May 24.

Why it matters

For Cliff this kills the simple 'moratorium yes/no' flag and forces a richer data structure. Each parcel needs (a) which template the local jurisdiction is using or likely to copy, and (b) the carve-out thresholds (sq ft, MW, water vs air cooling, in-progress-project named exemptions). A Minneapolis-template moratorium does not kill a 300k sq ft downtown deal — but a developer relying on a binary 'no moratorium here' field would not know that. Three concrete to-dos: (1) Add a 'moratorium_template' enum to the jurisdictions table: {pre_emptive_total, sq_ft_carveout, mw_carveout, named_project_carveout, none}. (2) Capture the carve-out parameters: threshold_sqft, threshold_mw, exempted_zones (e.g. 'downtown_core'), exempted_projects (named-list). (3) For OKC specifically, log the named exemptions — Expand Energy Campus and the OKC Outlet Mall site — into the approval-outcomes corpus, because being named on an exemption list is the most valuable possible signal that a parcel is politically pre-cleared. The Cave City 'high-impact industries' wording is also worth a separate flag — it implies the same ordinance can be triggered against adjacent project types and is the cleanest signal that an anti-data-center council intends to catch re-brands.

Why it matters

Two takeaways for Cliff's siting work. First, the statewide-moratorium risk is real but currently ceilinged: even in a state that voted yes in the legislature, the executive branch's economic-development incentive killed it. That means the near-term blast radius stays local (city / county) rather than going state-level, which is good news for site readiness because local risk is much easier to map than state-level risk. Second, the 'veto + Advisory Council' template is the new compromise pattern to watch for in larger states (Virginia, Texas, Georgia, Ohio, Arizona). An Advisory Council typically takes 6-12 months to produce recommendations, then 6-12 more months for those recommendations to turn into rules — so the practical effect on a developer's near-term site map is small, but the medium-term effect is a new patchwork of state-level rules layered on top of the existing local patchwork. Track Maine's Advisory Council membership, charter, and first-meeting date as a leading indicator of what rules will appear in 2027.

Why it matters

This is the cleanest precedent of the year for what 'protective' data-center tariffs will look like in the rest of the country, and the three parameters — cost-allocation %, MW threshold, contract length — are now the variables every state's next docket will fight over. For Cliff's underwriting product this means: (1) Every parcel-level economics view needs a 'tariff exposure' field showing the local utility's data-center tariff status — pending, approved, or absent — with the WI 100% / 100 / 15 as the new presumptive worst-case. A developer comparing two parcels in different states needs to see the implied LCOE delta between a WI-style 100% tariff and a legacy-style 75% tariff. (2) The 100 MW threshold creates an obvious 'just under the line' siting strategy — split a 200 MW campus into two 95 MW phases — which is exactly the kind of structuring play that should be made visible automatically in the tool. (3) Track which states' next data-center tariff dockets cite Wisconsin as precedent: this is how you predict where the floor moves next. Likely next states: Indiana (AEP), Ohio (AEP / FirstEnergy), Iowa (MidAmerican / Alliant), Georgia (Georgia Power has a docket already pending), Michigan (DTE).

Why it matters

For Cliff this is the most product-relevant single bill of the week because the 25% on-site clean-generation mandate creates an entirely new parcel-selection requirement that doesn't currently exist anywhere else. It also gives an obvious adjacency path for the BTM-economics SKU: every NC site evaluation will need a co-located generation sizing calculation (peak demand × 0.25 → required nameplate solar/wind/storage capacity → required additional acreage → site-fit yes/no). The 40 MW threshold catches more projects than the 100 MW Wisconsin floor, so NC actually becomes the more rigorous filter even though Wisconsin gets the headlines. Two concrete to-dos: (1) Build the 'on-site clean offset' calculation into the parcel evaluator now, against NC's draft text — even pre-passage — so it is live the day the bill is signed. (2) Track which other state legislatures introduce 25%-clean mandates citing NC, because if NC passes and two more states copy in 2026, this becomes the new normal for state-level oversight rather than a single-state quirk. Most likely copiers given political composition + existing renewables build-out: Colorado, Minnesota, Washington, New Mexico, Oregon.

Why it matters

The Virginia July 1 deadline creates a hard product opportunity in the next five weeks: any developer with a Virginia site that has not yet filed its air-permit application has a structural reason to file BEFORE July 1 to lock in pre-Tier 4 BACT, and a structural reason to delay if their equipment selection isn't ready. Cliff's site-readiness view of every Virginia parcel in the diligence pipeline should show a countdown to July 1 and a flag for 'filed / not filed.' This is the cleanest possible application of the live-docket wedge — a specific permitting authority, a specific deadline, a specific change in standard. The federal EPA proposal is medium-term context: it will not be final this summer, but it confirms that the federal direction (DOE pushing FERC to accelerate large-load interconnection, EPA loosening preconstruction) is consistently pro-build, while state air agencies are tightening. Map this as a deliberate jurisdictional split in the underwriting tool — a Virginia site is now subject to two opposite-direction regulatory vectors at once, and the EPA layer doesn't help if the state DEQ is the binding agency on backup-diesel BACT (it usually is).

Why it matters

Nine weeks remain until July 24. This is the only live deadline in U.S. data-center site readiness where the structured filing that developers must produce (Form W Part A + modeling data including the per-study-year minimum firm-load limit) is also exactly the dataset Cliff's approval-outcomes corpus needs to capture. Two product implications: (1) Before June 1, the public ERCOT docket should be re-scraped to confirm the May 13 TAC vote text and any last-second amendments to the MW allocation / studied-load treatment language. Even one tweak there changes how a developer should size and time their Declaration of Intent. (2) The July 24 cutover is the natural cutoff for a 'Batch Zero filing assistant' surface: ingest a developer's project spec (location, peak MW, ramp curve, expected minimum firm load), produce a draft Form W Part A and modeling-data package, flag missing inputs. The same surface re-positions cleanly as the PCLR contract-negotiation aid post-July-24 for projects that won a slot. Pair with the PUCT financial-security rule (16 TAC §25.194, $50k/MW non-refundable) so the developer sees the ERCOT timing gate AND the PUCT cost gate in one underwriting view.

Why it matters

For Cliff this adds a buyer archetype to the segmentation model. The financial-sponsor + chipmaker JV is structurally different from a hyperscale or a colo in what it values out of a site-readiness platform: it needs permitting-risk diligence ABOVE everything else (because the debt won't fund without it), is willing to pay for risk-quantification, and is committing to parcels NOW. That maps cleanly to Cliff's core value prop in a way the traditional hyperscale-direct buyer doesn't (hyperscalers have in-house teams the size of consultancies and don't easily buy diligence software from outside). Two concrete to-dos: (1) Add 'Blackstone-Google JV' to the prospects sheet for outreach, route via Benjamin Treynor Sloss's known LinkedIn presence and Blackstone's infra team (Sean Klimczak / Eli Nagler are the public infra heads). The angle is parcel-level permitting-risk underwriting for the first 500 MW of 2027 capacity. (2) Watch for the next two announcements in this category — expect at least one more PE-shop + foundation-model-lab JV in Q3 2026 (the obvious shape: Apollo or KKR + Anthropic/OpenAI/xAI). Each one is a new buyer.