Meet the robots being trained to build solar farms faster than any human crew

Key points

• Gritt, a robotics startup founded by two Carnegie Mellon University engineers, exited stealth on Tuesday with a $26 million Series A funding round.
• Total funding now stands at $34 million, with Obvious Ventures leading the latest round alongside Union Square Ventures and Active Impact Investment.
• A typical eight-person crew installs around 800 solar panels per day; working alongside Gritt's systems, the same crew can install 3,000 to 4,000 panels per day.
• Gritt is contracted to help install 2.8 gigawatts of solar capacity, enough to power hundreds of thousands of homes, over the next 18 months.
• Three of the top ten US power construction companies are already Gritt customers.

Solar panels are going up everywhere. Governments and energy companies are racing to build new solar farms, and the demand for skilled installation workers is far outpacing the supply of people willing, or able, to do the job.

Gritt thinks robots can fill that gap.

The startup, co-founded by CEO Puneet Puri and CTO Vishal Dugar, both trained at Carnegie Mellon University, builds AI-controlled systems that handle the dirtiest and most physically punishing parts of a solar installation. Right now, that means one main job: picking up large, heavy glass panels, carrying them across rough ground, and positioning them on metal frames with sub-millimetre accuracy so human workers can bolt them in place.

Here is why that matters. Solar panels weigh around 45 kilograms (roughly 100 pounds) each. Workers lift them overhead, repeatedly, all day. Injuries are common. Getting crews out to remote sites, where many large solar farms are built, is also genuinely hard. One Gritt customer, speaking to TechCrunch AI but not willing to be named publicly, said the system should make remote sites more workable and reduce the number of injuries on his jobs.

Gritt does not build its own physical robots from scratch. Instead it uses off-the-shelf hardware, rented machines like wheeled skid-steer loaders and robotic arms made by companies such as Kawasaki, and controls them with its own AI software. That software is the product. Think of it like an experienced driver who can take the wheel of almost any car, rather than a car manufacturer who builds one vehicle and hopes it fits every road.

Does this put construction workers out of a job?

Not in the near term, and probably not in the way you might fear. The system works alongside existing crews, not instead of them. Workers still fasten the panels; the robots handle the carrying and positioning. The productivity jump, from 800 panels a day to 3,000 to 4,000, means more sites can be built with the same number of people rather than replacing the people already there.

The company has two systems operating in the field right now. It wants 48 deployed within six months. Competitors including Luminous Robotics, Cosmic, and China's Trinabot are building similar machines, but those companies are developing their own hardware. Gritt's bet is that focusing purely on software and renting existing equipment will let it grow faster and at lower cost.

Longer term, Gritt wants its systems to fasten panels, drill posts, build the racks panels sit on, and eventually take on other heavy construction tasks like tying steel reinforcing bars before concrete is poured. Training the system to stack cinder blocks took weeks; using the same software, training it to handle rebar took just one day. That is the part the founders say new AI models made possible.