Photo credit: Jeff Gritchen / MediaNews Group / Orange County Register via Getty Images
In a stretch of desert southwest of Phoenix, a robotic arm grabs a solar panel off a wooden pallet, lifts it high overhead, and slots it vertically onto a conveyor belt. After the arm positions eight panels in a row, the belt pulls the panels along, while steel tubing travels to meet them. Technicians wait under a shaded awning to drill the pieces together, before they’re loaded onto a truck, which ferries them away to be incorporated into a larger array.
This automated system, which finished its first project in November, is the brainchild of Terabase, a solar company that sets up barebones panel assembly lines right on a project site to make installation easier.
Like many industries, the renewable energy sector is increasingly looking to incorporate technologies like automation and artificial intelligence, which companies see as potential antidotes to challenges like a tight labor supply and high interest rates. Companies like Terabase are particularly interested in how these technologies could help slash soft costs, which include all non-hardware expenses. These have proved resilient for solar projects even as the industry has pushed hardware costs to ever lower levels.
“Solar panel prices are extremely low now,” said Matt Campbell, Terabase’s CEO. “But the soft costs keep going up, not down. And the only way, in my view, to drive soft costs down is through software and robotics.”
In 2023, solar module prices fell by more than 30%. The fall was less pronounced in the United States due to trade duties, but the overall trend has been years of precipitous hardware cost declines. Campbell, a former vice president at SunPower, sees driving down soft costs as the last frontier in utility-scale solar. Today, those costs account for 65% of the cost stack for large projects, and bringing them down poses an intractable challenge.
“How do I make an engineer 10% more efficient? Or how do I make a project manager 20% more competent?” said Campbell. “Those are squishy things. How do I even measure that?”
Many developers think cutting-edge tools can drive development efficiencies, which they say are needed to preserve profits and deploy the record-busting amount of clean energy that fighting climate change requires.
That automated and efficient future, if it’s reached, remains years off. But developers are hopeful that newer technologies like robots, drones, and AI could change how projects are sited, designed, built, and managed.
A growing number of technology-forward startups have hinged their success on the rise of automation.
Terabase hopes its automated installation system ultimately helps to lower a project’s levelized cost of electricity by more than 10%. Thus far, the company says its in-field factory increases “installation productivity” by two times over standard construction, which requires many workers using hand tools spread throughout a large site.
Another startup, Built Robotics, has designed a robot that automates pile driving using GPS and a machine that looks like your average excavator. The company says the machine can drive three times the number of piles in a day that a manual solar crew could. And PrecisionHawk, a tech startup that’s partnered with Equinor, uses drones to inspect wind sites, including offshore, reducing the need for human inspectors to climb tall towers.
But more traditional companies are also flirting with automation. Lightsource BP, one of the largest solar asset owners in the U.S., has experimented with robotic mowers and inspectors. The company’s head of innovation and operational excellence Kevin Christy said using those technologies can be challenging, though, because the tech is evolving at the same time that users like Lightsource are experimenting with how to use it effectively.
Furthermore, certain processes like mowing grass around solar panels already have low-tech, low-labor options available. Lightsource, for example, employs sheep to munch around its panels at certain projects.
Christy sees onsite inspection as a more promising route for automation with robots. At one of its projects in Texas, Lightsource is working with a company called OnSight Robotics that uses AI and an autonomous vehicle to inspect sites for up to 12 hours a day.
“There's nothing that does comprehensive inspections like this bot does,” he said. “It's not economically feasible to do with people. And with the robot we could feasibly do 100% quality control on all of the DC connections, which is pretty stunning.”
Eventually, Christy envisions renewable projects saturated with sensors, gathering data and monitoring equipment in real-time.
“I call it ‘sensors everywhere,’ where basically the number and type of sensors on a project will proliferate over time, as it becomes lower and lower cost to measure more and more points throughout the system,” he said.
With the rise of automation, these systems are likely to create a lot more data. Already, companies are experimenting with software to process that information. And some companies are seeing the potential for software to go one step further, using AI to help drive decision-making.
Solar hardware supplier Nextracker now offers a product that uses machine learning to determine when hail poses a significant enough threat to prompt panels to move from horizontal to nearly vertical. In recent years, hail has become a significant concern for the industry because of the damage it can cause panels, said Venkata Abbaraju, Nextracker’s director of technical product management.
Since 2020, Christy at Lightsource said insurance companies have shifted from the usual all-risk contract to identifying several “sub-risks” in certain geographies for solar project coverage, each with its own deductible. For Lightsource’s largest projects in Texas, the deductible for hail is nearly $13 million, encouraging the company and its peers to find ways to mitigate risk. (Like Nextracker, Lightsource is also automating the panel stowing process, but without using AI.)
Nextracker has rolled out a system that uses digital twin technology — which creates a virtual copy of a physical site — along with onsite data processing and algorithms to change the angle of panels on different rows, allowing them to capture more sun.
Terabase also creates digital twins for certain projects, specifically to manage construction onsite. Drones capture images of the site and AI processes information contained in the images to determine how construction is progressing.
“The software is equally if not more important, in terms of making things more productive and efficient,” said Campbell.
Campbell said Terabase has sold its construction product to five gigawatts of projects that are either underway or already completed. So far, Terabase has partnered with companies including Intersect Power, Leeward Energy, and First Solar.
Still, these applications are novel rather than the norm. All of them are in very early stages, said Abbaraju at Nextracker. And Christy said they’re too new to have yet had a measurable impact.
BloombergNEF solar analyst Jenny Chase said solar is already so cheap that there isn’t much need for these types of technologies. She attributes the interest — particularly in AI — to venture capital “buzzing on the latest buzzword.”
“Solar is infrastructure. It’s not something that can necessarily be revolutionized by changes in software,” she said.
However, she added, soft costs are much higher in the U.S. than the rest of the world, so there may be more potential to cut costs there using new tech. But she also argued that structural shifts, like changes to the permitting system, could be more meaningful.
Still, developers remain animated about the potential for new technologies to change their work.
Abbaraju envisions automation adoption increasing to the point where the technology has a hand to play in more than 50% of the development process. In a couple of years, Christy also expects that automation will be central to more parts of the process. Lightsource’s innovation team’s motto, he said, is “better, faster, cheaper, safer.”
“When we're confronted by higher interest rates, tighter labor supply, higher supply chain costs, longer lead times, and — unfortunately, sadly to say — declining quality and reliability, then we need to find ways to leverage technology to start mitigating some of those issues,” he added.