Recently, alongside the full rollout of its next-generation Jetson Thor platform, NVIDIA announced a collaboration with U.S.-based agricultural robotics company Carbon Robotics, focusing on the latter’s laser weeding robot—LaserWeeder G2. Leveraging NVIDIA’s computing power and ecosystem support, the intelligence and performance of this product are set to reach new heights.
This development coincides with NVIDIA’s robotics developer community surpassing 2 million members, highlighting how AI computing platforms are rapidly accelerating innovation in traditional industries such as agriculture.
LaserWeeder G2 Originally Launched in February
As early as February 2025, Carbon Robotics officially released its new flagship product—the LaserWeeder G2. Utilizing laser and AI recognition technology, the system has attracted global attention as a “chemical-free weeding” solution.
The collaboration with NVIDIA does not mark a new product launch, but rather a computational upgrade and ecosystem integration based on the existing system, providing new technological momentum for its large-scale application.
Technical Highlights: 24 GPUs Power “Laser Weeding”
The LaserWeeder G2 is a 20-foot-wide towable smart weeding robot equipped with 24 high-performance lasers and 24 NVIDIA GPUs. It operates at a speed of 10,000 weeds eliminated per minute—equivalent to 600,000 weeds per hour, or approximately 167 weeds per second.
Its core advantages include:
-
Sub-millimeter precision: Powered by the NVIDIA Isaac development kit and Carbon’s custom “Large Plant Model,” it ensures weeds are targeted without damaging crops.
-
Self-built data repository: Each LaserWeeder also functions as a data collection unit. The system has already gathered and labeled over 65 million crop images, laying a solid foundation for future algorithm enhancements and cross-crop applications.
This approach resembles the training logic of large language models like ChatGPT, giving Carbon Robotics a long-term competitive edge in agricultural AI.
Applications: From Weeding to Autonomous Tractors
One of the major global challenges in agricultural production is labor shortage. Statistics show that over 25% of edible crops in the U.S. go unharvested due to a lack of qualified tractor operators.
After tackling the weeding challenge, Carbon Robotics also introduced AutoTractor, an autonomous retrofit kit that can be installed on existing tractors. It enables all-weather remote monitoring and operation, further alleviating labor shortages.
Commercial Progress and Industry Impact
Since its founding in 2018, Carbon Robotics has delivered over 150 LaserWeeder units to 14 countries worldwide, eliminating more than 30 billion weeds.
In October 2024, the company completed a $70 million Series D funding round led by Bond, with continued participation from existing investors including NVIDIA’s NVentures. Total funding has now reached $157 million.
This in-depth collaboration with NVIDIA signals that Carbon Robotics is not only a hardware leader but is also advancing toward global scalability bolstered by computing power and software ecosystem support.
Agricultural Tech Insight
Since its release in early 2025, the LaserWeeder G2 has already delivered efficient chemical-free weeding solutions to farms in multiple countries. NVIDIA’s involvement provides stronger capabilities in computing, model training, and ecosystem expansion.
This exemplifies a typical path for agricultural robotics industrialization: initial mature product deployment, followed by large-scale model and computing platform integration for accelerated upgrades and expansion. As hardware costs gradually decrease, this model is expected to be adopted by large-scale farms in the U.S., China, and beyond, driving the transition toward greener, more efficient, and intelligent agriculture.
