Grid edge

Aclara joins Nvidia and Utilidata in bringing AI to the smart meter

Utility customers will deploy meter-embedded AI modules at scale next year, while pilot programs will continue to use Utilidata’s “meter collar” add-ons.

March 12, 2024
Listen to the episode on:

Photo credit: John Paraskevas / Newsday RM via Getty Images

Photo credit: John Paraskevas / Newsday RM via Getty Images

Utilidata and Nvidia today announced their inaugural partner for their grid-edge module collaboration. Aclara Technologies — acquired by electronics giant Hubbell Inc. in 2017 — will be the first meter maker to embed the technology, which combines Nvidia’s chip and Jetson platform with Utilidata’s distributed AI software. 

  • The top line: The announcement marks the first hardware partnership for Utilidata’s Nvidia-powered distributed AI platform, Karman, and will ultimately provide real-time insights into what is happening on the grid. Utilities tend to be reticent about betting on new products and entrants, Aclara’s new meter has the experience of well-established players in hardware, chips, and artificial intelligence behind it — players that have been actively aligning themselves with influential companies and organizations in the utility sector.
  • The market grounding: Smart meters are hardly new to the grid. In 2009, nearly $8 billion in funding went to implement advanced metering infrastructure in the U.S., leading to the installation of more than 100 million smart meters. While those initial AMI rollouts received mixed reviews, a new generation of meters that integrate advanced technology like AI is beginning to be deployed. In 2019, for instance, Landis+Gyr and Sense Energy Monitoring partnered to put AI in smart meters, which are still being rolled out in the Northeast. Utilidata estimates that around 50 million meters in the country are nearing the end of their useful lifetimes, which means utilities are once again looking to replace them.
  • The current take: Utilidata president and COO Jess Melanson said that deployment of the meter-embedded chip will open up new applications of AI for utilities without pushing them too far out of their comfort zones: “This is a really significant leap forward in terms of compute, in terms of AI capabilities, and in terms of a company like Nvidia turning their focus to the electric grid,” he said. “[Utilities] can buy AI products from AI experts and hardware products from hardware experts, and have a really integrated, effective solution.”

Nvidia and Utilidata announced their partnership almost exactly a year ago, promising to develop and deploy a “lower-cost, lower-power module to deliver real-time AI at the edge of the grid.” The pair’s chip uses a combination of on-chip and cloud-based software, plus distributed AI, to collect real-time data and manage distributed energy resources.

Several thousand units of the first product to stem from that partnership — a “meter collar” that can be added to existing meters — are set to be deployed this year. 

But for utility customers, Melanson said, there are some major benefits of a combined electric meter and distributed AI module. That’s where the embedded version of Karman comes in, via an implementation with Aclara’s existing residential meter, which the companies say will provide utilities with 100 times more processing power than traditional solutions.

“Utilities did say they appreciated the flexibility that the collar gave them in terms of doing a deployment without having to think about billing,” Melanson told Latitude Media. “But at scale, as we start thinking about getting into 10% of the endpoints or more at a utility, there seemed to be a preference for the meter-embedded version.”

Beyond pilot programs

Melanson said Utilidata is currently in discussions with existing Karman customers, including Portland General Electric, ComEd, and Duquesne Light Company, about which product they’ll deploy. Those three companies received awards from the Department of Energy in October, in part to fund the deployment of over 150,000 Karman units.

Generally speaking, though, the meter-embedded product is for scaled projects, while early-stage deployments will continue with the meter adaptor.

The partnership with Aclara isn’t exclusive on either side, which, according to Melanson, reflects Utilidata’s desire to give utilities more choice: “Utilities should be able to pick their favorite hardware and their favorite distributed AI platform, and those may well come from different companies,” he said.

Nor is the Aclara meter the end of the road for the partnership. Melanson said Utilidata is currently in discussions to embed the module in other types of hardware in Hubbell’s massive portfolio, which, in addition to smart meters, includes everything from electric vehicle charging products and line construction hardware to demand response units and panelboards.

And while many utilities have been dipping a toe into AI-powered pilots that range from virtual power plant deployments to EV charging analytics, Melanson said the installation of Utilidata’s module will help to ensure utilities don’t have to spend the first few months of a pilot organizing and filling in missing data points. Instead, they’ll be able to focus on getting better at using that data — including by applying machine learning and AI models.

“Those are huge foundational capabilities that will drive better outcomes through all those projects, and accelerate the speed, so we don’t get stuck in pilot cycles for too long,” Melanson said. “We are all on the clock — the changes on the grid are underway and moving at a consumer-driven and market-driven pace.” 

Today’s innovations therefore have to focus on enabling utilities to move more quickly than they typically have, and a team-up that combines and packages the skills of three known, trusted companies, targets that exact need, he added: “To me, that’s the huge opportunity for acceleration here, accelerating the speed with which we’re getting experts doing their thing on the electric grid.”

Listen to the episode on:
No items found.
No items found.
No items found.
No items found.
No items found.