June 6.

Why it matters

Cliff has been tracking the 'regulated tariff' template — Oregon, Pennsylvania, Wisconsin, Florida — as the converging answer to 'who pays for data center growth.' Minnesota just put a different template on the board, and the two now compete inside every state PUC. The deciding variable is: does the state have one giant hyperscaler showing up at a single campus, or a dozen smaller projects competing for the same grid headroom? Single giant → bilateral wins (Minnesota). Many projects → tariff wins (Oregon, PA, WI). For Cliff's underwriting product, this means the 'tariff exposure' field needs a second axis: 'bilateral feasibility' — is there a hyperscaler in this state already executing a Minnesota-style ESA, and is the developer in a position to negotiate a comparable deal? The 1,900 MW CEAC piece is the bigger strategic clue. It's the first time a hyperscaler has functionally taken over a utility's clean-energy procurement function as a condition of being served. That maps onto the BTM-generation-economics layer Cliff has been queueing — developers who can pre-fund their own clean generation now have a negotiating instrument that didn't exist 18 months ago.

Why it matters

Three load-bearing implications. (1) In every Dillon's-rule state — Texas, Virginia, North Carolina, Ohio (with carve-outs), Kentucky, Tennessee, Alabama, Georgia — a data center moratorium is legally fragile and a well-resourced developer can break it in weeks. That changes the underwriting risk profile of those states: 'local moratorium risk' is real for project timing but not for ultimate project viability. (2) The 'disclosure checklist' replacement is the path of least political resistance for counties that want to *look* responsive without inviting a lawsuit, and the checklist itself becomes the document a developer needs to prepare on day one. Water-use disclosure, in particular, is the variable least covered by existing pre-development tools — which is a natural Cliff feature. (3) The Hill County checklist plus PA's model tariff plus Oregon's emissions-linked interconnection are now three different ways the same underlying question — 'what disclosures does a developer have to make before a project can proceed' — is getting structured. Cliff's regulatory-knowledge-graph play has to treat these as one schema with three sub-schemas, not three separate workflows.

Why it matters

The local-moratorium map is now diverging from the state-tariff map. State tariffs are converging (Oregon ≈ Pennsylvania ≈ Wisconsin ≈ Florida on the structural template). Local moratoriums are diverging — different durations, different thresholds, different replacement ordinances. For Cliff, that means the state-level layer of the regulatory knowledge graph can be modeled in 5-10 canonical templates, but the county/municipal layer has to be modeled as a per-jurisdiction tracker. The Pulaski County 'threshold-based' moratorium is also the first time we've seen the regulatory framing acknowledge a difference between AI compute and crypto/blockchain. That distinction is going to matter for siting decisions — once enough counties carve out 'AI' specifically (with friendlier rules than crypto), large hyperscaler projects will route toward those jurisdictions and the moratorium map becomes a routing tool, not just a risk tool.

Why it matters

Two product-level implications. (1) The July 31 effective date in PJM lines up almost exactly with the ERCOT July 24 Declaration of Intent deadline and the (likely) ERCOT July 10 Batch Zero effective date. So July 2026 is the month where both PJM and ERCOT — the two largest data-center grids in the country — formally codify their large-load interconnection rules. A developer with a PJM project and a Texas project will have to file structured forms in both markets within a two-week window. That's exactly the cross-jurisdictional workflow Cliff's filing-drafting wedge is designed for. (2) The convergence of PCLR and Non-Firm Contract Demand into the same structural product — 'cheaper grid service in exchange for curtailment when transmission is tight' — turns the 'interruptible load economics' analysis Cliff has been building into a portable tool. The math is the same in both markets; only the curtailment-frequency inputs change. That's the first piece of Cliff's product surface where one analytical engine covers PJM and ERCOT without a market-specific rewrite.

Why it matters

Two follow-throughs for the Cliff Batch Zero filing-assistant product. (1) Procedural risk has dropped enough that the 7-week developer window (now ~7 weeks until July 24) is the binding constraint on go/no-go decisions. The marketing pitch is no longer 'might want to file' — it's 'must file by July 24 or wait years.' (2) The 100 GW number is the right denominator for a TAM calculation. If even 1% of those projects use a third-party filing assistant at a few thousand dollars per filing, that's a real-money initial wedge — and it's the right size for a single-quarter sprint to validate before extending to PJM (where July 31 starts the same cycle). The Form W Part A structure plus the minimum-firm-load disclosure are the two pieces of the filing that should be productized first; the rest is paperwork.