The Transmission Capacity Question FERC Just Made Impossible to Dodge

On June 18, 2026, the Federal Energy Regulatory Commission issued Show Cause Orders to six of North America's largest grid operators — PJM, MISO, CAISO, SPP, NYISO, and ISO New England — directing each to justify or revise its transmission tariffs for connecting large electricity consumers to the grid. The trigger was well-understood: AI data centers are attaching to the grid at unprecedented speed and scale, creating concentrated load at specific transmission nodes that the existing interconnection queue never anticipated.

The headlines focused on transmission policy and data center economics. But for asset managers at those grid operators, and at every utility feeding into those systems, the orders carry a more uncomfortable subtext: FERC is now asking grid operators to account for transmission capacity in precise, defensible terms. And transmission capacity isn't a planning formula — it's the live sum of the health of every asset on the network.

Capacity Is an Asset Management Problem

When FERC asks a grid operator to justify what transmission capacity is available on a given corridor, it is implicitly asking about every transformer, every conductor segment, every switching station, and every protection relay on that path. The theoretical line rating assumes ideal conditions. Real available capacity is what remains after you account for the actual age, maintenance history, current health, and operating context of every physical asset along the corridor.

That's not a question a generation planning model can answer. It's a question that requires real-time, accurate, and unified asset data.

For most grid operators and utilities, this is where the split EAM problem becomes an operational liability rather than a technology inconvenience. The transmission network — corridors, towers, underground cables, conductor segments — is managed spatially, as a linear network with GIS-native data, topology rules, and network-graph dependencies. The equipment at each node — transformers, breakers, protection systems, substation facilities — is managed hierarchically, with parent-child failure taxonomies, OEE-grade telemetry, and work order trees tied to individual assets.

Two different asset logics. Two different platforms. One question from FERC that neither system can answer alone.

The Scale of What's Connecting

The data center surge adds urgency to a problem that was already growing. More than two-thirds of maintenance teams across asset-intensive industries plan to adopt AI-driven maintenance capabilities by end of 2026. The EAM market is projected to grow from $7.27 billion in 2025 to $16.42 billion by 2034, driven in part by the need to instrument and analyze assets that were previously managed on inspection cycles.

But instrumenting more assets with IoT sensors and AI models doesn't solve the split-system problem — it compounds it. A machine learning model trained on transformer health data at a substation won't account for the network load that a new hyperscale data center connection is placing on that transformer's upstream corridor. A GIS-based network analysis won't know that the transformer it's routing capacity through had an anomalous dissolved gas reading three weeks ago.

Predictive accuracy within a silo is not the same as predictive accuracy for the network. And when FERC or a transmission system operator asks for justification of available capacity, the answer has to account for both.

What Unified Reasoning Actually Looks Like

Managing linear assets well means understanding that every segment is part of a network with physical laws. A thermal limit on a transmission conductor isn't independent — it's influenced by ambient temperature, load flow from adjacent paths, and the health of the switching equipment at each end. GIS-native reasoning, cascade failure modeling, and dependency graph analysis are the operating assumptions, not optional features.

Managing discrete assets well means understanding the hierarchy. A transformer failure isn't just a single asset event — it propagates up to the substation, to the transmission zone, to the load pocket. ISO 14224 failure taxonomies, criticality rankings calibrated against OEE targets, and parent-child maintenance scheduling all depend on seeing the full asset tree.

When those two models are held in the same platform — with AI agents reasoning across the boundary, not just within it — you get answers that neither system can produce alone. A capacity query returns results informed by both the network topology and the health state of the equipment at each node. A maintenance recommendation accounts for both the asset's failure probability and its network position.

And every reasoning step — every anomaly flag, every risk ranking, every maintenance deferral — is captured in a decision log that can be produced for a regulator, an auditor, or a board without a quarterly spreadsheet reconciliation exercise.

Compliance Is No Longer a Lagging Indicator

NERC CIP-012-2, which took effect July 1, 2026, extends protection requirements to real-time operational data exchanged between control centers. CIP-003-9 enforcement began April 1, expanding governance requirements for vendor remote access and supply chain across the low-impact BES Cyber System tier. Penalties for non-compliance can reach $1 million per violation per day.

What these standards presuppose is that the organisation can produce a clear, consistent record of what assets exist, what data they generate, and what decisions were made about them. Two systems, two data models, and two decision logs don't produce that record — they produce a reconciliation problem that grows with every new sensor, every new AI model, and every new regulatory update layered on top.

The transmission capacity question FERC just made explicit is one version of a question that regulators across sectors are starting to ask in parallel. The answer, for any organisation managing both linear networks and facility assets, starts with whether those two asset worlds share a single source of truth.

---

Curious how UniEAM handles both? Request a demo.