“Smart city” has been an active vendor category for over a decade. The honest verdict is mixed: some specific AI applications in utilities are genuinely mature, a few XR uses are proving their value, and a large share of the broader vision remains aspirational. This piece separates the working pieces from the hype.
What Is Actually Working in Smart City Utility AI
AI applications in utilities that are generating real operational outcomes today fall into a short list.
AMI-driven analytics are the most broadly deployed. Itron and Landis+Gyr smart meter networks produce interval data that supports load forecasting, non-technical loss detection, and, in water utilities, pressure and consumption anomaly detection that helps identify distribution leaks. These are production systems, not pilots.
DERMS platforms, with AutoGrid as a notable example, use ML-based forecasting to dispatch distributed energy resources, rooftop solar, batteries, demand response, against grid constraints. Several grid operators in North America and Europe are running these in production to defer distribution upgrades.
ADMS platforms from GE Vernova, Schneider EcoStruxure, and Oracle NMS use AI-assisted FLISR (fault location, isolation, and service restoration) to reduce outage duration. The AI element here is primarily pattern matching on historical switching data to recommend restoration paths, with human operators approving the final actions.
Customer-facing AI, chatbots and usage advisors, is rolling out through CIS platforms. Cayenta CIS (Harris Computer) has added AI tools through its Cayla AI module. SAP C4U includes customer engagement analytics tied to S/4HANA Utilities. Oracle CC&B integrations with Oracle CX layer in AI-driven service suggestions. These are incremental improvements to existing CIS workflows, not replacements for the billing system.
Where XR Fits in a Smart City Utility
XR in a smart city utility context is useful in three specific scenarios, and it does not belong in a fourth.
Field inspection: AR overlays on mobile devices or headsets surface asset records, last-inspection dates, and wiring diagrams when a technician is standing at a piece of equipment. This reduces paper-based lookups and speeds up fault isolation. The data comes from the ERP asset module or GIS.
Crew training: VR simulations for substation operations, gas pipeline emergency response, and distribution switching let new workers practice procedures safely. A 45-minute VR session simulating a substation energization sequence is far safer and cheaper than equivalent live training. See also VR for utility workforce training.
Design and capital project review: VR walkthroughs of planned substations or pipeline reroutes let engineers and planners identify clearance and routing conflicts before breaking ground. This is a design tool, not an operations tool.
The scenario where XR does not fit: routine customer billing, CIS operations, or SCADA monitoring. The systems of record for those functions are the CIS (SAP IS-U, Oracle CC&B, Cayenta) and the ADMS. An operator running switching procedures is working from an ADMS console, not a mixed-reality headset.
The “Metaverse for Utilities” Problem
Several consulting firms and vendors have published material on the “utility metaverse”, persistent virtual environments where operators manage grid assets, customer service runs in 3D, and field crews collaborate immersively. This framing is mostly marketing.
The practical barriers are not primarily technical. They are organizational: utilities run on safety-critical processes with regulatory oversight. Changing the operator interface for grid switching or the intake workflow for a CIS requires regulatory approval, workforce retraining, and extensive testing. Utilities move carefully for good reasons. The realistic near-term outcome is incremental AR tooling for specific field tasks, not a wholesale shift to immersive operating environments.
Data and System Integration Remains the Hard Problem
Smart city utility initiatives that have stalled most often did so because of data integration problems, not technology limitations. Smart meters, GIS, ADMS, CIS, and ERP all hold different views of the same physical infrastructure and customer base. Getting those views synchronized well enough to support AI analytics requires data governance work that is unglamorous but essential.
Utilities that have made genuine progress on smart city AI have typically invested heavily in their data platform first, MDM (master data management) for customer and asset records, standardized AMI data pipelines, and ERP data quality improvement, before layering analytics on top.
For the AI and operations angle, see AI for utility operations and the smart grid optimization coverage.
An Honest Summary
Smart city utility AI is real in a bounded set of applications: AMI analytics, DERMS dispatch, ADMS-assisted outage management, and CIS-embedded customer tools. XR is useful for field inspection support, crew training, and design review. The broader “metaverse for utilities” vision is not supported by current procurement, regulatory, or organizational reality. The ERP and CIS stay at the center of utility operations regardless of what the smart city stack adds around the edges.