What Public Service Company of New Hampshire Does

Public Service Company of New Hampshire (PSNH) is a regulated electric utility serving over 500,000 customers in New Hampshire. Established in 1926 and headquartered in Manchester, it operates, maintains, and constructs the electrical distribution infrastructure—poles, wires, substations, and meters—that delivers power to homes and businesses. As a subsidiary of Eversource Energy, it functions within a strict regulatory framework, prioritizing reliability, safety, and compliance. Its core mission is to provide affordable, continuous power, manage outage restoration, and integrate an increasing share of renewable energy sources while navigating the challenges of an aging grid and severe New England weather.

Why AI Matters at This Scale

For a midsize utility like PSNH (501–1,000 employees), operational efficiency and capital expenditure optimization are critical. With annual revenue estimated around $750 million, margins are often tight under regulatory oversight. AI presents a lever to do more with existing assets and personnel. Unlike giant investor-owned utilities with massive R&D budgets, PSNH must be selective, focusing on AI applications that deliver clear ROI in reliability, cost avoidance, and regulatory performance. The transition toward distributed energy resources (like rooftop solar) and electric vehicles also adds grid complexity that manual processes cannot manage effectively. AI can be the force multiplier that allows a regional player to maintain reliability standards and customer satisfaction without proportionally increasing rates or workforce.

Three Concrete AI Opportunities with ROI Framing

1. Predictive Asset Failure Modeling (High Impact) PSNH manages thousands of miles of overhead lines and aging substation equipment. An AI model ingesting historical failure data, real-time sensor readings (like temperature, load), and weather forecasts can predict transformer or cable failures weeks in advance. By shifting from time-based to condition-based maintenance, PSNH could reduce unplanned outages by an estimated 15–20%. The ROI comes from avoiding major capital replacement costs (a failed transformer can cost $50k+), reducing emergency repair labor, and mitigating regulatory penalties for poor reliability metrics. A pilot on the highest-risk circuits would validate the model.

2. AI-Optimized Vegetation Management (High Impact) Tree contact is a leading cause of outages and wildfire risk. Currently, trimming cycles are schedule-based. AI can analyze high-resolution satellite, LiDAR, and drone imagery to identify species growth rates and proximity to conductors. This enables risk-prioritized trimming schedules, focusing crews where they prevent the most outages. For a utility of PSNH's size, this could reduce vegetation-caused outages by 30%, directly improving SAIDI/SAIFI reliability indices that regulators monitor. The ROI includes reduced storm restoration costs and potential liability from fire incidents.

3. Intelligent Customer Engagement Chatbots (Medium Impact) During major storms, call centers are overwhelmed. An AI-powered chatbot and voice assistant, integrated with the outage management system (OMS), can automatically answer 40–50% of routine inquiries (e.g., “Is my power out?”, “When will it be restored?”, “How do I report a downed wire?”). This deflects calls, reduces wait times, and frees human agents for complex issues. The ROI is calculated through reduced temporary staffing costs per storm and improved customer satisfaction scores (CSAT), which are increasingly tied to rate-case outcomes. Deployment via existing web and mobile apps keeps implementation costs manageable.

Deployment Risks Specific to This Size Band

PSNH faces distinct implementation challenges as a midsize utility. First, data readiness: Legacy supervisory control and data acquisition (SCADA) systems and siloed departmental databases (engineering, GIS, customer care) make creating unified data lakes for AI training difficult and expensive. A phased approach, starting with the most valuable data source (e.g., smart meter data), is essential. Second, talent gap: Attracting and retaining data scientists is harder than for tech hubs or larger utilities; partnerships with specialized AI vendors or system integrators may be more viable than building in-house teams. Third, regulatory hesitation: State public utility commissions may be cautious about approving rate recovery for unproven AI investments. PSNH must frame pilots as grid modernization efforts with clear consumer benefits (reliability, cost containment) to secure approval. Finally, cybersecurity: Any new AI system connecting to operational technology (OT) expands the attack surface; robust security protocols and air-gapped testing environments are non-negotiable for a critical infrastructure provider.