Gritt exits stealth with $34M for AI solar-install robots

The interesting thing about Gritt's stealth exit is not the funding number, it's what the founders think construction robotics is now ready to do. According to TechCrunch's reporting, the Carnegie Mellon-founded startup closed a $26 million Series A led by Obvious Ventures, with Union Square Ventures and Active Impact Investment joining, bringing total funding to $34 million.

The core claim is a big productivity delta on one narrow task. An eight-person crew installs about 800 panels a day the traditional way. With Gritt's systems handling the unloading and placement, that same crew is reportedly doing 3,000 to 4,000 panels a day, with sub-millimeter accuracy onto the installation frames. The hardware is deliberately unromantic, rented skidders paired with off-the-shelf Kawasaki robotic arms, and the proprietary bit sits in the AI models that drive them.

Why it matters right now: utility-scale solar buildout in the US is constrained less by panels or capital than by crews who can actually put modules on the ground. Gritt already holds contracts for 2.8 gigawatts of installation over 18 months, reportedly serving three of the top 10 US power construction companies. If those numbers hold, the company is not selling a demo, it is selling to the incumbents who otherwise have to hire and train field labor at a pace they cannot sustain.

The honest caveat is that only two systems are actually operating in the field today. Founders Puneet Puri and Vishal Dugar plan to have 48 deployed within six months, which is a steep ramp for a hardware-in-the-loop business. What the reporting does not give you is installed cost per watt, uptime in bad weather, or what the unit economics of a rented-skidder plus Kawasaki-arm setup actually look like once you strip out the Series A subsidy. Those are the numbers that decide whether this is a durable business or a well-funded pilot.

The forward-looking piece is the founders' pitch that panel-laying is only the first task. Their stated roadmap extends to fastening panels, drilling posts, tying rebar, and building racks, essentially the argument that modern AI is finally general enough to attack construction task by task from the same base of rented commodity hardware. If that generalization holds up outside solar, the customers who benefit first are the same construction majors who are already writing them checks.