Pricing AI Data‑Center Load: GS‑5 Minimums, Riders, and Queue Risk

Opening Insight

AI/data‑center demand has moved from anomaly to default. That shift resets how price formation, siting, hedging, and operations work. Utilities are codifying recovery in large‑load classes and minimums (for example, GS‑5 14‑year terms with 85% T&D and 60% generation minimums), while interconnection queues, milestone deposits, equipment lead times, and rider volatility push fixed‑cost and availability risk onto campuses and portfolios. Add water stress and cooling choices, and engineering constraints become P&L variables that must be priced, hedged, and operationalized.

This post defines those exposures and the operating model to manage them. We quantify the costs of inaction—fixed‑charge lock‑ins, queue deposits and delays, curtailment penalties, rider whiplash, credit concentration, and systems drift—and then show the operational and financial gains from encoding minimums and take‑or‑pay into curves and ETRM, integrating event‑driven queue/tariff data, optimizing PUE/WUE, and scheduling around congestion and curtailment windows.

We lay out a market‑to‑operations control plane—rules‑as‑software, agentic monitoring, counterparty intelligence, and a sidecar integration with legacy ETRM—plus a sequenced roadmap and leadership accountabilities that deliver T+0 re‑pricing, tighter hedge attribution, and lower settlement variance. For the grounding facts and immediate implications, continue to Context and Analysis.

Costs of Inaction

Ignoring the data‑center‑driven shift turns manageable exposures into fixed costs, missed schedules, and avoidable P&L volatility.

• Fixed‑charge lock‑in under Virginia’s GS‑5: >25 MW campuses face 14‑year terms with minimum monthly demand at 85% of contracted T&D and 60% of generation, so under‑utilization still bills and distorts campus‑level P&L.
• Schedule slip and cash drain from queues: a 230‑kV transformer arrived 22 months late while minimums accrued; in Ohio, AEP added $10,000–$100,000 interconnection study deposits and milestone payments, pushing in‑service dates and tying up cash.
• Curtailment hits operations and penalties: when curtailment protocols meet take‑or‑pay and minimum‑demand clauses, ramps miss, nominations break, and balancing costs rise.
• Rider and tariff whiplash: ERCOT TCRF/DCRF resets and PJM transmission formula‑rate updates move non‑bypassable charges, shifting delivered cost and basis even without a spot‑price shock.
• Credit and concentration exposure: co‑ops projecting data centers near 95% of sales and fragile neoclouds amplify default and stranded‑asset risk on upgrades and leases.
• Water and permit tightness: evaporative‑hour caps can force a pivot to dry cooling, as in Phoenix, adding about 600 kW; with local water around $3–$6 (Loudoun/Fairfax) and $5–$8 (Phoenix) per 1,000 gallons, the choice moves both utility bills and compliance burden.
• Controls and systems drift: static

ETRM/reference data that ignore new large‑load thresholds (>25 MW) , ratchets, and minimum payments drive latency, errors, and settlement disputes.

• Competitive slippage: peers that lock long‑dated PPAs, storage hedges, or geothermal/cooling options secure advantages you won’t match later.

Operational and Financial Gains

• Event‑driven data into ETRM collapses decision latency to hours, not weeks, as queue milestones and GS‑5 thresholds (>25 MW; 14‑year, 85%/60% minimums) update positions and budgets—cutting surprise riders and mis‑timed hedges.
• Optimizing PUE/WUE—shifting to sub‑0.2 L/kWh cooling where water binds—and hedging to tariff structure ($/kW, $/kWh, riders) reduces delivered cost and permit complexity while improving peak‑cooling control.
• Scheduling builds resilience by planning around congestion and curtailment windows and, where feasible, siting near stronger backbones—reducing forced ramps and balancing risk that disrupt throughput.
• Pricing GS‑5 minimums (85%/60%), take‑or‑pay, and riders into curves tightens hedge attribution and stabilizes collateral—shrinking the “millions per month” P&L swings that under‑modeled minimums can trigger.
• Treating interconnection fees and AEP‑style study deposits ($10,000–$100,000) as screened pipeline risks improves forecast accuracy and capital discipline, focusing credit on executable MWs.
• Settlements variance falls as dynamic rate structures, pass‑throughs, and take‑or‑pay provisions are encoded in ETRM, with cleaner handoffs across front, middle, and back office via event‑driven workflows.

Market-to-Operations Control Plane

The point is to fuse market, tariff, and operational signals so teams act faster with tighter risk and cleaner P&L.

• Data and architecture: an event‑driven backbone that streams ISO/RTO queue milestones, tariff updates (including GS‑5 terms and dates), rate cases, water‑stress indices, and interconnection capacity into ETRM and planning with lineage.
• Forecasting and optimization: ML‑driven forecasts of locational price and basis impacts from build‑outs; explicit modeling of take‑or‑pay and minimum‑demand clauses; hedge optimization across power, gas, storage, geothermal, and Cold UTES.
• Agentic automation: agents watch dockets, permits, and campus announcements, auto‑triggering playbooks for credit review, PPA repricing, and logistics or scheduling adjustments.
• Rules‑as‑software : encode large‑load thresholds (>25 MW), GS‑5‑style minimums (85% T&D/60% generation), curtailment protocols, and ratchets directly into risk and settlement controls.
• Counterparty intelligence : continuously score counterparties on revenue concentration, lease obligations, and contract tenors; align collateral and covenants to regime‑shift exposure and default risk.

Expect faster, more accurate decisions, lower cost to serve, lower variance in settlements, and stronger campus‑level P&L as queue, tariff, and rider signals drive coordinated action across trading, risk, and operations.

Arcelian Architecture and Roadmap

Arcelian turns

Turning Grid and Tariffs into Executable Controls

This approach builds a market-to-operations control plane that pipes tariffs, interconnection milestones, water and cooling signals, and counterparty risk into ETRM, credit, and settlement. The goal is to keep trading, risk, and operations working from the same facts before costs hit P&L, pairing modern data architecture with rules and operating-model changes that reduce leakage and raise hedge precision.

Architecture: Control Plane and ETRM Integration

Shared fact base linking siting, hedging, scheduling, and settlements

A market-to-operations control plane unifies the data that drives siting, hedging, scheduling, and settlements so physical and financial moves stay aligned in the ETRM and downstream processes.

Event-driven backbone for ISO/RTO queues, tariffs, and rate cases

An event-streaming backbone ingests ISO/RTO queue milestones, tariff and rate updates (for example, Dominions GS-5 terms and dates), rate-case filings, water-stress indices, and interconnection capacity changes with full lineage into ETRM and planning systems.

Rules-as-software embedded in risk and settlements

Codify thresholds and commercial minimums as executable rules: >25 MW trigger; GS-5 14-year contracts with 85% T&D and 60% generation minimums effective Jan 1, 2027 ; plus curtailment and take-or-pay logic wired directly into risk, credit, and settlement engines.

Contract and rider models that reflect true delivered cost

Normalize demand/capacity and rider impactsincluding TCRF/DCRF and TVA ratchetsso delivered-cost curves and settlements reflect actual bill drivers instead of averages.

Forecasting and optimization across power, gas, storage, and alternatives

Project locational effects from grid build-outs; price queue costs such as AEP Ohio interconnection study deposits of $10,000$100,000 ; and optimize hedge mixes spanning power, gas, storage, and options like geothermal and Cold UTES.

Agentic monitoring of dockets, permits, and campus announcements

Autonomous monitors watch regulatory dockets, permits, and campus news, pushing curtailment and rate-case alerts to positions, nominations, and dispatch decisions within hours.

Counterparty intelligence for neoclouds and developers

Score exposure across counterpartiesrevenue concentration, lease obligations, and tenorsand align collateral and covenants to improve credit and collateral outcomes.

Cooling and water signals tied to $/kWh and $/kgal exposure

Link evaporative vs. dry/liquid-immersion cooling choices to energy and water costs; capture permit caps and operational pivots such as the ~600 kW Phoenix shift to keep settlements and delivered-cost curves accurate.

Roadmap: Near-Term Sequence and Trade-offs

• 1) Baseline exposures and ETRM reference data  load shapes, GS-5 thresholds, riders (TCRF/DCRF), queue positions, and water/permit limits; map leakage in settlements.
• 2) Stand up the event-driven backbone  integrate PJM/MISO/ERCOT queue feeds, tariff and rate-case updates, and water-stress indices into ETRM and planning with lineage.
• 3) Encode rules and minimums  implement >25 MW triggers, GS-5 85%/60% minimums and 14-year terms effective Jan 1, 2027 , plus take-or-pay and curtailment logic; target lower variance in settlements.
• 4) Deploy agents and counterparty analytics  monitor filings and campus news; trigger credit reviews and collateral changes for concentration risk among neoclouds; aim for improved credit outcomes.
• 5) Integrate queue milestones into siting and 

Hedging: Probability‑Weighted Upgrades and Scheduling Alerts

Treat upgrades as priced risks with probability‑weighted in‑service dates; decide interim mobiles versus full GIS builds during gaps like the 22‑month transformer delay ; wire alerts to scheduling playbooks.

Pilot Forecasting and Cooling Optimization

Run price and cost forecasts; test evaporative‑to‑dry pivots against local water tariffs (for example, Loudoun/Fairfax or Phoenix) to cut leakage and raise hedge precision.

Extend Governance for New Load Classes

Formalize model governance for new load classes and long take‑or‑pay provisions; add campus‑level curtailment alerting that reaches positions/nominations within hours.

Human & Organizational Changes

• CFO — chair a standing CFO/CIO/CRO forum that steers capital and hedges using scenario bands (for example, 6.7%–12% U.S. electricity share by 2028 ); ensure take‑or‑pay and queue fees are budgeted and monitored.
• CIO — deliver the event‑driven backbone, ETRM lineage, and rules‑as‑software; own model deployment and data quality.
• COO/Operations — translate curtailment protocols, queue milestones, and cooling choices into site playbooks; choose interim mobiles versus GIS builds and coordinate with schedulers so alerts flow within hours.
• CRO/Head of Trading — embed tariff thresholds, GS‑5 minimums, riders, and queue costs into pricing, limits, and collateral; refresh model governance for new load classes and long‑dated take‑or‑pay.
• Commercial and risk teams — retrain to price water and permitting constraints with energy basis; adopt a shared cadence so updates land in positions and nominations quickly, driving reduced leakage and faster hedge precision.

Unify Market and Operations

The core reality is that data‑center demand is now the standing load that sets terms for pricing, hedging, and operations. It shows up in faster T&D buildouts, long equipment and interconnection queues, and bills that carry large‑load classes, contract minimums/ratchets, and riders that move non‑bypassable charges. Water adds a second constraint, forcing cooling choices that shift both kW and kgal exposure and tie siting to local permits and rates.

For trading desks, this rewrites basis and capacity dynamics and makes interconnection timing a P&L variable; for risk, fixed take‑or‑pay and counterparty concentration harden downside; for operations, delays and curtailment windows demand tighter scheduling and contingency paths. Leadership’s job is to keep CFO, CIO, CRO, and trading aligned on scenarios, tariff changes (for example, GS‑5 85%/60% minimums), and queue milestones. Build a market‑to‑operations control plane so trading, risk, and operations act from the same facts, fast.

Operationalize the Control Plane

Arcelian operationalizes the market‑to‑operations control plane so trading, risk, and operations act on

the same facts. We implement the data, tariff, and workflow backbone that lets you price, hedge, and settle against baseline data‑center load with control.

• Trading and procurement intelligence that integrates ISO/RTO queues and rate cases to anticipate basis shifts, curtailment windows, and grid‑upgrade timelines.
• Contract and tariff engineering that encodes GS‑5‑style 85%/60% minimums , take‑or‑pay, and interconnection/queue deposits into ETRM and credit.
• Architecture and workflow redesign using event‑driven integration, rules‑as‑software, and agentic monitoring to cut latency across trading, risk, scheduling, and finance.
• Water and community risk mapping that ties site water‑stress and permit constraints to cooling choices and compliance exposure.
• Counterparty and credit analytics that quantify neocloud concentration and lease obligations to align collateral and off‑ramps.

Book the 45‑minute working session at calendly.com/arcelian/45min-working-session or email contact@arcelian.com to map exposures now and outline a 90‑day modernization sprint.

Integrating Agentic AI with Legacy ETRM: A Control‑Plane Modernization Strategy

For organizations facing AI/data‑center load reshaping tariffs and interconnection risk, the priority is not an ETRM rewrite but a control‑plane that binds market events to commercial obligations.

Practically, this means an event‑driven layer that ingests ISO/RTO queue milestones, tariff/rider changes (e.g., Dominion GS‑5 85%/60% minimums), curtailment protocols, and water/cooling constraints; normalizes them into canonical events; evaluates rules‑as‑software ; and pushes priced outcomes into deal capture, credit, and settlements.

Agentic monitors continuously parse tariff updates, compare them to portfolio exposures, and open workflow tasks or auto‑reprice take‑or‑pay and minimums under strict approvals and audit.

This is an ETRM architecture extension by integration—not replacement—aligned to a modernization strategy that reduces decision latency and prevents settlement leakage.

• (1) Embedded ETRM customizations
• (2) A sidecar control‑plane with adapters and a streaming bus
• (3) A data hub that feeds analytics first

For most, the sidecar pattern is the pragmatic integration roadmap: it isolates change, supports low‑latency write‑backs, and enables versioned rules.

Selection criteria should include:

• Round‑trip latency from market event to re‑price
• Explainability and auditability of rules
• Lineage from tariff source to settlement line item
• Ability to attribute hedge effectiveness
• Operational resilience under curtailment events

Sequence delivery to reduce risk:

• Phase 0: Map obligations to data sources
• Phase 1: Ingest ISO/RTO queues and tariff riders
• Phase 2: Implement rules‑as‑software with CI/CT and canary releases
• Phase 3: Enable selective write‑backs to ETRM, credit, and invoicing
• Phase 4: Introduce agentic automation for exception handling

This operationalizes the post’s thesis

that surging AI/datacenter demand must be wired directly into trading, risk, and operations.

Measurable outcomes:

• T+0 repricing in <5 minutes for tariff/rider changes
• 95% hedge attribution explain
• 30% reduction in settlement adjustments
• Credit exposure refresh aligned to event timestamps

Risk controls:

• Humanintheloop approvals for material price impacts
• Immutable rule/version logs
• Model monitoring for tariff parsers
• Fallback playbooks for ETRM writeback failures

Frequently Asked Questions

What is Dominions GS5 largeload tariff and how should we price its minimums?

For campuses above 25 MW, GS5 sets 14year contracts with minimum monthly demand equal to 85% of contracted transmission and distribution and 60% of generation starting Jan 1, 2027. Those minimums bill most fixed charges even when a site is underutilized, distorting sitelevel P&L if ignored. Price the minimums and takeorpay into forward curves, encode the thresholds and ratchets in ETRM/settlements, and align hedges and collateral so basis and availability risk are reflected before costs hit the invoice.

How do we upgrade a legacy ETRM to handle concentrated AI load without replacing it?

Stand up a sidecar controlplane: an eventdriven layer that streams ISO/RTO queue milestones, tariff and rider updates (e.g., GS5), curtailment protocols, and water/cooling signals into a shared fact base. Evaluate rulesassoftware to reprice deals and minimums, and use agentic monitors to watch dockets and trigger credit, hedging, and scheduling playbooks with auditability. Sequence delivery from ingesting queues/riders to rules with versioning and selective writebacks; target T+0 repricing in under 5 minutes, tighter hedge attribution, and fewer settlement adjustments.

Which queue and rider changes most affect delivered cost and basis right now?

Interconnection study deposits and milestone fees (e.g., AEP Ohios $10,000$100,000), plus long equipment delays like a 230kV transformer arriving 22 months late, push inservice dates and tie up cash. ERCOT TCRF/DCRF resets and PJM transmission formularate updates shift nonbypassable charges, moving delivered cost and basis even without a spotprice shock. Treat these as priced pipeline risks in forecasts and ETRM, wire alerts to scheduling to manage curtailment windows, and update credit and collateral as dates and riders change.

Trend Watch

AI data center power demand is accelerating faster than regulatory cadence, locking in fixed costs and redefining siting math. As data center electricity consumption deepens to baseline, utilities are hardening grid upgrades and tariffs with instruments like Dominion GS5 85% T&D 60% generation minimums and rider volatility (ERCOT

TCRF/DCRF, PJM transmission formula rate). For portfolios, that means minimum demand charges (take‑or‑pay) become a first‑order driver of delivered cost and hedge attribution, not a footnote.

This is where ETRM modernization must meet operations. A market‑to‑operations control plane with agentic automation should continuously ingest ISO/RTO interconnection queues, tariff/rider filings, and water stress indices, then re‑price basis and curtailment risk within minutes.

Interconnection queue delays and AEP Ohio interconnection study deposits tie up capital and shift in‑service probabilities; agents should push those changes into curves, collateral, and nominations the same day to prevent settlement leakage and collateral shocks.

Cooling choices now move P&L as much as power. Track PUE alongside data center water usage (WUE) and permit caps; model pivots to dry cooling, geothermal, or Cold UTES as part of tariff strategy. In water‑stressed nodes, the right WUE/PUE mix can be worth more than a penny in LMP by avoiding curtailment windows and ratchets.

Commercial edge:

• wire minimums and curtailment protocols into deal logic.
• stress‑test basis against phased queue outcomes.
• monitor neocloud counterparties for credit concentration risk.

The firms that integrate AI with legacy ETRM via a sidecar control‑plane will price faster, hedge smarter, and site where constraints are monetizable—not fatal.

Closing Insight

Data‑center load is no longer a variable to be negotiated quarterly; it is the standing condition shaping price formation, siting, and control. The edge goes to firms that translate GS‑5‑style 85%/60% minimums, rider volatility (TCRF/DCRF, PJM formula rates), and interconnection‑queue slippage into rules‑as‑software in a sidecar, event‑driven control plane—enabling T+0 re‑pricing , disciplined collateral, and fewer settlement surprises. Treat cooling and water as financial levers: align WUE/PUE and curtailment protocols with take‑or‑pay, and let agents propagate those constraints into curves, nominations, and credit to contain basis and availability risk while watching neocloud concentrations. The organizations that unify trading, risk, and operations on this market‑to‑operations spine will lower delivered cost, harden resilience, and modernize without ripping out ETRM—turning AI’s structural demand into a durable advantage rather than a source of volatility.

Partner with Arcelian

Market conditions are hardening fast: GS‑5‑style 85%/60% minimums, ERCOT/PJM rider volatility, and interconnection delays are turning data‑center load into fixed obligations that rewrite delivered cost, credit, and operations. Arcelian partners with CFO/CIO/CRO and trading to stand up a sidecar, event‑driven control plane that integrates with legacy ETRM, encodes tariff/rider and take‑or‑pay rules, and links cooling and water choices to pricing—driving T+0 re‑pricing , tighter

hedge attribution, and lower settlement variance. Connect with our team to explore how this architecture and operating model can de‑risk your AI growth, prioritize siting and hedges, and deliver measurable P&L stability over the next 90 days.