Discussions about AI and the power grid often emphasize the strain that emerging technologies will place on the nation’s aging energy infrastructure. However, Josh Brumberger, CEO of Utilidata, envisions AI as a tool to enhance grid management. Utilidata, based in Rhode Island, is developing “edge AI intelligence, ” embedding AI chipsets designed by Nvidia into smart meters and grid control devices. These devices can process vast amounts of data rapidly, enabling utilities to manage increasingly complex power grids more effectively. On Tuesday, Utilidata announced $60. 3 million in funding to scale production and deployment of this technology with expanding utility partnerships. This brings the company’s total venture funding to $126. 5 million. The round was led by Renown Capital Partners and included existing investor Keyframe Capital, Nvidia, and Quanta Services, a major grid and energy infrastructure firm. Brumberger remarked, “We want to make it easy for hardware manufacturers to embed AI and distributed intelligence into their devices. AI will be crucial for next-generation infrastructure like the power grid; a few years ago, we were on the edges of these discussions, now we are central. ” Utilidata and Nvidia began collaborating in 2021 and launched a consortium the following year with U. S. utilities and Sunrun, a leading residential solar and battery installer, to support deployment. Since then, their technology has been selected for several advanced utility projects. In 2023, Portland General Electric received a Department of Energy (DOE) grant to use Utilidata and Nvidia’s “grid-edge computing platform” to integrate batteries, EVs, and community solar. Pennsylvania’s Duquesne Light won a similar DOE grant to use these devices for data collection to better handle climate-induced grid threats. Consumers Energy in Michigan started a project with Utilidata to analyze the effects of home EV charging on the grid. These projects share key features: collecting massive data streams such as subsecond voltage and frequency readings, processing this data through integrated circuits executing complex calculations, then enabling grid operators to make informed decisions. The field devices must be reprogrammable to handle diverse and evolving tasks autonomously—something traditional, slower, centralized control systems cannot achieve. Utilidata, originally focused on grid voltage control equipment, pivoted in 2020 to address these “grid-edge” challenges, developing faster versions of older utility tech that lacked the speed for modern grid demands. For example, over 100 million smart meters in the U. S. deployed since the mid-2000s collect basic billing data and outage alerts but rely on AMI 1. 0 chipsets with limited, predefined capabilities designed for low cost and mass deployment.

Meanwhile, chip technology has since improved. Leading vendors like Itron and Landis+Gyr are enhancing AMI 2. 0 systems for more complex functions. Utilidata and Nvidia’s “Karman” platform claims superior capabilities, albeit likely at a higher price, reflecting typical industry variability in pricing. Utility adoption is slow; early pilot projects from 2022 embedded Nvidia chips on existing meters. In 2023, Utilidata partnered with Aclara, a smart meter manufacturer owned by Hubbell, to develop integrated meters using Karman. Current deployments won’t immediately replace all existing meters but target strategic grid areas needing granular data, especially to assess impacts of distributed energy resources (DERs) like rooftop solar, batteries, and EV chargers. Utilities increasingly seek to leverage these devices to relieve grid stress via coordinated actions such as storing solar energy midday or scheduling EV charging to avoid infrastructure overloads. Virtual power plants (VPPs) and distributed energy resource management systems (DERMS) can manage DERs through centralized control via wired or wireless communication, but advanced tasks demand local computations and real-time device-to-device communications—precisely what Karman enables. Brumberger stated, “For VPPs as reliable and large as gas-fired plants, accelerated computing and AI are essential to unlock the untapped value of disparate resources. ” Portland General Electric, planning to rely heavily on DERs, views technologies like Karman as crucial to understanding VPP reliability. Senior manager Ananth Sundaram noted such platforms help analyze grid services, power disaggregation, and customer behavior, creating a solid foundation while harvesting crucial grid-edge data. Utilidata hopes utilities and regulators consider these future needs in upcoming large-scale smart meter rollouts. Quanta, a new investor, plays a significant role in large utility infrastructure projects. However, these deployments rely on DOE grants authorized under the Biden administration. The Trump administration has paused and threatened funding and workforce reductions affecting these programs. Brumberger affirmed no cancellations yet but recognizes challenges ahead. The new funding will enable Utilidata to expand internationally and beyond utilities, exploring applications like embedding Nvidia-enabled energy control tech within data centers to monitor and manage power use at the server level. Brumberger remarked that data centers, viewed as powerful microgrids, increasingly integrate generators, batteries, and energy controls. Additionally, Utilidata and Nvidia’s platforms will analyze and “train” on collected data locally, akin to how large language models learn from massive text and image datasets. For example, they can differentiate voltage signature patterns caused by household appliance use versus external disturbances like tree branches. “This isn’t just a sensor, but a local activity hub for training, reducing the need to transmit all data offsite, ” Brumberger explained. The critical question is the proportion of a utility’s territory needing such advanced technology—whether 10%, 50%, 80%, or 100%.