Planted, an Oakland company that builds solar power plants using its own construction robots, has raised $31.8 million to expand its fleet and launch a next-generation machine. The round matters because it bets that the real bottleneck in solar deployment is not panel efficiency or the interconnection queue - it is the speed of field construction.
Piva Capital and RA Capital Management Planetary Health co-led the round, announced on 15 September. Breakthrough Energy Ventures, Gigascale Capital, Google, and Khosla Ventures also participated. The company frames its product not as a solar array but as a single system combining planning software, high-density terrain-following racking, storage, and autonomous field robots - an attempt to replace bespoke, site-by-site project delivery with something closer to a production line.
The speed argument
Chief executive Eric Brown put the thesis in plain terms: "Globally, solar needs to be deployed about five times faster than it is today to hit the scale experts project we'll need by 2050, and the gap is growing as the industry confronts constraints involving supply chains, land, interconnection, and skilled labor. The way to close that gap is to turn the construction of a power plant into a manufacturing problem: building robots and deployment systems, not just projects."
Planted's most recent build, a 28 MW behind-the-meter installation serving a neocloud data center, went from first call to power in ten months, with field construction completed in under three. The company also completed Bowes Solar, an 11 MW community-solar project in Illinois for Cultivate Power. Its next job, Aligned Climate Capital's Armoracia project, will sit on 16 acres - 10 fewer than the original design required.
Gigascale Capital founding partner Mike Schroepfer reinforced the manufacturing analogy: "When industries go from bespoke to manufactured, they unlock orders of magnitude more scale at a fraction of the cost. Solar construction is still building custom projects, site by site. Planted's model is the engine this industry needs, and it's already proven in the field."
What the numbers mean for construction
Planted says its arrays can be built on slopes of up to 27% without grading. That figure should interest anyone who has priced earthwork on a marginal parcel. Grading is often the line item that makes a site uneconomic, and a system that tolerates that slope is a claim about which land becomes buildable - a question upstream of everything else on the project.
The company deployed more than 10 MW in 2025 and says it is on track for 100 MW in 2026, against a project pipeline of more than 20 GW. Its existing fleet is fully committed through 2027. The new capital funds fleet expansion and the launch of Sage, a next-generation robot entering the field in late 2026 that the company says more than doubles field productivity compared with its current machines.
Piva's managing partner Ricardo Angel named the constraints directly: "By combining intelligent planning and project development software, high-density racking, and autonomous field robotics, Planted dramatically reduces the land, steel, and labor bottlenecks slowing projects down." Land, steel, and labor are construction constraints, not generation constraints. That is where Planted has pointed its robots.
The robotics play is more conservative than it sounds
Planted's machines work on repetitive, largely obstacle-free sites, installing identical components in rows. That is one of the more tractable environments in outdoor construction - repetitive does not mean flat, given the 27% slope claim, but it does mean structured. Compare this to autonomous excavators working live earthmoving sites with unstructured terrain and crews moving through them. Choosing the easier half of the problem first is a reasonable strategy, not a shortcoming.
The demand backdrop reinforces the bet. Data-center construction has been a dominant story in real estate and infrastructure this year. OpenSpace has documented 1,000 data-center projects, and Buildots raised $130 million last week on a customer list weighted toward that buildout. Planted's most recent named project serves a data center as well. The capital chasing construction AI is increasingly chasing the same underlying event.
Caveats to keep in view
Every operational figure - the ten-month timeline, the sub-three-month field build, the 20 GW pipeline, the 27% slope tolerance, Sage's productivity gain - comes from Planted's announcement. None has been independently verified, and no customer has published a matching account.
A pipeline is not a backlog. "More than 20 GW" describes projects Planted has identified, not projects it has contracted. The release does not break the figure down. Google is named as an investor only; the announcement describes no commercial agreement. And "fully committed through 2027" is a capacity statement that cannot be sized without knowing the fleet's absolute throughput.
Why this matters for real estate and construction
For developers, general contractors, and anyone managing horizontal construction, Planted's model tests a proposition that matters beyond solar: whether repetitive outdoor installation work can be productized rather than project-managed. If Sage ships on time and the 100 MW year lands, the company will have a second data point on a cost curve - the minimum needed to argue that solar construction is genuinely becoming a manufacturing problem. If it slips, Planted is a well-funded EPC with unusually good software. Either outcome tells the industry something about how fast robotics can compress field timelines. Professionals tracking the intersection of construction and automation may find relevant context in AI for Real Estate Courses.
Your membership also unlocks:
