The question of who pays for AI’s power grid has been building for years. It arrived in Oregon on May 12.
The Oregon Public Utility Commission reportedly adopted Schedule 96, a dedicated billing class for large-scale data centers within the Portland General Electric service area. The framework is specific enough, a named regulatory instrument, a named regulator, a named utility service area, a named cost mechanism, that the underlying regulatory action is almost certainly real even before the primary OPUC order is confirmed. What remains to be verified is the precise applicability threshold, the effective date mechanics, and the exact language of the clean energy contract provisions.
This deep-dive answers a different question than the daily brief. Not what Schedule 96 is. What it means for operators who fall under it, what the financial impact model looks like, and whether the policy logic is exportable.
What Schedule 96 Actually Requires
The framework creates a dedicated customer class. That’s significant before the 85% floor even enters the analysis. Creating a separate billing class for AI data centers acknowledges that these facilities have a fundamentally different load profile from standard commercial and industrial customers, higher peak demand, more variable utilization, larger infrastructure footprint per dollar of consumption. Standard commercial tariffs weren’t designed for 100-megawatt-plus facilities with GPU clusters running inference at variable intensity.
The 85% minimum capacity payment is the mechanism that changes operator economics most directly. Under standard utility billing, a customer pays for what they consume. Under Schedule 96, a data center subscribing to a given capacity level pays for 85% of that subscription in every billing period. Usage below 85% of subscribed capacity doesn’t reduce the bill. The operator is paying for available capacity, not consumed energy.
This structure exists for a reason that’s defensible from the utility’s perspective. Grid infrastructure serving a 200-megawatt data center has to be built and maintained whether the facility runs at 60% utilization or 100%. The marginal cost of keeping that capacity ready is not zero during low-utilization periods. The 85% floor is a cost-recovery mechanism, not a penalty, it ensures the infrastructure investment gets funded across the billing cycle, not only during peak draw.
The Peak Growth Modifier is the second major mechanism. Allocating grid upgrade costs to the fastest-growing customer classes creates a dynamic pricing signal: as data center operators expand their subscribed capacity, they bear a larger share of the infrastructure investment required to serve that expansion. This is a direct departure from the default utility model, where infrastructure costs are spread across all ratepayers regardless of who drove the need for new capacity.
The Financial Impact Model
The 85% floor’s financial impact varies by workload profile. Consider two hypothetical operator scenarios, labeled generically to avoid inventing specific figures.
An operator running continuous, near-maximum inference workloads, training runs, sustained API serving, likely operates at or above the 85% threshold regularly. For this operator, Schedule 96 changes the billing structure but may not materially increase costs relative to prior consumption. The minimum floor is close to actual usage.
An operator running highly variable workloads, batch processing, development environments, inference capacity held in reserve for demand spikes, may operate well below 85% in many billing periods. For this operator, the 85% floor creates a meaningful fixed cost premium above actual consumption. A facility subscribed to 100 megawatts but averaging 60% utilization now pays for 85 megawatts in every billing period. That’s a 42% increase in energy costs relative to actual consumption during low-utilization periods.
The clean energy contract provisions add a third dimension. Data centers that enter special clean energy contracts under Schedule 96 reportedly gain the ability to directly support new clean energy infrastructure. The financial mechanics of these contracts, whether they represent a premium, a credit, or a cost offset, aren’t detailed in the available reporting. But the structure suggests Oregon is using the tariff framework to advance both cost-allocation and clean energy policy goals simultaneously.
The National Backdrop
Oregon’s Schedule 96 didn’t emerge in a policy vacuum. The federal conversation about AI infrastructure and grid capacity has been accelerating throughout 2026.
Congress held a subcommittee hearing on AI power demand, permitting reform, and ratepayer protection. The White House ratepayer protection framework included commitments from major AI operators, though the enforcement mechanism for those voluntary commitments remains unclear. Prior coverage here documented a 71% surge in planned natural gas capacity driven largely by data center demand, alongside the ESG implications of that surge for operators with clean energy commitments.
Schedule 96 is the first state utility commission response to that national pressure that takes the form of a dedicated billing class rather than a general rate case adjustment. Virginia, home to the largest concentration of data center capacity in North America, has had its own utility commission proceedings around data center load growth. Texas, Georgia, and Arizona are also high-concentration states where similar proceedings could emerge.
The policy logic is transferable. Any state utility commission facing the same dynamic, rapid data center load growth straining grid capacity, ratepayer cost allocation pressure, clean energy commitment requirements, has the same regulatory toolkit. Schedule 96 gives those commissions a named precedent to cite, a mechanism to model, and a political argument to make: we’re not penalizing data centers, we’re ensuring they fund the infrastructure their growth requires.
Who Falls Under Schedule 96, and Who’s Watching
The applicability threshold isn’t precisely defined in available reporting. “Large-scale data centers” and “high-demand users” are the operative terms. The specific megawatt or customer class threshold that determines which operators must file under Schedule 96 rather than existing commercial tariffs is a detail that requires the primary OPUC order to confirm.
For hyperscale operators, Amazon Web Services, Microsoft Azure, Google Cloud, Oracle Cloud, Meta’s infrastructure footprint, with existing or planned Oregon capacity, the threshold question is urgent. These operators already have legal, regulatory affairs, and energy procurement teams monitoring utility commission proceedings. The question is whether their Oregon capacity crosses whatever line the OPUC has defined.
For mid-market data center operators, regional colocation facilities, enterprise private cloud operators, the threshold matters more acutely, because their teams may not have the same regulatory monitoring capacity as the hyperscalers.
Verification Status and What Remains Unknown
Three elements of Schedule 96 require primary OPUC order confirmation before compliance decisions can be made. First, the precise applicability threshold, what megawatt level or customer classification triggers Schedule 96 rather than existing tariff schedules. Second, the effective date mechanics, whether the May 12 adoption date is the effective date for all provisions or whether there are phased implementation periods. Third, the clean energy contract terms, whether these are mandatory, optional, or available only to qualifying operators.
The Renewable Energy World reporting provides the framework outline. The OPUC docket is the authoritative source. Operators in the PGE service area should not rely on T3 coverage alone for compliance planning.
TJS Synthesis
Schedule 96 is a cost-allocation decision dressed in regulatory language. Oregon answered the “who pays” question by directing it back to the operators creating the demand, a structurally sound policy position that will be politically easier to defend than general rate increases.
The question isn’t whether this approach is correct. It’s whether it spreads. The policy logic is sound enough to survive export to other jurisdictions. The mechanism is specific enough to serve as a template. And the national pressure driving Oregon’s action, grid capacity stress, ratepayer protection politics, clean energy commitment expectations, is present in every state with significant data center concentration.
Watch the Virginia and Texas utility commission dockets through Q3 2026. If a second state utility commission adopts a dedicated AI data center billing class before year-end, Schedule 96 stops being an Oregon story and becomes the leading edge of a national regulatory pattern. That’s the signal that matters.