STANFORD, CA, May 27, 2022
The road from research to full commercialization for sustainability technologies is long, full of surprises and devoid of certainty.
Amit Narayan with a UC-Berkeley PhD in electrical engineering launched a couple semiconductor startups when he had the epiphany that perhaps the complex software algorithms used in chip design could also be used to optimize the electric grid. After all, the chips, though tiny, carry signals at gigahertz speeds making nano-scale metal wires behave as transmission lines. He broached this idea in 2010 to Abbas El Gamal, a Stanford University professor of electrical engineering and an advisor to one of Narayan’s semiconductor startups. Intrigued, El Gamal invited Narayan to Stanford to explore this further. On May 11, Autogrid, the climate-AI startup that Narayan began working on at Stanford, announced that it is being acquired by Schneider Electric, a Fortune Global 500 company in energy efficiency and sustainability technologies.
Here, Narayan describes his decade-long journey and the road ahead.
Congratulations on the acquisition. How does that make you feel?
It’s a great outcome. When I started the company, my son was in kindergarten. He will start college this fall. So, I feel like two of my kids have graduated at the same time! Schneider is the right place to maximize the impact of our technology. They operate in more than a hundred countries and have many complementary products. Most importantly, they share our values and commitment to decarbonize our planet. From an impact perspective this gives us the opportunity to scale faster.
From the beginning this has been a sustainability issue for you, right?
I vividly recall the day when I was driving my 5-year-old son to school. Stuck in traffic, with all the smoke around him, he started crying afraid that he was going to run out of oxygen. This is when I started thinking about the world he will have to grow up in and the dangers of climate change. I realized that the electricity sector was one of the major sources of carbon emissions, but also the biggest opportunity to transition into a sustainable world. If we could deploy solar and wind, and use this renewable energy to power our cars and homes, we would make a big dent in the problem.
However, I soon realized that the architecture of our electricity grid would need to change. For decades what the utilities did when demand peaked on a hot summer day was fire up these old, fossil-fuel based peaker plants. These peaking generators are 10 times more expensive and four times more polluting than average fossil fuel plants. And more than half of them are located in the most vulnerable communities: low-income neighborhoods and ethnic minorities with limited political power. Part of what we’re trying to do is get rid of these environmentally and ethically unjust peaker plants.
How far into the future do you think an integrated smart energy system will be effective and widely deployed?
I think it is right here, right now. We’re deploying more and more renewables. The wind doesn’t always blow, and the sun doesn’t shine at night. If, for every megawatt of renewable generation, we have to add an equivalent megawatt of a fossil-fuel plant as a backup, it defeats the whole purpose.
Fortunately, there’s a better way. When I started working on this problem, I realized that by utilizing the power of AI and the cloud, we can break the century-old assumption that power can only flow one way. We could harness the flexibility of all energy assets to keep the grid balanced in a cheaper, cleaner, more reliable manner than before. We called these ‘software-defined’ virtual power plants.
I think smart grid will have arrived when I can give my utility some control over my air conditioning and large appliances in exchange for a lower rate and a more reliable system. I don’t have that option.
Most Americans have that option today. In California, we now have over a million EVs. If we can use the batteries of parked EVs, we can back up the entire California grid. And, by renting these EVs, the owners of these vehicles can earn money. It’s like making your car available for the Uber fleet, but unlike Uber you don’t have to drive your car. It can stay at home, plugged into the wall, and can make money by helping stabilize the grid. It’s better than a self-driving car version of Uber!
A lot of utilities are offering this, but they are not very good at marketing it to consumers. Also, many times the incentives are not structured properly. So, the bigger thing is the increased sale of these assets – electric vehicles, solar panels, smart thermostats and appliances. The point of sale is the time to get people into these optimization programs, which can offset the cost of purchasing the devices by 30 to 40 percent.
From the consumer perspective, do you think “set it and forget it” technology will be key?
Absolutely. Nobody has time to think about this on a daily basis. Nor should they. In our view, it should all be seamless, automated, behind the scenes. As long as your car is charged when you need it in the morning, you don’t worry about it.
That’s also where artificial intelligence comes in. It understands your preferences and personalizes the use of your assets so that you don’t notice. You do know that you’re getting 100 percent carbon-free electricity.
How did this journey for you start at – and continue to intersect with – Stanford?
The TomKat Center for Sustainable Energy had just started and was looking to fund its first batch of research projects, specifically for a smarter electricity grid. Stanford was very active in solar, wind, batteries, EVs. But at the time no one was really looking at the whole system.
We identified that the fundamental challenge lies in dealing with the intermittency of renewables. We made the system open and extensible, just like the Internet. We said it doesn’t matter if the asset is an EV or solar panels, we only need to know how much power it can produce or use, and then optimize the entire system. So, (with professors El Gamal, Stephen Boyd, and Benjamin Van Roy, and adjunct professor Daniel O’Neill), we got the research project GridSpice funded by TomKat. I was a visiting professor at the time. The project was about modeling and simulating a smart grid system, how that could work and optimize millions of distributed assets.
Once we showed that it could work, I wanted to build the software for actually controlling assets and optimizing real systems. That was the focus for AutoGrid.
Over the years, we have collaborated closely with several Stanford professors. Steve Eglash, who is now at SLAC but was at Precourt then, introduced me to executives of the City of Palo Alto Utilities, which became our first customer. And the unit that manages the university’s endowment made a small, early venture investment in us, which validated what we were doing in the eyes of follow-on investors. And quite a few Stanford graduates have and continue to work for us.
Stanford gave me the opportunity to enter the energy field when I didn’t know much about the domain. The amount of encouragement and support I received, despite being an outsider to the industry, was remarkable. I really don’t think any other university would have done it. The environment of taking risks, collaborating with industry, connections, freedom to explore is unique.
You can read the Full Interview at : Q&A: Stanford smart grid project launched a decade-long journey for a recently acquired climate-tech startup