AMI Data + AI: Balancing Grid Affordability - Utility Analytics Institute

An aging grid, surging AI and EV energy demand, and rising inflation are intensifying affordability concerns for utility customers. This pressure has placed utilities onto a dual track: requesting rate increases to modernize infrastructure while aggressively expanding relief programs for income-qualified customers struggling to keep pace with rising costs (S&P Global, 2026). The solution can be found within the AMI data utilities have invested in for years.

Even where funds are available for assistance programs (think bill relief, dynamic rate structures, and weatherization rebates), utilities can face a persistent gap between program availability and actual customer participation. According to the American Council for an Energy-Efficient Economy (ACEEE), low-income households represent 27.5 percent of U.S. households but receive only 13 percent of utility efficiency spending (2022).

Why the disconnect? Because identifying, prioritizing, and engaging the right households at scale remains a massive operational hurdle.

Reliance on coarse, fragmented census data can create a blind spot where many income-qualified customers go undetected. Simultaneously, middle-income households—who typically do not meet strict federal eligibility thresholds despite facing significant financial strain—are overlooked. Ultimately, failing to account for the full spectrum of eligible households can make regulatory targets hard to reach, leave critical funding on the table, and miss opportunities to help reduce arrears.

Overhauling these outcomes requires a total paradigm shift in how utilities target and communicate with vulnerable residents by adopting predictive, data-driven analytics that optimize enterprise-wide efforts.

From Raw Intervals to Household Insights

Fortunately, the foundation for this data-driven shift already exists. Over the last 13-plus years, U.S. Department of Energy shows billions of dollars have been invested in grid modernization, including Advanced Metering Infrastructure (AMI) (n.d.). By layering machine learning (ML) and artificial intelligence (AI) onto foundational AMI data, utilities are tapping into behavioral insights that shift the focus from when energy is used to how it is used.

AMI data demonstrates value across multiple applications, including:

AMI data has demonstrated value across multiple applications, including:
EV Detection & Optimization: Pinpoint charging behavior in real-time to drive off-peak incentives and prevent grid-level transformer strain.

Appliance Inefficiency & Anomaly Detection: Spot patterns that indicate wasted energy or failing hardware.

Rate Design: Granular data allows for more creative and fair rate structures (like peak-time rebates).

These AI-driven strategies can be pivoted to bolster affordability programs. By transcending broad ZIP-code-based assumptions, AI enables utilities to analyze granular usage data at the meter level, providing a precise calculation of an individual household’s actual energy burden.

Identifying Inefficient Energy Drivers

Because AI can isolate the distinct electrical signatures of aging HVAC systems or failing water heaters (two drivers of high energy burdens), utilities can target the root cause of a high energy burden rather than just treating the symptoms.

Segmenting ‘Leaky’ vs. ‘Peaky’ Homes

Not all energy-distressed homes require the same intervention. Machine learning can distinguish between homes with high baseline loads (leaky) that indicate poor insulation and best served by weatherization teams and homes with dramatic spikes during high-rate hours (peaky) that are ideal candidates for automated demand response.

Proactive Arrears Prediction

By correlating load profiles with local weather and gas rates, AI can identify at-risk behavior—such as a household drastically cutting heat during a cold snap. This makes it possible for utilities to offer subsidized rates or payment plans weeks before a missed payment occurs.

Virtual Eligibility Proxies

Consumption patterns can act as a proxy for socioeconomic factors (e.g., high occupancy density). Utilities can use these signals to pre-qualify households for assistance, bypassing the administrative hurdles that often prevent customers from enrolling in programs.

The Analytical Backbone of Modern Affordability

With this foundational intelligence, utilities can deploy programs like distributed energy resources (DERs), virtual power plants (VPPs), and electrification at the scale, cost, and speed required to reach millions of households, protect utility margins, and meet regulatory requirements.

An immediate application of this data is the ability to replace broad marketing outreach with highly accurate propensity modeling. Instead of sending generic weatherization flyers to an entire ZIP code, for example, analytics teams can rank customers by their specific likelihood to benefit from a given intervention. This is designed to ensure that every program dollar is spent where it will have the greatest impact on a customer’s bill while significantly lowering the customer acquisition costof traditional marketing.

Beyond outreach, these insights can allow utilities to de-risk the transition to Time-of-Use (TOU) or EV rates through shadow billing. By simulating how a new rate structure would have affected an income-qualified household based on their actual history, utilities can ensure customers are moved to plans that are designed to deliver savings rather than inadvertently increasing their energy burden. This level of precision extends to demand side management (DSM) and energy efficiency portfolios, where data can identify specific homes that are positioned to provide the most relief to both the resident and the grid.

As these data-driven strategies reduce localized loads through non-wires alternatives (NWAs), utilities can also defer and/or avoid massive capital expenditures on infrastructure upgrades. These avoided costs represent effective ways to prevent future rate hikes that typically fund grid expansion.

Paving the Path Forward with Comprehensive Intelligence

When generic, siloed outreach struggles to reach the customers who need support most, utilities face chronic enrollment shortfalls—even with fully approved budgets sitting ready. To break this cycle, utilities can leverage AI and existing AMI data to transform isolated programs into a synchronized, measurable strategy that moves the industry beyond reactive bill credits toward holistic solutions that can help avoid missed payments before they occur.

References

American Council for an Energy-Efficient Economy. (2022, November 18). Report: Despite progress, low-income households underserved by utilities’ efficiency programs [Press release]. https://www.aceee.org/press-release/2022/11/report-despite-progress-low-income-households-underserved-utilities

S&P Global Market Intelligence. (2026). Record amount of utility rate requests in 2025 amid affordability concerns. https://www.spglobal.com/market-intelligence/en/news-insights/research/2026/01/record-amount-of-utility-rate-requests-in-2025-amid-affordability-concerns

U.S. Department of Energy. (n.d.). Recovery Act: Smart Grid Investment Grant (SGIG) program. https://www.energy.gov/oe/recovery-act-smart-grid-investment-grant-sgig-program

About the Author

Ted Nielson is the Chief Product Officer at Bidgely, leading product strategy, innovation, and AI-driven solutions to empower utilities worldwide. With 15+ years of experience scaling organizations and delivering enterprise-grade products, he combines a builder’s mindset with a strategic, data-driven approach. Ted is passionate about creating impactful customer experiences and driving measurable business outcomes.