By: Luis Reyes
Published: Jun 26, at 4:30pm ET
Solar panels have gotten almost embarrassingly cheap. A module that cost a small fortune fifteen years ago now sells for a fraction of that, and the price keeps sliding downhill every year.
So you would figure the hard part of building a solar farm is long since sorted. It isn’t. The slow, expensive, back-wrecking part of utility-scale solar was never the panel itself. It’s getting a few million of them clamped onto steel tubes, by hand, in the kind of desert heat where finding crews is half the battle.
And in California, a fleet of four robots has now done exactly that at a scale nobody had pulled off before: 100 megawatts of solar, installed at better than one panel a minute, with the heavy lifting taken off human hands.
The machine is called Maximo, and the friendly name is doing some work, because the crews on site reportedly treat it like a coworker rather than a threat. It was incubated by the utility AES, runs now as a construction service rather than a product you buy, and this spring it crossed the line that separates a science project from a real job.
The 100-megawatt run happened at AES’s Bellefield complex in Kern County, on the dry southern end of California’s Central Valley, and it’s the clearest sign yet that robotic panel installation has graduated from “one robot doing a demo” to a coordinated fleet finishing utility-scale work.
Here’s what actually went down at Bellefield. The setup grew from a single machine to four Maximo units working in parallel, each one an industrial six-axis robotic arm bolted to a tracked platform that crawls along the rows. It runs off its own battery microgrid, which matters more than it sounds, since a half-built solar site doesn’t have grid power to plug into yet.
A skid steer drops a pallet of panels into the back of the unit. The arm picks them up one at a time, lines each one against the steel torque tube using AI vision, and either sets it in place for a worker to clamp down or does the whole installation start to finish on its own.
To understand why that’s a big deal, you have to picture the version a human does. Deise Yumi Asami, who founded Maximo, told The Robot Report that installing panels onto the tall torque tubes normally takes “three people on ladders on uneven ground.” The robot does it from the dirt, on tracks, in the dust and wind that she said you simply can’t control on a desert site.
No spam. Unsubscribe anytime. Privacy policy (opens in new window)
AES and Maximo describe the 100-megawatt result as the moment robotic module installation shifted from early validation to sustained commercial production, which in plain reading means it stopped being a pilot.
The numbers underneath are the part that holds up. Version 3.0 of the system consistently placed more than one module per minute, and crews running the robots installed as many as 24 modules per shift hour per person, close to double what traditional methods manage in that region.
Nick Hegeman, Maximo’s chief commercial officer, put the consistency this way to PV Tech: with the robotic system you know you’re getting 400 to 500 modules per shift, every shift, instead of whatever a given crew happens to produce on a given day in the heat.
Why does a robot that places panels matter more than it sounds? Because a utility-scale solar module is nothing like the panel on your neighbor’s roof. It’s a slab of glass and aluminum that can weigh 60 to 80 pounds, and a single installer might lift, carry, and set a few hundred of them in one shift, bending and reaching each time, usually under a sun that makes the work miserable and the labor scarce.
Do that across the buildout the country has lined up, where market analysts expect the U.S. to deploy hundreds of gigawatts of new solar this decade, and the thing rationing how fast solar gets built stops being panels or money. It becomes human backs and the number of people willing to spend a career using them this way.
That’s not a hypothetical shortage. The International Energy Agency has projected that annual solar additions need to roughly triple and the solar workforce nearly double by 2035 to keep up with demand, a lot of it driven by the same data-center boom now eating electricity faster than anyone planned for.
It’s the same demand wave that has California leaning hard on grid-scale batteries to cover the evening ramp, and that has China stringing a 250-mile wall of panels across a desert. Everybody wants more solar in the ground tomorrow. The chokepoint is the unglamorous task of physically getting it there.
It’s also not the only company chasing the problem, which is the part that should keep this from reading like one firm’s press release. Out of the same California solar world, Terabase Energy has been building a system it calls Terafab, and it attacks the bottleneck from a completely different direction.
Instead of one arm placing panels out in the field, Terabase erects a hardened assembly line right on the construction site, runs panels and steel torque tubes through an automated inspection-and-build process, then loads the finished sections onto unmanned rovers that drive them out into the array.
The second-generation version, Terafab V2, has finished field testing and is cleared for commercial use, the company told PV Magazine USA, financed largely by a $130 million round led by SoftBank. CEO Matt Campbell summed up the company’s entire reason for existing in one line: to build solar “10 times faster for half the cost.”
The catch is a useful one. Terabase’s rovers still hand the assemblies off to human crews who unload them and bolt them onto pre-placed mounts, and full automation of that last step isn’t due until 2027. Which is the honest state of solar robotics right now. The machines have taken over the carrying and the assembly, but a person is usually still in the loop at the exact moment a panel meets its mount.
A few caveats are worth keeping straight, because the marketing tends to blur them. First, AES has been consistent that Maximo isn’t there to cut headcount.
The company says its staffing on job sites hasn’t changed since the robot showed up, and the pitch is that automation widens the door to solar work for people who can’t or won’t spend years lifting 70-pound panels in the heat. Whether that holds as fleets scale up is a fair thing to keep an eye on, but no one has been laid off for a robot so far.
Second, the “half the time and half the cost” figure that follows Maximo around is AES’s own claim, in the company’s materials since the robot launched back in 2024, not an independently audited result out of Bellefield.
What Bellefield actually demonstrated, in numbers the company put on the record, is throughput: better than a panel a minute, and crews placing close to double what conventional methods manage in that region. That’s a real, measurable jump. It just isn’t the same thing as a verified 50 percent cost cut, and the two are worth not blending together.
Third, this ran on a lot of borrowed brainpower. Maximo built and tested its fleet in simulation using NVIDIA’s Omniverse and Isaac Sim before pushing updates to the actual robots, and Amazon Web Services handled the data and software end.
Amazon is the offtaker for Bellefield’s output too, so the panels these robots installed feed Amazon data centers, which gives the company an interest on both sides of the deal. The next version, v4.0, is the one Maximo says will build on what version 3.0 pulled off at Bellefield. As of now, it hasn’t shipped.
So the headline version, that robots are building solar farms now, is true with an asterisk. Four machines really did clamp down 100 megawatts of panels in a California desert at better than a panel a minute, and it’s the first time that’s been done at scale instead of as a demo. What they automated wasn’t the clever part of solar.
It was the dumb, heavy, repetitive part that wears out crews and drags on schedules, which happens to be the exact part standing between the U.S. and the hundreds of gigawatts it keeps saying it wants. Whether Maximo and Terafab end up as standard kit on every site or stay a niche for the biggest desert jobs comes down to the next versions and whether the companies actually building these farms decide a robot arm pencils out cheaper than another crew.
Right now, the robots have the worst job on the site, and the people get to keep their backs.
Don’t bite your tongue. Speak up.
Olivia Richman · Jun 26, 2026
Luis Reyes · Jun 8, 2026
Luis Reyes · Jun 4, 2026
Luis Reyes · Jun 24, 2026
Luis Reyes · Jun 7, 2026
Luis Reyes · Jun 1, 2026
Luis Reyes · Jun 26, 2026
Luis Reyes · Jun 26, 2026
Luis Reyes · Jun 26, 2026
Olivia Richman · Jun 26, 2026
Luis Reyes · Jun 26, 2026
Autonotion is the English-language automotive editorial by Autonocion.com — car news, reviews, and industry analysis for American readers.
Other links
Company
Subscribe
Get the latest car news in your inbox:
By submitting your email you allow autonocion.com to send you news or promotions. More info

source