The archive · Climate & Energy · Product decision · 2023–2026
Gritt bets AI finally tames construction sites, retrofits equipment to build solar farms
Gritt retrofits construction gear with AI to install solar panels; its CMU-trained founders exited stealth with a $26M Series A led by Obvious Ventures.
Gritt
What the business is
Gritt is a San Francisco startup building 'physical AI' for construction: AI-controlled platforms attached to standard equipment that automate labor-intensive jobs, starting with installing utility-scale solar panels.
Starting capital：$26M Series A led by Obvious Ventures (July 2026); earlier seed from First Round Capital, Climactic, Congruent Ventures and VSC Ventures
How it started
Gritt was founded by Carnegie Mellon-trained roboticists Puneet Puri (CEO) and Vishal Dugar (CTO). Their starting point was the global solar build-out, which is bottlenecked by a limited supply of construction workers, while industrial robots had historically failed in unstructured, chaotic outdoor environments. The founders bet that the latest generation of AI models changed that equation, making robotic manipulation generalizable enough to work on real construction sites.
What happened
Gritt exited stealth on 21 July 2026 with a $26 million Series A led by Obvious Ventures, with Union Square Ventures and Active Impact Investment participating, bringing total funding to about $32 million per TechCrunch. Two systems are deployed in the field; the company says a typical eight-person crew that installs 800 panels a day can install 3,000-4,000 with its systems. Gritt reports contracts to help install 2.8 GW of solar over the next 18 months, customers including three of the top ten US power-construction firms, and a goal of 48 systems within six months. On 25 August 2026, Burns & McDonnell announced a strategic partnership to deploy the technology on utility-scale solar projects after a year of joint evaluation.
How it ended up
Still operating and scaling; no exit. Burns & McDonnell strategic partnership announced 2026-08-25.
Background
Gritt is a San Francisco startup building what it calls physical AI for construction: AI-controlled platforms attached to standard, off-the-shelf equipment - so far rented skidders and robotic arms - that automate the most labor-intensive jobs on building sites. Its first market is utility-scale solar, where its systems unload glass panels, carry them to metal frames and position them with sub-millimeter accuracy for workers to fasten.
The bet is that the latest generation of AI models finally makes robots reliable in the unstructured, chaotic environments of construction sites, and that the same generalizable intelligence can then expand across construction. Gritt's founders point to training speed as evidence: stacking cinder blocks took weeks, while a similar rebar-tying demo took a day on the same software.
Gritt exited stealth on 21 July 2026 with a $26 million Series A led by Obvious Ventures, with Union Square Ventures and Active Impact Investment, bringing total funding to roughly $32 million. It says an eight-person crew installing 800 panels a day can install 3,000-4,000 with its systems, and reports contracts for 2.8 GW of solar over 18 months, with three of the top ten US power-construction firms among customers. In August 2026, Burns & McDonnell announced a strategic partnership to deploy the technology on utility-scale solar projects.
What has to be true
- The solar build-out is labor-constrained: an eight-person crew goes from about 800 panels a day to 3,000-4,000 with Gritt's systems - a 4-5x productivity jump.
- Retrofitting equipment crews already rent avoids the capex and process disruption that sank earlier construction robots.
- New AI models made the approach generalizable: rebar tying was demonstrated in a day after cinder-block stacking took weeks.
- The wedge task is the dullest, most dangerous one - repeatedly lifting roughly 100-lb panels - which also improves worker safety at remote sites.
- Real revenue pipeline before scale: 2.8 GW contracted over 18 months with top-ten US power contractors.
What can be applied
If a market is bottlenecked by labor and robots failed before, revisit it when the underlying AI capability changes - retrofitting familiar gear can be the wedge that makes automation adoptable.
Aftermath
As of 2 September 2026, Gritt is scaling: two systems are in the field, targeting 48 within six months, with a contracted pipeline of 2.8 GW of solar over 18 months. Customers include three of the top ten US power-construction firms, and on 25 August 2026 Burns & McDonnell formalized a partnership after a year of testing at utility-scale sites. Gritt is extending its AI to fasten panels, drill posts and build racks, and later to tasks like tying rebar, positioning itself as a physical-AI layer for infrastructure buildout.
Sources
- Gritt exits stealth with $32 million for robots to build solar plants - then, everything else
- Gritt Robotics raises $26M Series A to build robots for solar plants
- Burns & McDonnell and Gritt Partner to Advance the Future of Solar Construction
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card