Stratenity POV, Energy & Power Industry Scan

The grid was designed for a load curve that is now obsolete

For two decades US electricity demand was essentially flat, growing under 0.5 percent per year, and utilities optimized for reliability and rate-base growth on a predictable duck curve. That world is gone. Data center load, reshored manufacturing, EV charging, and building electrification are now driving national demand growth forecasts of 2.5 to 3 percent annually, with some interconnection territories in Virginia, Texas, and Georgia seeing localized peak growth north of 10 percent. PJM's 2024 capacity auction cleared at roughly 833 dollars per megawatt-day, nearly ten times the prior year, a direct market signal that firm capacity is scarce.

The bind is structural. Adding load to a distribution feeder that is already near its thermal limit is not a software problem, it is a copper, steel, and permitting problem. Large power transformer lead times have blown out to 100 to 210 weeks, distribution transformers to 12 months or more, and the interconnection queue backlog exceeds 2,600 gigawatts nationally, more than double the entire installed US generation fleet. A utility can want to say yes to a 500 megawatt data center campus and physically be unable to serve it before 2031.

Rate base is a blunt instrument when capital must be sequenced

The regulated utility earns a return on prudent capital investment, typically an authorized ROE between 9.4 and 10.5 percent. That model incentivizes building, but it does not reward building in the right order, and regulators are increasingly skeptical of load forecasts that justify billions in transmission spend against speculative data center interconnection requests, many of which are duplicate submissions across multiple utilities. The financial risk is asymmetric: overbuild and you strand assets in the rate base, underbuild and you throttle economic development and risk reliability penalties.

The worked example that matters: deferring a 120 million dollar substation rebuild with a 30 million dollar storage-plus-flexibility package is a 90 million dollar avoided cost and a faster path to serving load, yet under a pure rate-base incentive the utility prefers the substation. The strategic question is not what to build, it is how to align cost recovery with the lowest-total-cost path.

The grid workforce cliff meets a new skills demand curve

Roughly 25 percent of the utility workforce is eligible to retire within five years, and lineworker, protection engineer, and substation technician roles are the hardest to backfill. At the same time the work itself is changing: a grid saturated with distributed energy resources, inverters, and bidirectional flows requires power-electronics literacy and data fluency that the legacy workforce was never trained for.

• Protection and controls engineers who understand inverter-based resource behavior and can model fault current in a low-inertia grid are the single scarcest skill set.
• Distribution planners now need to run hosting-capacity analysis and probabilistic load forecasting, not deterministic peak-day sizing.
• Field crews need to interpret advanced metering and grid-edge telemetry, shifting from break-fix to condition-based work.
• Cyber-OT specialists who bridge NERC CIP compliance and real-time operations are commanding a 20 to 30 percent wage premium.

Most utilities have data exhaust, not a data asset

Advanced metering infrastructure generates enormous volumes: a single AMI meter reporting at 15-minute intervals produces roughly 35,000 reads per year, and a mid-size utility with two million meters generates 70 billion reads annually. Yet most of that data dies in a meter data management silo, disconnected from the GIS, the outage management system, and the distribution planning tools. The result is that planners still size feeders on nameplate assumptions while sitting on the empirical load data that would let them plan probabilistically.

AI is genuinely useful here, but only on a governed data foundation. Load forecasting models that ingest weather, economic, and AMI signals can cut peak forecast error materially, and computer-vision inspection of transmission assets from drone and satellite imagery can prioritize vegetation management and detect equipment degradation before failure. The precondition is a unified, workspace-scoped data model where every model output carries provenance: which inputs, which version, which assumptions.

FERC, NERC, and the reliability standards are the operating envelope

Nothing ships in this sector without clearing a dense regulatory lattice. NERC reliability standards, including the CIP suite for critical infrastructure protection, carry penalties up to 1 million dollars per violation per day. FERC governs wholesale markets, transmission cost allocation, and interconnection under Order 2023, which reformed the queue toward a cluster study, first-ready-first-served model. At the state level, integrated resource planning and rate cases run through public utility commissions on multi-year cycles.

• NERC CIP: cyber and physical security of the bulk electric system, audited on a defined cycle with severe penalty exposure.
• FERC Order 2023: interconnection queue reform, penalties for withdrawn projects, and firm study deadlines.
• FERC Order 1920: long-term regional transmission planning on a 20-year horizon with mandatory cost allocation.
• State IRP and rate cases: the mechanism through which any capital or program earns cost recovery.

The governance implication is that every planning artifact, forecast, and program justification will be litigated in front of a commission and intervenors. Explainable, versioned, source-traceable analysis is not a nicety, it is the evidentiary standard.

Reliability metrics are the currency of trust and regulation

The customer contract in this sector is measured in SAIDI and SAIFI, the system average interruption duration and frequency indices, and increasingly in the resilience of service under extreme weather. A typical US utility targets SAIDI in the range of 90 to 120 minutes per year excluding major events, but that average masks equity gaps: rural and lower-income feeders often see multiples of the system average.

Extreme weather has made the tail the story. The 2021 Texas winter event and repeated wildfire-driven public safety power shutoffs in the West show that the low-probability, high-consequence event now dominates both customer harm and regulatory scrutiny. Modern reliability strategy must optimize for the worst 1 percent of days, not just the annual average, and must weigh the affordability impact of hardening against the resilience it buys, because the same customers pay for both.

The utility is becoming a platform orchestrator, not a monopoly builder

Serving the new load curve requires capacity the utility cannot build alone and, increasingly, should not own outright. Independent power producers, storage developers, aggregators of distributed energy resources, and hyperscale customers are all now counterparties in grid planning.

• Hyperscalers are signing power purchase agreements and even co-locating behind-the-meter generation, turning the largest customers into partial supply partners.
• Virtual power plants aggregating residential storage and smart thermostats can now bid into wholesale markets, giving utilities a dispatchable resource they do not own.
• Transmission developers and merchant lines are financing bulk capacity that a single utility could not rate-base alone.
• Equipment supply partnerships and reserved manufacturing slots for transformers and switchgear are now a strategic procurement discipline, not a purchasing formality.

The integrated resource plan as a living, governed artifact

The recurring failure in this sector is that strategy is produced as a static document on a triennial cadence while the load, supply, and regulatory environment shift monthly. Stratenity's position is that the integrated resource plan, the interconnection strategy, and the capital sequencing plan should be a single, versioned decision artifact with defined inputs, constraints, and dependencies, updated continuously and always carrying its provenance into the rate case.

Concretely, that means load forecasts, non-wires alternative screens, and reliability targets are typed units that AI agents can update as new AMI and queue data arrives, but every consequential output passes a human prudency checkpoint before it becomes a commission filing. The strategic edge is not a better forecast, it is a faster, defensible, and traceable decision loop.

Where to focus the next twelve months

• Build a large-load interconnection playbook with a flexibility-first tariff so speculative data center requests are de-risked before they justify billions in transmission capex.
• Stand up a non-wires-alternatives screening gate that must be cleared before any distribution capital project above a set threshold is approved for rate base.
• Unify AMI, GIS, and outage data into a single governed model so distribution planning moves from deterministic peak sizing to probabilistic hosting-capacity analysis.
• Reserve transformer and switchgear manufacturing slots under multi-year supply agreements and treat lead time as a first-class planning constraint.
• Reframe the IRP as a continuously versioned artifact so the rate case is defended with traceable, current analysis rather than an 18-month-old snapshot.

Sector-specific levers with a metric on each

• Deploy dynamic line rating and advanced conductors to unlock 10 to 30 percent more transfer capacity on existing corridors, measured as megawatts added per rights-of-way dollar.
• Contract curtailable large-load flexibility to serve new demand faster, measured as gigawatts of load served within 18 months versus the interconnection baseline.
• Defer distribution capital with storage and DER packages, measured as avoided capex dollars per deferred substation upgrade.
• Harden the worst-performing 5 percent of feeders, measured as reduction in SAIDI on the bottom-decile circuits, not the system average.
• Cut peak load forecast error with AMI-fed models, measured as percentage-point improvement in forecast accuracy feeding capacity and IRP decisions.

Related reading

Put this sector view to work with the cross-cutting Stratenity frameworks.