Cliff Daily — May 1.

Why it matters

This is the cleanest precedent yet for the firm-equivalent-mw-underwrite.md / interruptible-load-derate.md thesis: a state regulator just stamped 'curtailment commitment in exchange for tariff discount, transmission upgrade waived' as the standard structure for hyperscaler interconnection. Add to interruptible-load-derate.md as the first IURC-approved template alongside the ERCOT PCLR work. This is the exact regulatory primitive Cliff's de-rate calculator needs to price — what's the $/MW-yr discount for X% curtailment availability, and how does that compare to the $50K/MW interconnection-fee path in TX or the cost-shift path in VA. Open question for outreach: does this template extend to AEP / Indiana Michigan Power's pending Amazon and Meta filings (also IURC-jurisdictional)? Worth a 1-day read of the IURC docket.

Why it matters

Add to byog-byong-state-patchwork.md as a new utility-led cost-allocation template. The structure (hyperscaler pays 100% of interconnection + take-or-pay capacity + contributes generation) is the demand-side mirror of the VA cost-shift fight and the Texas SB 6 financial-commitment requirement — three different regulatory mechanisms converging on 'data center pays its own way.' OG&E's July 1 OCC filing is a live regulatory clock to add to the open-investigations list; the Electric Service Agreement and Capacity Purchase Agreement will be public docket material. This is also the first deal where a hyperscaler is *contributing generation capacity* as part of the tariff structure — that's a structural precedent BTM-product-surface.md should track.

Why it matters

75 MW is now the shared threshold across TX SB 6 (PUCT 25.194), ERCOT Batch Zero, and OK HB 2992. That's the canonical 'large load' definition for the regulatory-knowledge-graph wedge. The 60-day land-purchase + 5-mile public-meeting trigger is a *new* regulatory primitive — it imports a county-zoning-style transparency requirement into state law, and it's the first rule that fires *before* a developer even files for interconnection. This is the kind of pre-interconnection notification surface paces.md / landgate.md / transect.md don't currently track. Add a row to byog-byong-state-patchwork.md and flag for v1-strategy-audit.md as a new pre-IA regulatory clock to watch in OK.

Why it matters

Negative-result data point for the cost-allocation template wave: DTE is structurally tying a hyperscaler campus to ratepayer relief in the *opposite* direction from VA / OK / IN — instead of making the hyperscaler pay its own share, DTE is asking ratepayers to subsidize via rate increases now in exchange for *future* relief contingent on the data center actually energizing on time. This is the structure NRDC/Sierra Club already flagged in the November 2025 Michigan PSC contested case (referenced in ai-paperwork-automation-loop-log.md). Worth a new entry on byog-byong-state-patchwork.md showing the MI structure as a counter-example. Also relevant for hyperscaler-lease-vs-build.md — the contractual ability to *delay* energization in a take-or-pay world is itself a financial structuring question.

Why it matters

Direct update to longhorn-air-permit-case-study.md, air-permit-case-study-scope.md, air-consultant-outreach-plan.md. This is a hard regulatory clock with two consequences: (1) any VA data center that wants the cheaper Tier-2 / Tier-3 generator package needs to file before July 1 — that creates a 60-day rush window where the air-permit GTM is at peak urgency for VA developers; (2) presumptive BACT means the permit-engineering judgment call shifts from 'what control technology is BACT' to 'what's equivalent to Tier-4' — exactly the kind of unstructured-text-to-structured-data interpretive work the Cliff air-permit wedge is positioned to do. Pair with rice-neshap-dr-carveout.md: VA's Tier-4 stance combined with the federal RICE NESHAP 100-hour limit is a stack of constraints that rules out most cheap diesel BTM. The case for natural gas / fuel cell / battery primary backup just got stronger in VA.

Why it matters

Refines yesterday's $650B number to $725B and adds the Amazon-specific signal that's most underwriting-relevant: AWS growth re-accelerated to a 15-quarter high in the *same* quarter that capex grew 77% YoY. That's the demand-side validation for the supply-side bottleneck — capacity is selling through faster than it's coming online, which is what the de-rate-calculator-gtm.md / firm-equivalent-mw-underwrite.md narrative depends on. Apple's structural absence from the cloud-capex race is the other side: ~1 of 5 incumbent 'hyperscalers' isn't bidding for grid power, which makes the demand concentration on the other four even more important. Update the macro paragraph in de-rate-calculator-gtm.md from $650B → $725B.