When all of us drove around with paper maps in our gloveboxes, we never would have dreamed that we’d soon route around traffic in real-time, be able to find an open coffee shop that only adds four minutes to our trip, and never have to pull over and ask for directions again. That is the power of distributed computing and machine learning software. We desperately need that kind of power on the grid. That’s why Utilidata and NVIDIA are working together to build a smart grid chip.

The weak link in the energy transformation. 

We’ve spent decades refining public policies to foster thriving markets for solar, offshore wind, electric vehicles, energy efficient products, and a modern transmission system. While the work is not done, the momentum that has collectively been created in these sectors is powerful. 

Distribution grid modernization does not have that same momentum. Although billions of dollars have been invested in the smart grid, it’s not nearly as smart as it needs to be. The poles and wires that bring power to our homes and businesses remain more analog than digital. Grid operators still have very little real-time visibility or control over the distribution grid. The electric distribution grid is the weak link in the energy transformation. 

Today, we feel the pain of the analog grid in a myriad of ways. Solar interconnection is slow and expensive. Demand management is still a sparsely utilized blunt instrument. Outages take too long to recover from. And when extreme system conditions hit, like the cold snaps in Texas or wildfires in California, rolling blackouts are one of the few real-time tools we have. 

The distribution grid lacks the kind of modern, future-proof solutions it needs to deal with this emerging complexity. The grid modernization solutions the market has developed are predominantly siloed and offer only small improvements, layered on top of hardware-centric core infrastructure. We need much larger leaps forward, and we need them fast.

The skill sets to tackle this challenge. 

Utilidata has refined machine learning grid operations software for the past decade. Today, that software has supervisory control of hundreds of distribution circuits across the country, making the grid more efficient, reliable and clean. Over the last several years, we’ve leveraged smart meters to make that software more distributed and impactful. But we’ve been limited by the processing power and hardware-centricity of existing grid mod solutions. Like trying to run Google Maps on an old Blackberry, the potential benefits are severely limited.

NVIDIA has already brought distributed machine learning to other industries. They invented the graphics processing unit (GPU) for the gaming world. They then figured out how to combine that unparalleled processing power with distributed AI software to transform industries like health care and autonomous driving. Their autonomous driving chip has already driven millions of simulated miles to rapidly get smarter and become the industry standard for the majority of new autonomous vehicles. 

By embedding Utilidata’s grid operations software onto NVIDIA’s processors, we will finally provide the industry its version of Waze or Google Maps. The grid will have situational awareness and real-time controls all the way to the edge, but instead of coffee shops and traffic, it will understand power flow, EV charging demand, solar production, and customer usage profiles. The processing power will allow each node to monitor for anomalies and compare them in real-time against a growing library that will learn the difference between a tree branch, cyber attack, or impending equipment failure.

A grid with this type of computing power could provide instantaneous solar interconnection; seamlessly integrate millions of EVs without expensive grid upgrades; engage continuous demand management with machine-to-machine communications that the customer would authorize but never notice; predict and prevent outages, and offer grid operators the ability to execute surgical, targeted demand shedding during an emergency instead of life-threatening rolling blackouts.  All of these new capabilities would be delivered at a fraction of the cost of hardware-centric alternatives. 

A better future is possible. 

Most other industries, from shipping to manufacturing, have modernized their networks in this way. And for the electric grid, this transformation is essential. A dozen states now have binding laws committing to 100% clean energy in less than 30 years. That cannot be achieved without this level of grid operational sophistication. 

Utilidata and NVIDIA aim to start by embedding the smart grid chip in smart meters. In the U.S. alone, nearly 50 million smart meters will be due for replacement in the next five years. The first rollout of smart meters has generally underwhelmed, and we cannot afford to squander this next opportunity. We are placing powerful computers on the side of every home and business at the dawn of electrification and decarbonization – these meters need a future-proof distributed platform like the smart grid chip.  

Over the next few months, we’ll make more announcements about the development of this product, its use cases, our partner utilities and our ongoing collaboration with regulators and other stakeholders about how we fund and deploy this solution. 

Climate experts agree that the next few years will determine if we can stave off the worst effects of climate change. The distribution grid needs new technology solutions to be ready for this challenge. Utilidata and NVIDIA are working as fast as we can to bring the grid the solution it needs.