OK / HB 2992 / 5th state ratepayer protection / 75 MW threshold

Why it matters

Three updates. (1) The cluster is now 5 states + 1 regional ISO in one ~30-day window: Pennsylvania (PUC model tariff May 8), Florida (SB 484 signed May 7), Wisconsin (We Energies VLC tariff PSC-approved April 24), Oregon (PGE Schedule 96 effective June 10), Oklahoma (HB 2992 to governor May 5), and PJM at the regional level (NCBL framework paper May 6, FERC filing targeted in time for the 2028/29 Base Residual Auction in June 2026). This is no longer 'a handful of states reacting' — this is convergent regulatory drift across PUCs and legislatures at sufficient velocity that any data-center site evaluated for a 2027+ in-service date should now be underwritten assuming a ratepayer-protection tariff exists or will exist in the relevant jurisdiction. The competitive-landscape page should add an 'in-state ratepayer-protection tariff' field to every site row with values {none, statutory, regulator-issued, in-flight}. The five May states should be marked 'statutory' (OK, FL), 'regulator-issued' (PA, WI, OR), and the rest should be marked 'in-flight' until proven otherwise. (2) The 75 MW threshold is the second data point above the 50 MW modal convergence. The regulatory community is converging on a 50-75 MW band as 'large load = systemically significant = different rules.' For site-readiness underwriting, this means any campus that buildouts past 50 MW (basically every hyperscale AI site) crosses into the systemically-significant zone in nearly every jurisdiction. The 'sub-threshold colocation' workaround — building two 49 MW facilities on adjacent parcels to stay under the threshold — is closing because the PA, PPL, and We Energies definitions all include aggregation rules within a geographic radius. Oklahoma's HB 2992 does NOT include an explicit aggregation rule (verify on the final bill text once signed), which is a structural loophole an aggressive developer could exploit if Stitt signs as-is. The de-rate calculator should add an 'aggregation-rule presence' boolean to each state's tariff representation, and Oklahoma's value should be 'no' as a working assumption (verify on signature). (3) The 60-day land-acquisition notice provision is novel and structurally important. Most ratepayer-protection regimes operate at the tariff layer (rates and infrastructure cost recovery), not the land-use layer. By requiring notice to adjoining landowners and county commissioners BEFORE the developer closes on the land, Oklahoma is creating a pre-zoning political-trigger mechanism that effectively gives local opposition a 60-day organizing window for every hyperscale parcel. This is the first state to bridge tariff-layer protection and land-use-layer transparency in a single bill, and it's a template other states will copy. For Cliff's regulatory knowledge graph, this creates a new field type ('pre-zoning notice trigger') that needs to be tracked separately from the existing 'tariff threshold' and 'aggregation rule' fields. Investor-update translation (for investor-updates/2026-05-13.txt): Oklahoma becomes the 5th state in ~30 days to formally tell hyperscale data-center developers 'you pay your own infrastructure costs, and you pre-disclose your land purchases.' The 5-state pattern in a 30-day window is the kind of regulatory-velocity signal Cliff is built to capture — the per-state tariff structures are different enough that a hyperscaler operating across 5 states needs 5 different tariff lookups, 5 different threshold rules, 5 different notice timelines. That heterogeneity is the moat: the cost of an LLM-powered cross-state tariff lookup approaches zero, but the cost of a human-consultant doing the same thing scales linearly with state count.