The AES Corporation, through Maximo (its incubated solar robotics company) · Energy and utilities (utility-scale solar construction) · United States (Bellefield, Kern County, California)
AES and Maximo: solar installation robots developed in Omniverse and Isaac Sim
Maximo, the solar robotics company AES incubated, reported 100 MW of panels installed by a fleet of four robots at AES's Bellefield complex in March 2026. It credits NVIDIA Omniverse libraries and Isaac Sim with letting it test robot updates in simulation before sending them to the field.
- Status
- In production
- Status as of
- 25 Mar 2026
- NVIDIA technologies
- NVIDIA Omniverse, NVIDIA Isaac
The challenge
Building a utility-scale solar plant means lifting and fastening very large numbers of heavy modules, and AES stresses that a robot doing this must cope with many climates and lighting conditions. NVIDIA frames field robotics as a way to narrow the gap between fast-rising electricity demand and the construction capacity available to build new generation.
AES launched Maximo in July 2024 to automate the lifting, placing and attaching of modules and said it expected the robot to help build up to 5 GW of its solar backlog and pipeline over three years. Because the robots work next to people on active sites, each software change needs testing before it reaches the field.12
What was implemented
According to AES's 2024 launch release, Maximo uses AI computer vision to place panels precisely, learns from operation to improve, and runs a proprietary generative AI pipeline that reconstructs camera images washed out by glare. It had been validated at several US sites, with Oak Ridge Solar in Louisiana as its first utility-scale deployment, and was due to start at Bellefield, a 2 GW solar-plus-storage project in Kern County, in August 2024.
Maximo's March 2026 release says the team built, tested and refined robot capabilities in physics-based simulation using NVIDIA AI infrastructure, NVIDIA Omniverse libraries and the Isaac Sim robotics simulation framework before pushing updates to the field, and credits this with shorter development and validation cycles. AWS supplied compute, automated software delivery and analytics on operational data. At Bellefield the deployment grew from a single robot to four version 3.0 units working in parallel inside normal construction workflows alongside union technicians. A version 4.0 is announced but not described.123
Reported outcomes
100 MW of utility-scale solar capacity installed by Maximo robots at AES's Bellefield complex13
100 MW installed
MeasuredVersion 3.0 robots consistently exceeded a technical rate of one module per minute3
more than 1 module per minute
MeasuredCrews installed as many as 24 modules per shift hour per person, which Maximo puts at nearly double traditional methods in the region3
up to 24 modules per shift hour per person; about 2x regional baseline
Measured
Each outcome links to its source; the label there says whether the company, NVIDIA or a third party reported it.
What others can learn, and the limits
What others can take from it: simulating the robot and its site before each field update lets a team change perception and motion software without stopping construction, and pairing robots with existing crews avoids redesigning the whole build process. Limits: all performance figures come from Maximo itself, the productivity figure is a peak value against an unquantified regional baseline, and the 2024 release had projected 100 MW by 2025 while the milestone was announced in March 2026. The case depends on AES owning both the robot company and a multi-gigawatt project pipeline; most developers would buy installation as a service rather than build a robot, and the simulation tools matter mainly to the robot maker, not to the plant owner.
Explore a similar project
Use the Startup-to-NVIDIA Match with your own assumptions. Results are independent of this case.
Sources
Thank you. Your correction was sent.
The editors check it against the sources. If you left an email address, they may reply about it.