The 2-3 year window is also a tell that NV Energy's Greenlink transmission is the binding constraint and Fleet has chosen to bridge with gensets rather than wait — that's the same arbitrage pattern the de-rate calculator models for ERCOT Batch Zero in Texas, with Nevada as a clean parallel.

Why it matters

Three near-simultaneous suits (IL, MN, OH) targeting hyperscaler-tied data centers in the same week is a meaningful escalation of the local-opposition mechanism beyond the moratorium-vote channel covered yesterday. Worth tracking in byog-byong-state-patchwork.md as a separate 'litigation-pause' category distinct from 'moratorium-pause' — the procedural surface is different (zoning-defect lawsuits seek to invalidate already-issued approvals; moratoriums freeze new applications), and the resolution path is different (settlement, injunction, or appeal — typically 6–18 months versus the 90-day-rolling moratorium pattern). The Hermantown vote today is also a same-day pivot point: city council can vote up the Google development agreement before the lawsuit is heard, but the suit then becomes a post-approval challenge similar to PW Digital Gateway. This is the chain that Compass just walked away from — and it's now duplicating in the rust-belt corridor (OH) and the Upper Midwest (MN, IL). For Cliff's submission validator product, the implication is that 'submission-ready' must include an injunction-resilience module: did the public-notice path comply with state-specific procedural minimums that survive judicial review? PW Digital Gateway invalidation came from procedural defects in the 27-hour hearing, not substantive zoning grounds — and the Hermantown / Yorkville complaints are reading similarly. Open a new strategy/data-center-litigation-tracker.md page to track these systematically; will compound into a litigation-outcomes corpus alongside the approval-outcomes corpus already mapped in competitive-landscape-and-adjacencies.md.

Why it matters

Both moves are MISO-Central / PJM-West concentration plays that further reinforce the Indiana / Ohio / Wisconsin rust-belt corridor as the new center-of-gravity for hyperscale capacity, displacing growth from saturated NoVa and Phoenix. AWS's 2.4 GW Indiana number maps directly onto Meta's 1 GW Indiana announcement in Q1 capex (yesterday's coverage) — combined Big-3 concentration in Indiana is now north of 4 GW announced. That's exactly the pipeline that the Ohio constitutional ballot initiative (yesterday's tail-risk section) would freeze if it qualifies and passes. The Mount Pleasant village approval of 15 additional buildings is also worth contrasting against the Hermantown / Yorkville local-opposition pattern in the section above — Mount Pleasant is the model of *successful* hyperscale entitlement (2017 deal still expanding, no court challenges, Foxconn-pivoted-to-Microsoft narrative), and the contrast with the simultaneous opposition lawsuits is the canonical bimodal outcome distribution Cliff's approval-outcomes corpus should be capturing. Indiana also doesn't have an active state-level moratorium bill or constitutional ballot drive, which is exactly why hyperscalers are concentrating there — and which is why an Ohio ballot win in November would push even more pipeline into Indiana, raising the local-pushback risk there as well. For the de-rate calculator: the Indiana-versus-Ohio political-risk spread is now wide enough that it should be a discrete input for any developer underwriting MISO-side sites.