Cliff Daily — May 3.

Why it matters

This is the cleanest data point so far that the *threat* of a local moratorium has direct underwriting impact, separate from whether one passes. Sabey is not a marginal player — they're a long-tenured Tacoma/Tukwila operator with an existing campus, and they walked from a 68 MW expansion at a site they already own and operate, before any ordinance touched them. Add to byog-byong-state-patchwork.md as the canonical 'moratorium-threat-as-effective-veto' case. For the regulatory-knowledge-graph: this means moratorium-status modeling needs three states, not two — passed, pending, and *announced-but-not-introduced* — because the third already has a measurable price effect. Useful for de-rate-calculator-gtm.md framing too: the right denominator for site-level regulatory risk is not just enacted ordinances but the rumor-ordinance pipeline that's been growing for ~6 months in jurisdictions with rising data center concentration. Also note the asymmetry: Sabey withdrew but Equinix and Prologis stayed — likely a function of how far along their interconnection studies are and how exposed each is to the specific water / land-use concerns. That's a heterogeneous response Cliff's air-permit / interconnection / siting integration is positioned to map and price.

Why it matters

Three nested votes (county, city P&Z, city council) on the same physical site within an 8-day window is a clean test case for the regulatory-knowledge-graph: outcomes can interact (county moratorium passes but P&Z denial moots it; or P&Z approves but city moratorium kicks in 7 days later and freezes the project). For southeastern Georgia specifically, Camden / Kingsland is the closest deepwater-port jurisdiction to Jacksonville's data center cluster and the proposed Nassau County FL development; if both Camden GA and Nassau FL freeze in the same month, the FL/GA coastal corridor effectively closes for new entitlement applications until Q3 2026. Useful to track because it pressure-tests the 'local-vs-state pattern' framing in byog-byong-state-patchwork.md — Georgia's HB 1059 (state-level moratorium) hasn't moved, so the entire enforcement is happening at the county/city tier, exactly the post-Maine pattern.

Why it matters

Sedgwick County is the largest Kansas county by population and the only one with multiple active data center site purchases in flight; the freeze pattern (Jan freeze → Mar extension → likely May extension) is exactly how a 'temporary' 90-day pause turns into a de-facto 12-month entitlement halt without ever passing a permanent ban. Worth modeling in the regulatory-knowledge-graph as a separate 'rolling moratorium' status — distinct from a one-shot 12-month freeze because the political cost of *not* re-extending is asymmetric (rolling extensions get less press than the original ban). The Kansas state senate's separate 3-year pause request is also worth noting as an upside-risk on Kansas exposure. For the developer-financing capital stack: a rolling 90-day freeze is the worst possible state for site readiness underwriting because the option value of waiting is high but the carrying cost of land is also high — exactly the spread Cliff's de-rate calculator should be able to surface.

Why it matters

Denver is a small data center market, but it's the first major U.S. metro to consider a city-wide moratorium that blocks all new entitlement processing rather than just zoning approval. The procedural nuance matters for Cliff's submission-validator product: a moratorium that blocks 'acceptance, processing or approval' freezes filings at the intake step, which means the data center developer can't even compile a complete record while the moratorium is in effect. That's a different surface than an after-the-fact zoning rejection (which leaves the record buildable) and it changes the site readiness timeline math materially. Pair with Denver's separate effort to reduce city-wide water use 20%; the data center water/cooling angle is now a primary moratorium driver in arid-climate metros and feeds directly into the air-permit case study scope (cooling-tower drift, water-temperature effluent in some jurisdictions ladders into Title V applicability).

Why it matters

Update on yesterday's lead. Two new layers worth pulling out: (1) the budget hostage taking by the same legislators on data center sales-tax exemption is a tell that the cost-shift fight is not just about HB 1393 / SB 253 in isolation — it's about the entire VA fiscal treatment of data centers, and Spanberger's posture on the bills will read as a preview of her budget posture. (2) The 19-day window means the next news cycle for VA cost-shift is the Spanberger announcement itself. For virginia-data-center-legislation-2026.md: add a countdown to May 22 and a contingent branch for each of the three outcomes, with the price-of-load delta for each. The Dominion 51 GW contracted figure (Q1 lead) is what monetizes — at ~$50/MW-month delta on a 15% cost-shift, that's ~$300M/year in reallocated cost across the contracted book, dropping to whatever fraction of new contracts get touched depending on effective date. That's the input Cliff's de-rate calculator should be modeling for Virginia exposure.

Why it matters

This is the cleanest fully-off-grid hyperscale template announced to date and a direct input to BTM-product-surface.md and off-grid-feasibility-by-uptime.html. Three structural details to copy into the wiki: (1) CCHP captures waste heat to drive cooling — this is the architectural choice that makes a gas-fired data center pencil environmentally and ratepayer-wise (no liquid water consumption for cooling at scale, less effluent, smaller air permit envelope per MW), (2) grid-forming battery at 1.1 GWh provides the inertia and ride-through that gas gensets alone can't, replacing the synchronous-condenser role utilities play in grid-tied data centers, (3) the Cat dealer (Wheeler) is the actual O&M counterparty — that's a different supply chain than utility T&D, and it materially compresses the site readiness timeline because Cat fleet delivery slots are the binding constraint, not interconnection studies. The Utah siting choice is also worth tracking: Millard County is rural, low-population, low-water-rights pressure, and adjacent to the Intermountain Power Project transmission corridor (a hedge if the off-grid posture ever needs to grid-tie). For Cliff's wedge: this is the structural answer to the local-moratorium wave (sections above), and it implies the BTM build-vs-buy economics calculator needs to model 'fully off-grid' as a discrete deployment mode, not just 'BTM-with-grid-export-credit.' The Cat fleet model also means the de-rate calculator's air-permit module needs to handle multi-engine RICE NESHAP at scale (4 GW of G3520K is roughly 1,800+ engines) — that's where the air-permit case study scope hits next.

Why it matters

This is currently the largest tail risk to U.S. data center site readiness that Cliff's wiki doesn't have its own page for. Per the May 2 ConstructConnect data, Ohio + Illinois + the rust-belt corridor saw $22B in Q1 2026 data center starts — the largest regional concentration in the country, much of it driven by AEP's announced 10 GW data center campus 765 kV buildout. A constitutional ban on >25 MW facilities would functionally cancel essentially all of that future pipeline if it qualifies for the ballot AND passes. Signature thresholds in Ohio for constitutional amendments are notoriously hard (last successful one was Issue 1 abortion, 2023, ~700K collected); 413K in 12 weeks against organized hyperscaler money is a steep climb but not zero. Worth opening a new strategy/ohio-data-center-ballot-2026.md page tracking: signature progress (deliverable July 1), polling once on ballot, hyperscaler counter-spend, and the legal question of whether already-permitted facilities are grandfathered (the Hunterbrook reporting on Meta's 'secretly-behind' $1B Ohio facility is the lead worst-case). For the regulatory-knowledge-graph: ballot-initiative bans are a different regulatory class than legislative bans — the political-economy risk is concentrated in 6 weeks of signature gathering + 4 weeks of campaigning, not in legislative log-rolling. Ohio is a leading indicator for whether the ballot-initiative channel becomes the next vector after county-level moratoriums.

Why it matters

The ~$620–630B midpoint is the demand-side denominator that the supply-side stories (Dominion 51 GW contracted, Saline $16B closing, Joule 4 GW Utah, AEP's 765 kV Ohio buildout) ladder up against. Two new things vs yesterday's lead: (1) memory/component cost inflation is now an explicit budget driver, which means the per-MW cost of *equipping* a data center is rising fast enough that hyperscalers are choosing to lock in compute via leasing (yesterday's OpenAI/Stargate retreat to bilateral leasing) rather than build-and-own — that durably reinforces the site diligence-platform-for-developers positioning. (2) The Meta Indiana 1 GW announcement adds another rust-belt site to the Ohio/Illinois cluster Hunterbrook flagged as the secretly-Meta $1B Ohio facility — i.e., Meta is concentrating heavily in MISO/PJM-Central, and that's exactly where the Ohio ballot-initiative ban would hit. The capex denominator is durable; the question is which of the 8 functional layers in strategy/competitive-landscape-by-layer.md gets share of that denominator. Cliff's wedge is that ~5–7% of capex (the site diligence fraction) is currently going to hourly consultants with no productized ingest of regulatory text — that's the $30–45B/year addressable layer the regulatory knowledge graph is positioned against.