Oracle has been building a visible AI story in utilities across two fronts: Opower, the long-running behavioral science and engagement platform, and newer infrastructure like Energy and Water Data Intelligence that aims to unify data across the Oracle Utilities portfolio. A press release dated April 13, 2026 gave the clearest public tally yet of Opower’s cumulative scale. This review looks at what Oracle is actually shipping, what is still on the roadmap, and where independent caution is warranted.
Opower at scale: the Oracle-reported figures
Oracle reported in April 2026 that Opower now benefits “nearly 45 million” North American households, with the precise enrolled figure given as 44.6 million residential households cumulative since 2009 through March 2026. Over that period, Oracle reported 3.5 billion personalized communications delivered across channels, with cumulative energy savings of 44.23 TWh and nearly $4.3 billion in cumulative bill savings. Of that bill savings total, Oracle attributed $369 million to calendar year 2025.
These are Oracle-reported figures, not independently audited. Measurement and verification methodologies for behavioral energy programs vary across utilities and regulators, and attribution is difficult to standardize at scale. Utilities filing with an IRP or DSM proceeding should request program-specific measurement and verification documentation, not portfolio-level totals.
What the AI and behavioral science actually do
Opower’s core engine compares a household’s consumption against a peer group of similar homes, then produces a personalized Home Energy Report. The framing draws on behavioral science research around social norms: showing customers that comparable neighbors use less energy consistently produces modest but measurable reductions in a way that generic efficiency appeals do not.
The AI layer sits on top of usage data from the utility’s meter data management system. It segments customers, selects which communications to send (print, email, in-app, digital), calibrates timing, and over time adjusts the program based on response. The system is not replacing the CIS as a system of record. It is consuming usage and account data from the CIS and MDM, running its models, and pushing outbound communications. That architecture matters: Opower adds engagement capability to an existing CIS, it does not substitute for one.
For a broader view of what Oracle’s CIS portfolio looks like as a foundation for tools like this, see the Oracle Utilities overview and the Oracle vs SAP comparison.
Energy and Water Data Intelligence: what is live vs. what is roadmap
Oracle announced Energy and Water Data Intelligence on September 11, 2024 as a data unification and AI acceleration layer. At launch, it was pre-integrated with Oracle Utilities Customer Cloud Service and Customer To Meter NOW.
The announcement also noted planned integrations with Work and Asset Cloud Service (WACS), DERMS, and Opower as roadmap items. Those integrations were not described as shipped at launch. Buyers evaluating Energy and Water Data Intelligence should verify current integration status directly with Oracle rather than treating all four integrations as generally available.
The intent is practical: utilities on Oracle Utilities products accumulate usage data, meter events, work orders, and customer records across systems that have historically required bespoke integrations to join together. Energy and Water Data Intelligence is Oracle’s answer to that fragmentation, providing a pre-built data layer so AI models have a unified input.
AI in the contact center and in meter data management
Two additional Oracle AI announcements from 2025 are worth noting for utilities on the Oracle Utilities Customer Platform.
In May 2025, Oracle added AI call summarization and tagging to Oracle Utilities Customer Platform at no extra cost. After each customer service call, the AI produces a structured summary and applies tags, reducing the time agents spend on after-call work and creating a more consistent record for quality review. This is an internal-facing feature, not a customer-facing one.
In June 2025, Oracle announced AI anomaly detection inside Meter Data Management targeted at cutting billing exceptions and truck rolls. The proposition is straightforward: meter reads that fall outside normal patterns get flagged automatically before billing runs, reducing the manual review workload on the CIS billing team and avoiding estimated bills that generate customer calls.
Taken together, these two capabilities address two well-known cost drivers in utility customer operations: after-call handle time in the contact center, and exception-driven billing work in MDM. Neither requires a new CIS implementation; both are incremental additions to an existing Oracle Utilities footprint.
An honest independent view
Oracle’s Opower footprint is real and large. The cumulative household counts, TWh saved, and bill savings figures in the April 2026 press release represent a genuine track record across dozens of North American utilities over 17 years. The caveat stands: these are Oracle-reported, program-level aggregates, and the underlying measurement and verification methodology is not standardized across all utility programs in the portfolio.
For utilities already on Oracle Utilities products, the newer AI additions (Energy and Water Data Intelligence, call summarization, MDM anomaly detection) are incremental expansions of capability on an existing platform rather than architectural shifts. They lower the integration cost of adopting AI because the data plumbing is already in place.
For utilities not yet on Oracle, these AI capabilities are not a standalone reason to choose Oracle over alternatives. The CIS selection decision belongs at a different layer, as covered in the full Oracle Utilities platform review and the side-by-side SAP and Oracle comparison. The AI layer is most valuable when the foundational CIS data is clean, current, and well-integrated with MDM.