AI Agent Operational Lift for San Diego Gas & Electric in San Diego, California

San Diego Gas & Electric (SDG&E) is a regulated investor-owned utility providing electricity and natural gas to over 3.7 million people in San Diego and southern Orange counties. As a critical infrastructure operator in a region prone to wildfires and pushing aggressive renewable energy goals, SDG&E manages a complex network of power generation, transmission, and distribution assets, supported by millions of smart meters and IoT sensors.

Why AI matters at this scale

For a utility of SDG&E's size (5,001-10,000 employees), operational efficiency and grid resilience are paramount. The sheer volume of data generated from grid sensors, smart meters, and weather stations is beyond human-scale analysis. AI and machine learning offer the only viable path to transform this data into predictive insights, enabling proactive management of an aging grid, integration of volatile renewable resources, and compliance with stringent state safety and climate mandates. At this scale, the ROI from even marginal improvements in asset utilization, outage prevention, and workforce efficiency can amount to tens of millions annually.

1. Predictive Maintenance for Grid Reliability

SDG&E's vast physical asset base—from transformers to power lines—requires constant maintenance. AI models can analyze historical failure data, real-time sensor readings (temperature, vibration, load), and environmental conditions to predict equipment failures weeks or months in advance. This shifts maintenance from a reactive, costly model to a scheduled, proactive one. The ROI is clear: reducing unplanned outages improves key reliability metrics (like SAIDI), avoids costly emergency repairs, and enhances customer satisfaction. For a company of this size, preventing a single major substation failure can save millions in equipment and restoration costs.

2. Optimizing Renewable Integration

California mandates force utilities to integrate high levels of solar and wind, which are intermittent. AI-driven forecasting models that fuse weather data, historical generation patterns, and satellite imagery can predict renewable output with high accuracy. This allows for optimized scheduling of conventional power plants and utilization of battery storage, reducing the need for expensive and carbon-intensive "peaker" plants. The financial impact is direct: more efficient grid operations lower fuel costs and reduce penalties for imbalance energy.

3. Enhanced Wildfire Risk Mitigation

In California's high-fire-threat districts, utilities face enormous liability. AI-powered risk analysis, using computer vision on satellite imagery to monitor vegetation growth near lines and machine learning to synthesize weather, fuel moisture, and historical fire data, can dramatically improve Public Safety Power Shutoff (PSPS) decision-making. More precise risk modeling allows for smaller, shorter outages, balancing public safety with customer disruption. The ROI includes reduced wildfire liability—which can be existential—and improved regulatory standing.

Deployment risks specific to this size band

While SDG&E has the resources to fund AI initiatives, its large, regulated nature introduces specific risks. Legacy IT system integration is a monumental challenge, requiring data unification from decades-old operational technology (OT) and new IT platforms. The regulatory environment, while potentially providing cost recovery, also slows experimentation and requires extensive justification for capital expenditures. Furthermore, a company of this size must navigate complex internal change management; convincing seasoned engineers and field crews to trust and act on AI-driven recommendations requires careful change management and proven pilot results. Cybersecurity for AI models and their data pipelines is also a heightened concern given the critical infrastructure involved.