POTENTIAPower

Energy & Power

Power companies face rising load on constrained grids. We train utility, generation and trading leaders to apply AI to forecasting, dispatch, asset health and interconnection, and to govern it as a critical system.

IThe landscapeAI for the grid, the fleet and the trading floor.

The landscape

Many power companies are planning for sustained load growth for the first time in a generation. Data centers, electrification and new manufacturing are adding demand, while interconnection queues, transmission limits and equipment lead times slow what can be built. AI sits on both sides of that equation. It drives much of the new load, and it can sharpen how assets are forecast, dispatched, maintained and financed. Arclight trains generation, grid and trading leaders to make those calls with precision, using their own fleet, market rules and regulatory obligations as the material.

  1. I

    Load growth against grid limits

    Data centers, electrification and reshored manufacturing are requesting capacity faster than interconnection studies, transmission build-out and equipment supply can deliver it. Every load forecast now carries a capital commitment and a reliability risk.

  2. II

    Reliability with a changing fleet

    Retiring thermal units, variable renewables and battery storage change how reserve margins, ramping and outages are planned. The reliability obligation is unchanged, but the fleet is harder to forecast and extreme weather tests it more often.

  3. III

    Every dollar under review

    Rate cases, merchant price volatility and investor expectations keep capex and opex under constant scrutiny. Boards must show that AI spending improves reliability, cost or speed to market, and that it is governed like any other critical system.

IIApplications06

Where AI earns its place.

Six applications with proven economics in this sector, taught with the techniques behind them and the ways they fail.

  • 01

    Load, renewable and price forecasting

    Forecast load, wind and solar output and nodal prices at the horizons that matter for bidding, hedging and resource planning. Executives learn to read forecast error in dollars and to assign accountability for model performance.

    • Probabilistic time-series forecasting
    • Gradient-boosted models
    • Bayesian hierarchical models
  • 02

    Asset health and outage prevention

    Detect degradation in turbines, transformers, inverters and boilers from SCADA, sensor and inspection data before it becomes a forced outage. The decision is when to intervene, and what that does to availability and the maintenance budget.

    • Anomaly detection
    • Physics-informed ML
    • Survival analysis
  • 03

    Dispatch, bidding and storage

    Optimize unit commitment, battery charging and market bids against physical limits, market rules and price uncertainty. Leaders learn where mathematical optimization already pays, where reinforcement learning is ready and where operators must stay in control.

    • Mixed-integer optimization
    • Stochastic programming
    • Reinforcement learning for control
  • 04

    Grid planning and congestion

    Estimate hosting capacity, congestion and the effect of distributed energy resources across the network, fast enough to test many scenarios. Planners screen options with machine-learned approximations, then confirm the shortlist with full engineering studies.

    • Surrogate power-flow models
    • Graph neural networks
  • 05

    Interconnection and site origination

    Screen sites against queue position, available capacity, land, permitting, environmental constraints and community factors before committing development capital. Geospatial models turn a long list of parcels into a ranked, defensible pipeline.

    • Geospatial ML
    • Satellite imagery analysis
    • Multi-criteria decision analysis
  • 06

    Regulatory and contract intelligence

    Search and compare tariffs, interconnection agreements, PPAs and regulatory filings with citation-grounded language models. Teams prepare rate case responses, contract reviews and compliance evidence faster, with every answer traced to its source.

    • Retrieval-augmented generation
    • Document intelligence
    • LLM agents
IIIThe sector track04 modules

The Energy curriculum.

Taught within the Command Intensive, as an on-site immersion, or privately in one-to-one advisory.

  1. I

    AI across the power value chain

    Where AI creates measurable value in generation, networks, retail and trading, and where conventional engineering remains the better tool. Participants map use cases to reliability metrics and the P&L, using their own fleet as the case.

  2. II

    Forecasts, markets and risk

    How probabilistic forecasts feed bidding, hedging and resource adequacy. Executives learn to read error distributions, set model risk limits and question a trading model the way they would question a trader.

  3. III

    Operational technology and safety

    Deploying models near SCADA, energy management and control systems without compromising safety or NERC CIP compliance. Covers human-in-the-loop design, cyber exposure and change control for models that influence physical equipment.

  4. IV

    Capital allocation and governance

    Managing AI as a capital program, with business cases, stage gates, vendor selection and build-or-buy decisions. The module closes with a board-ready plan the team can defend to regulators and investors.

Who attends

  • CEOs and executive committees
  • CFOs and heads of corporate development
  • COOs and fleet operations leaders
  • Heads of trading, origination and risk
  • Transmission and grid planning directors
  • Chief digital, data and technology officers

Outcomes

  • A ranked AI portfolio tied to reliability, margin and capital plans
  • The fluency to challenge forecasts, models and vendor claims
  • Governance suited to operational technology and reliability standards
  • A 12-month roadmap with owners, stage gates and value measures

Begin

A sector briefing for your leadership, within seventy-two hours.

Tell us about your assets and the decision in front of you. A principal replies within one business day.