The Real P&L Drivers: Capacity, Riders, and Data Center Load

Opening Insight

Unit prices are rising even as volumes soften, and the reason is not elusive demand but structured pass‑throughs: capacity repricing (PJM’s 2025/26 BRA at $269.92/MW‑day embedded via PLC × BRA), higher T&D and wildfire riders, and a fuel mix that keeps gas on the margin as coal days‑of‑burn narrow. The effects are uneven—double‑digit increases in DC, Pennsylvania, and Maryland while Connecticut, Arizona, and California edge down, with California’s wildfire charges taking a mid‑teens share—but the operating result is consistent: budget variance, margin leakage, collateral stress, and operational friction. Concentrated data‑center load is amplifying these pressures, reshaping PJM and accelerating visible multi‑billion‑dollar capacity uplift, exposing operating models built for yesterday’s weather, tariffs, and legacy ETRM workflows. This post translates what’s changing into what to do: the costs of inaction, the upside of fixing the model, and a pragmatic answer—an event‑driven control plane with governed data, ETRM modernization, and agentic AI that turns tariffs, riders, and capacity math into first‑class, auditable objects. We map that design to an Arcelian architecture, a phased roadmap, and measurable KPIs to compress decision cycles and harden risk. For the market backdrop and mechanics that set up this thesis, continue to Context and Analysis.

Consequences of Inaction

Unit prices are rising even as demand softens; treating this as noise turns temporary variance into structural damage. Capacity repricing, rider acceleration, and tighter fuel optionality will bleed through budgets, controls, and customer outcomes.

• Margin leakage: Misaligned hedges that ignore capacity math and gas‑on‑margin days push spreads negative as PLC × BRA flows straight to bills (e.g., ~$269.92/MW‑day ≈ about $8/month per 1 kW PLC, roughly 0.8¢/kWh at 1,000 kWh).
• P&L distortion: The price proxy climbed 8.3% to 14.17¢/kWh while sales fell 1.7%, creating budget variance and accrual noise as energy, capacity, T&D, and wildfire riders reset on different clocks; over the next 60–90 days, gas, hydro, and CPUC decisions can shift true‑ups.
• Credit and collateral stress: Higher PJM capacity prices for 2025/26 and data‑center‑driven uplift push bills up (often double‑digit in parts of PJM), triggering collateral calls and weakening C&I collections when rate shocks hit.
• Regulatory and audit exposure: California’s wildfire and insurance charges now take a mid‑teens share of bills and are expected to drive most of the Q4 delta; thin documentation around pricing, procurement, and hedge rationale in competitive states invites audit findings.
• Operational fragility: Coal days of

burn narrowed (PRB ~55–65; ILB/Appalachian ~60–70), cutting dispatch optionality and leaning harder on gas; schedulers face more imbalance risk as fuel riders react to DA/RT volatility.

• Competitive disadvantage: Regional spreads are wide—DC (+30.3%), Pennsylvania (+21.7%), Maryland (+20.9%) versus Connecticut (‑5.1%) and California (‑0.5%)—so late reactions mean mispriced offers and customer churn where demand and fuel‑mix signals shift fastest.

Benefits of Fixing the Model

• Finance: Codifying PLC × BRA and rider mechanics makes bills forecastable, reducing budget variance and smoothing cash flow. Anchors are explicit—e.g., $269.92/MW‑day, mass‑market capacity adders ~0.6–1.0¢/kWh, T&D +2–4% YoY, and California wildfire/insurance in the mid‑teens share.
• Front office: Hedges align to real demand and capacity signals (HDD/CDD, data‑center load) and fuel‑mix cues (gas‑on‑margin, coal days of burn), improving P&L capture when adders reset (~0.81¢/kWh at 1,000 kWh from PLC math).
• Risk and credit: Earlier, policy‑driven triggers tied to rider and capacity repricing cut surprise collateral calls; exposures reflect tariff updates and BRA outcomes rather than stale assumptions, hardening recoveries.
• Operations and scheduling: Event‑driven workflows speed nominations and adjustments on gas‑on‑margin days, hydro swings, or coal logistics constraints, lowering rework and imbalance risk as schedules mirror live fuel‑mix telemetry.
• Compliance: Rules‑as‑software with audit‑ready workflows preserve rationale for pricing, procurement, and hedges in competitive states, shrinking audit exception risk while maintaining clear documentation.
• Technology and data: A governed data layer normalizes tariffs, riders, HDD/CDD, capacity prices, and fuel‑mix indicators; ETRM treats rates/riders/capacity adders as first‑class objects, cutting settlements variance and latency.
• Organizational cadence: Integrated front/middle/back office creates traceability from drivers to action, sharper attribution and accountability, and a faster weekly decision rhythm with threshold‑based credit re‑scores and hedge adjustments.

Unified Control Plane

Deploy a control plane—an event‑driven, data‑centric operating model—to align pricing, hedging, credit, and operations to live tariff and fuel‑mix signals. It addresses higher unit prices (average retail revenue up 8.3% to 14.17¢/kWh while retail sales fell 1.7%), uneven regional demand, capacity pass‑throughs, and tighter coal buffers that keep gas on the margin. It also makes PJM’s PLC × BRA math ($269.92/MW‑day, ~0.8¢/kWh at 1,000 kWh) and California’s wildfire riders actionable in forecasts, procurement, and collateral.

• Data architecture and lineage: Standardize tariffs, riders, HDD/CDD, capacity prices, and fuel‑mix indicators (gas share, nuclear/renewables availability, coal stock drawdowns) into governed data products with traceable lineage.
• ML‑driven forecasting and optimization: Refresh load, shape, and price views weekly or daily

Energy Risk Control Plane for Retail Power: ETRM Modernization, Agentic AI, and Event Streaming

• Procurement and hedging optimization: use weather, interconnection delays, outage data, and localized data center siting to optimize procurement and hedging with explainable constraints.
• ETRM modernization: treat rates, riders, and capacity adders as first-class objects; embed PJM/ISO capacity, congestion, and basis scenarios; wire settlements to real-time meter and tariff signals to cut variance.
• Agentic AI and workflow automation: use policy-guardrailed agents to watch tariffs, capacity auctions, and regional fuel mix and to propose hedges, nominations, and credit adjustments with audit trails.
• API and event integration: stream market, weather, and operational telemetry into risk and credit engines; trigger re-scores, collateral calls, and hedge adjustments on threshold breaches.

With a single control plane, clear decision rights, and a cadence of weekly reviews and monthly attestations, teams act sooner and reduce settlements and cash-flow variance.

Arcelian Architecture and Roadmap Setup

Setup: Prices are rising even as volumes wobble—average retail revenue per kWh climbed 8.3% to 14.17¢ while sales fell—driven by PJM capacity repricing, gas on the margin, tighter coal optionality, and wildfire riders.

Arcelian operationalizes a single, governed control plane so front, risk, credit, and ops act on the same signals before they hit bills and cash.

Architecture

• Control plane: one event-driven layer spans front, risk, credit, and operations with rules-as-software and human-in-the-loop approvals, creating audit trails and shortened decision cycles.
• Governed data products: normalize tariffs and riders, HDD/CDD anomalies, capacity prices, fuel-mix, and coal stock drawdowns (e.g., PRB days of burn in the 55–65 range) alongside price and demand proxies. Encode bill math like PLC × BRA $/MW-day → $/kW-month → ¢/kWh (e.g., ~ $269.92/MW-day ≈ ~ 0.81¢/kWh at 1,000 kWh) with lineage.
• ETRM modernization: treat rates, riders, and capacity adders as first-class objects; embed PJM/ISO scenario analysis; and tie settlements to real-time meter and tariff signals to reduce variance.
• Agentic monitoring: policy-guardrailed agents watch tariffs, capacity auctions, HDD/CDD, congestion, and fuel-mix/coal inventory telemetry, proposing hedges, nominations, and credit adjustments with documented rationale.
• API and event streaming: stream market, weather, and operational events into risk/credit to trigger re-scores, collateral calls, and hedge adjustments when thresholds trip.

Roadmap

• 1) Launch a 4–6 week diagnostic on pricing ingestion, demand forecasting, and fuel-mix telemetry to size margin at risk and control gaps.
• 2) Normalize tariffs, riders, HDD/CDD, capacity, fuel-mix, coal stockpiles/days-of-burn, and price/demand proxies into governed data products with lineage.
• 3) Stand up ML forecasting and optimization cadences (weekly/daily) for load, shape, and price, including

Execution Priorities for ETRM and Control Plane

Localized data center effects.

• 4) Redesign ETRM and the control plane: create rate/rider/capacity objects, add PJM/ISO scenario analysis, and wire audit‑ready workflows.
• 5) Deploy agentic monitoring and alerts with human‑in‑the‑loop approvals; convert manual rules into software artifacts.
• 6) Implement API/event integration to trigger credit/collateral/hedge actions and synchronize settlements to real‑time meter/tariff signals.
• 7) Redesign credit and collateral policies tied to capacity/tariff shocks; institute weekly cross‑functional reviews and monthly risk‑and‑controls attestations.

Operating Model and Governance

• Decision rights for portfolio managers to act on model‑driven signals with documented rationale for compliance.
• A single control plane with clear SLAs across front, risk, credit, and ops.
• Rules‑as‑software workflows with lineage and audit readiness; event‑driven triggers with human approvals.
• Event logs connecting market drivers to commercial action for traceability.

Roles and Culture

• CIO: Owns data architecture, ETRM modernization, and API/event enablement.
• COO: Owns operating cadence and the control plane; drives scheduling throughput.
• CFO: Sets risk appetite and working‑capital and collateral policy tied to capacity and rider shocks.
• Capability uplift: Train schedulers/analysts to interpret models; preserve portfolio manager decision rights with clear documentation.

KPIs

• Lower variance in settlements and cash flow.
• Higher scheduling throughput and fewer reworks.
• Earlier credit triggers and better recoveries.
• Stronger compliance and monthly attestations.
• Clearer attribution and accountability from driver to action.

Trade-offs

The design balances speed and governance via guardrails and audit trails, manages gas‑on‑margin reliance as coal optionality tightens, and prioritizes capacity and wildfire riders first, then fuel adjustments.

Align Trading and Risk

Unit prices are rising even as volumes wobble: average retail revenue per kWh climbed 8.3% to 14.17¢ while retail sales slipped 1.7%, driven by energy and capacity pass‑throughs (including PJM’s $269.92/MW‑day BRA ≈0.8¢/kWh at 1,000 kWh ), rising T&D and wildfire riders, gas on the margin, and coal stock drawdowns that cut dispatch optionality, with days of burn in the 50s–60s . The result is budget variance, P&L leak, collateral strain, and operational friction, amplified by data center load reshaping PJM and pushing uplift with double‑digit bill impacts . Long term, trading must align hedges to capacity and fuel‑mix signals, risk must tighten around rider and congestion scenarios, and leaders must own traceability from drivers to action. Adopt an event‑driven, data‑centric, control‑aligned operating model with a single control plane, clear decision rights, and a cross‑functional cadence so trading, risk, and operations act on capacity, rider, and

Fuel‑mix signals before they hit the bill.

Implement the Operating Model

Unit prices are up while retail sales soften, with energy, capacity, T&D, and wildfire riders flowing through as coal‑to‑gas shifts compress optionality. Arcelian links these pass‑throughs to front, risk, credit, and settlements—repairing the integrated operating model now driving structural margin risk.

• Market‑to‑machine integration: Codifies tariffs, riders, capacity prices (PJM BRA, PLCs), and fuel‑mix signals into one source—reducing P&L distortion and settlements variance.
• Forecasting and optimization accelerators: Refresh load/price forecasts with HDD/CDD and data center clustering to align hedges on gas‑on‑margin days—limiting margin leakage.
• ETRM and control‑plane modernization: Make capacity adders and rate components first‑class objects with audit‑ready workflows—synchronizing settlements and shrinking true‑up noise.
• Agentic monitoring and alerts: Watch HDD/CDD, congestion, auctions, and coal stock drawdowns; trigger actions before rider and capacity shocks hit P&L.
• Credit and collateral redesign: Tie capacity repricing and riders to exposure analytics—reducing collateral spikes and counterparty surprises.

Commission a 4–6 week diagnostic on monthly electricity market update process—pricing ingestion, demand forecasting, and fuel‑mix telemetry—to quantify margin at risk and deliver a roadmap this quarter.

Integrating Agentic AI with Legacy ETRM: a practical modernization strategy

For most shops, the right ETRM architecture choice is augmentation, not replacement. Start by externalizing a governed data plane (rates, riders, capacity, tariffs, meters, positions) and layering an event‑driven control plane on top of the core system. Treat rates/riders/capacity as first‑class objects with versioned schemas and APIs; stream meter and tariff changes via topics your agents can subscribe to; and expose settlements, credit, and exposure calculations as callable services rather than brittle batch jobs. This preserves the book of record while enabling AI/agentic monitoring, ML forecasts, and workflow automation to act on real‑time signals across front, middle, and back office.

Sequencing matters. In phase one of the integration roadmap, stand up CDC from the ETRM into a streaming backbone with a schema registry; publish governed data products for tariffs, meters, trades, and reference data; and wrap legacy functions with idempotent APIs. Phase two adds ML forecasts (load/price/renewables) and agentic workflows to reconcile exposures, propose dispatch adjustments, and pre‑clear settlements against real‑time tariff rules—enforcing policy‑as‑code for limits, approvals, and audit. Phase three expands to counterparty risk and compliance agents, wiring intraday credit checks and trade surveillance into the same event fabric. This sequencing operationalizes the event‑driven control plane described earlier—augmenting the legacy ETRM with governed

Modernizing ETRM with data products, streaming APIs, and agentic monitoring

Adopt data products, streaming APIs, and agentic monitoring rather than forcing a wholesale rebuild.

Key decisions and measurable outcomes

Criteria

• Acceptable data latency by control (e.g., P&L vs. credit)
• Vendor extensibility
• API coverage
• Lineage and entitlements for sensitive fields

Trade-offs

• Duplicate state vs. strict ETRM primacy
• Synchronous APIs vs. asynchronous events
• Model autonomy vs. human-in-the-loop for high-impact actions

Risks

• Model drift impacting limits
• Shadow ledgers from partial settlement offloads
• Weak identity and segregation on agent actions

Outcomes

• 50–80% fewer manual settlement adjustments
• Sub‑5‑minute intraday exposure refresh
• >30% reduction in exception cycle time
• Audit‑ready event trails spanning front, middle, and back office

Frequently Asked Questions

What’s driving higher unit prices even as volumes soften?

Pass‑throughs are the main driver. Capacity repricing (e.g., PJM’s 2025/26 BRA at $269.92/MW‑day), rising T&D and wildfire riders, and fuel‑mix dynamics are flowing more cost into the unit rate. Coal days‑of‑burn have tightened (roughly 55–70 days), keeping gas on the margin and lifting fuel riders where gas sets price. Weather muted winter load (fewer HDD), so demand fell even as bills rose. Impacts vary by region—DC, Pennsylvania, and Maryland saw the largest increases, while Connecticut, Arizona, and California edged down, with California’s wildfire charges taking a mid‑teens share of bills.

How do PJM capacity charges show up on customer bills, and what’s the practical impact?

Capacity is billed as PLC × BRA. With the BRA clearing around $269.92/MW‑day, a 1 kW PLC maps to roughly $8/month, or about 0.8¢/kWh at 1,000 kWh—often a 0.6–1.0¢/kWh adder for mass‑market. When capacity resets higher, it passes straight through to bills, widening budget variance and triggering collateral and credit stress if pricing and hedges aren’t aligned.

How can an event‑driven control plane with agentic AI help a legacy ETRM in this environment?

By aligning pricing, hedging, credit, and operations to live tariff and fuel‑mix signals. Normalize tariffs, riders, HDD/CDD, capacity prices, and fuel‑mix into governed data products; treat rates, riders, and capacity as first‑class ETRM objects; and use policy‑guardrailed agents to watch auctions and fuel‑mix and propose hedges, nominations, and credit actions with audit trails. Teams typically see lower settlements variance, earlier credit triggers, faster adjustments on gas‑on‑margin days, and clearer documentation. A 4–6 week diagnostic can quantify margin at risk and stand up the data and workflow foundations.

Trend Watch

Event‑driven control planes with agentic AI are shifting ETRM from a batch ledger to a living

Risk Surface and Electricity Demand Trends 2026

risk surface. That matters as electricity demand trends 2026 skew upward on concentrated data center electricity demand while retail electricity price increase dynamics persist (retail revenue per kWh ~ 14.17¢ ). The combination of PJM capacity prices 2025/26 from the PJM Base Residual Auction and California wildfire riders turns static assumptions into daily P&L drivers—exactly where legacy workflows lag.

What to Operationalize Now: PLC × BRA, Rider Intelligence, Fuel‑Mix, Load Localization

• PLC × BRA as a streaming feature: Publish governed data products for capacity tags and BRA updates; have agents simulate bill impact intraday and recommend hedge shifts when PJM capacity uplift or interconnection delays alter forward shapes.
• Rider intelligence: Track California wildfire riders and T&D resets as first‑class ETRM objects; trigger credit re‑scores and collateral calls proactively to cut settlements variance and surprise exposure.
• Fuel‑mix sentinels: Monitor gas on the margin conditions and coal stockpiles days of burn thresholds; auto‑propose nominations and shape hedges when PRB/ILB inventories tighten.
• Load localization: Fuse meter telemetry with siting news to flag hyperscale clusters; route offers and risk analytics to zones where elasticity is lowest.

Why This Is Different: AI in ETRM and Policies as Code

• AI in ETRM only pays when policies are encoded as software. Agentic AI uses audit‑guardrails to explain decisions, reducing credit and collateral stress while improving attribution.
• Firms that treat tariffs, PLC × BRA, riders, and fuel‑mix as code will price faster and clearer—turning negative sentiment into advantage as data center electricity demand collides with capacity scarcity.

Closing Insight

Markets are telling us that price is now a function of policy and physics as much as demand; winning shops will operationalize that reality. Treat PLC × BRA, wildfire riders, T&D resets, and fuel‑mix telemetry as code inside a unified control plane, so hedges, credit, and settlements adjust before volatility reaches the bill. With data‑center clustering tightening capacity and keeping gas on the margin, agentic AI must move beyond dashboards to policy‑guardrailed actions—re‑scoring collateral, reshaping nominations, and repricing offers intraday with audit trails. The strategic edge is digital resilience: modernized ETRM anchored in governed data and event streams that compress decision cycles and harden risk management. Start with a 4–6 week diagnostic to quantify margin at risk and institutionalize a weekly decision rhythm.

Partner with Arcelian

Prices are rising even as volumes soften, with PJM capacity repricing (PLC × BRA), wildfire/T&D riders, and gas-on-margin days turning variance into structural risk—especially where data‑center clustering tightens supply. Arcelian partners with

Operationalize a Single Control Plane Across Tariffs, Rates, and Capacity

CFOs, COOs, and CIOs: Operationalize a single control plane that unifies governed tariff and rider data, modernizes ETRM to treat rates and capacity as firstclass objects, and deploys policyguardrailed agents that align hedging, credit, and settlements.

• Governed tariff and rider data
• ETRM modernization that treats rates and capacity as firstclass objects
• Policyguardrailed agents aligning hedging, credit, and settlements

Outcomes: cutting variance, advancing credit triggers, and reducing collateral shocks.

Focused 46 Week Diagnostic

Connect with our team to scope a focused 46 week diagnostic that:

• Sizes margin at risk
• Pressuretests PLC BRA and rider exposure
• Builds a pragmatic roadmap to augment your legacy ETRM this quarter