Solar construction is not short only on capital or panels. It is also short on crews who can do heavy, repetitive work fast enough.
Gritt came out of stealth with a narrower claim than many humanoid startups: put an AI control layer on ordinary construction equipment and robotic arms, then use it to unload, move and position solar panels on real job sites.
TechCrunch reported that Gritt raised a $26 million Series A led by Obvious Ventures, bringing total funding to $34 million. The funding is useful context, but the larger signal is that the system is already being measured against construction throughput.
Start with the crew math
CEO Puneet Puri said construction needs intelligence that can work in outdoor, chaotic environments.
The reported operating benchmark is concrete: a typical eight-person crew installs about 800 solar panels a day. With Gritt's system assisting the same crew, daily output can reach 3,000 to 4,000 panels. Gritt's own site gives a similar frame, saying live jobsites have reached 4x output without added labor.
This does not mean the robot replaces the whole crew. Workers still fasten, inspect and coordinate the site. The machine takes on the heavy transport and precise placement step first.
Borrow the machines, train the AI
Gritt is not starting with a fully custom robot fleet. It uses rented construction vehicles and commercial robotic arms, then controls them with its own models. The first task is to unload large glass panels, carry them to the racking, and place them with sub-millimeter precision.
That choice lowers the adoption barrier. Solar builders already rent equipment; adding a control layer is easier than buying an unfamiliar fleet.
Gritt says its machines capture terabytes of field data per day, have worked across more than seven jobsites, and support pick-and-place, transport, assembly, verification and planning. The company also says it has installed tens of thousands of modules with zero breakage.
The contract is the hard part
The strongest commercial marker is a contract pipeline covering 2.8GW of solar installations over the next 18 months. TechCrunch also reported that Gritt customers include three of the top 10 U.S. power construction companies and that the startup wants 48 systems operating within six months.
At that scale, demo videos matter less than reliability. Weather, mud, delayed materials, safety incidents and machine downtime all decide whether a solar farm is delivered on schedule.
A race beyond humanoids
Gritt is entering a field that already includes Luminous Robotics, Cosmic Robotics, Trinabot and Built Robotics. Some competitors build more specialized machines. Gritt is betting on ordinary equipment plus a reusable physical AI stack.
For Chinese readers, the useful comparison is not another humanoid launch. It is the possibility that physical AI earns its first durable revenue in infrastructure jobs where the value of automation is visible in crew size, injury risk, daily output and unit installation cost.
The next checkpoints are practical: whether 48 systems come online, how much of the 2.8GW pipeline is publicly delivered, whether the 3,000-to-4,000-panel daily rate holds across different terrains, and whether breakage, downtime and safety metrics are disclosed.
Sources: TechCrunch, Gritt official materials, MassCEC/Luminous Robotics, CocoLoop, Canary Media; checked the $26 million Series A and $34 million total funding, eight-person crew output of 800 panels versus 3,000 to 4,000 with Gritt assistance, 2.8GW over 18 months, more than seven jobsites, 4x output, and competing solar-robot approaches.