These small sensors can detect fires at utility-scale solar sites before they spread

As utility-scale solar proliferates across the United States, project operators must contend with monitoring the landmass hosting these sprawling PV arrays. The footprint of grid-tied solar is increasing, and, separate to that, so is the spread of wildfires stoked by a warming climate, consuming acres of land in regions experiencing historically less precipitation.

In 2024, nearly 9 million acres of land was consumed by 64,897 wildfires in the United States, a staggering increase from 2.69 million acres in 2023, according to National Interagency Fire Center (NIFC) data. As of March 28, the NIFC reported that 707,228 acres have been burned in 2025 alone, which is above average from prior years.

Solar projects have become critical energy assets for the electrical grid, and, given their scale, it’s unrealistic for array technicians to monitor the hundreds of acres these power plants occupy without assistance. The earlier a fire is addressed here, the more likely it is to be prevented.

Torch, a monitoring technology startup based in San Francisco, has developed a remote sensor by the same name capable of detecting fires over large tracts of land, including land under solar modules.

“Since the PG&E wildfire fallout, we’ve noticed a major shift in how seriously solar operators are taking site-level fire risk and liability exposure,” said Vasya Tremsin, CEO and inventor of Torch.

SPW spoke to Tremsin and Torch COO Jo Morris about the origin of the technology and why the company is pushing the solar industry to care about wildfire prevention.

This interview has been edited for brevity.

SPW: Did I hear you correctly that Torch started as your high school science fair project?

Tremsin: Yep. Basically, I grew up in a family of PhD scientists. Since the eighth grade, I’ve been participating in these science fair competitions and doing projects together with my dad. We would build certain devices that would solve some kind of real-world problem, like a device for people with hearing disabilities, a device for soil moisture during drought. In 2017, there was the crazy Napa Valley wildfire that basically burned through half of California, and I’m from the East Bay area, so it was very close to me.

I saw that and was like, why do we always detect these fires too late when it’s already impossible to stop it? That’s when the idea started. I did that project, kind of a first prototype, very basic science. Then a couple years later, some of the judges at the science fair introduced me to some investors in Silicon Valley and I basically co-founded the company with my dad.

We worked on a lot of the R&D processes, like testing development; went to a lot of prescribed burns; burnt up big piles of wood to just keep testing and testing. Jo joined us as a late co-founder a couple years ago, and we finally developed the technology and we’re selling to a lot of different use cases.

SPW: Broadly speaking, how do Torch sensors work?

Tremsin: What we’ve built is a solar-powered sensor that combines a lot of different modalities in one device. You don’t need any outside power, you don’t need any Wi-Fi, Bluetooth, none of that stuff. It just connects with one click of a button to radio networks, either through the local radio networks or through a gateway. We have these different variables, such as infrared imaging, optical imaging, spectral analysis, gas sensing, temperature sensing and humidity sensing, all packaged in one device. Then we have these pretty complex back-end algorithms that are analyzing all of this data and combining it to see or smell or feel a fire within basically three minutes of ignition.

It’s a very important and especially good use case for the solar industry, specifically because we are able to look under the panels. We can go below the panels and look across the entire field, whereas if you put a traditional security camera or some kind of satellite or something, you would just be completely blocked by the panels. When a fire starts there, usually it’s a spark that happens because the solar panel overheats, falls onto the ground and the vegetation causes a brush fire.

SPW: Is there a specific reason why it’s three minutes? Is that a crucial time in terms of a wildfire starting?

Tremsin: In terms of remote monitoring, when you have a solar farm far away, where the land is less expensive, no one is there, no one sees anything. Then a fire happens, and nobody sees it for hours sometimes, and it spreads to acres and there’s a huge plume of smoke, and at that point someone comes. If you can decrease that by multiple-times and do it within minutes, even under 10 minutes, you can immediately come and put it out. In the kitchen, the chef puts out a fire. They don’t call in a firefighter, because the fire is still so small.

So, the reason we say three minutes is because our devices are very low-powered. They actually go in this complex system of “sleep mode” and “wake up time” to save power and to last basically indefinitely, just on solar power with no outside power. The most amount of time they sleep is three minutes.

SPW: What U.S. regions are you targeting to deploy Torch?

Morris: Taking a step back, we’ve created this really cool piece of technology, and I think we felt pull outside of utilities, also as a huge consumable with government for a variety of use cases. But where we found most velocity is with solar, and solar that typically has some kind of fire fuel, so vegetation, and that’s typically outside of California. Our first deployments were on the East Coast. We’re expanding to these 66 sites in and around the East Coast.

We’re not solar experts, but we’ve learned through this one partnership, and we’re also learning from others, including advisors that are coming on that work in the space.

SPW: Why has solar become one of your focus areas for deploying Torch sensors?

Tremsin: All of these potential customers that we’ve talked to say that it sometimes takes hours to detect a fire, and it’s up to three acres, four acres, 10 acres of fire by the time anyone notices. We know that every 10 minutes of that fire spreading is literally millions of dollars of these solar assets burning down and getting destroyed — the panels, the tracker systems, the whole package. It’s really important to save every minute.

SPW: Since being deployed, have your systems detected and prevented any fires at this point?

Morris: I’m glad to say, no. We’re still quite early on in this journey. We typically do live demos where we put a fire in the middle of sensors that are placed around at different lengths and show the thing working.

SPW: How much land mass can a single sensor account for?

Morris: The scientific claim is that a small, 2-by-2-m fire, we can detect up to 113 m. Given that we have a 360° view, that’s a 10-acre radius around a point. We know, roughly, that’s about $2 million worth of solar assets.