Opentrons integrates NVIDIA Isaac and NVIDIA Cosmos physical AI software with Opentrons’ global network of 10,000 laboratory robots to enable AI systems that learn from real-world biological experimentation
NEW YORK–(BUSINESS WIRE)–Opentrons Labworks Inc., a laboratory robotics company enabling AI-driven autonomous science, is accelerating the development and deployment of physical AI-enabled laboratory robotics with NVIDIA. Through this integration, Opentrons is using NVIDIA Isaac and NVIDIA Cosmos platforms to develop training data for physical AI models purpose-built for laboratory environments.
Opentrons delivers a unique set of real-world assets, including a global fleet of more than 10,000 deployed robotic systems across leading research universities and biopharma companies, as well as deep experimental expertise spanning both Opentrons platforms and a wide range of third-party laboratory instruments. Opentrons, with NVIDIA, is bridging simulation and reality to bring physical AI into everyday laboratory workflows.
BioNeMo provides the foundation for training and deploying AI models for biological discovery, while Opentrons supplies the standardized physical execution layer required to close the loop between digital design and experimental validation.
Until now, AI in drug discovery has largely been constrained to prediction: proposing molecular structures, suggesting targets, and analyzing datasets. Experimental execution now represents the bottleneck to novel discoveries. By standardizing experimental execution and generating high-quality training data from real laboratory operations, Opentrons enables AI systems to continuously learn from wet-lab results.
“We see a future where physical AI unlocks autonomous experimental execution throughout laboratory environments,” said James Atwood, CEO of Opentrons. “AI models and agents propose a hypothesis and experimental plan; our systems execute that experiment. The results are then fed back to the AI in a closed loop to refine the experiment further. When that cycle runs continuously across thousands of labs, discovery timelines compress from years to weeks.”
“Connecting computational models with experimental validation is essential to accelerating AI-driven drug discovery,” said Stacie Calad-Thomson, North American Business Development Lead for Healthcare and Life Sciences at NVIDIA. “With NVIDIA AI, Opentrons provides the standardized physical infrastructure that turns experimental designs into consistent, reproducible results – helping generate the training data needed to develop physical AI models that can operate across diverse laboratory environments.”
Opentrons builds AI-enabled laboratory robots that automate experimental execution across workflows ranging from antibody discovery to proteomics and genomics. With systems deployed across every top-20 U.S. research university and 14 of the top 15 global biopharma companies, Opentrons operates the world’s largest standardized network of lab automation.
Opentrons and NVIDIA will present the partnership at the SLAS International Conference and Exhibition, February 9–11, 2026, in Boston. The joint session, “Opentrons × NVIDIA: Extending Lab Automation into the Era of Physical AI,” will explore how AI-driven planning and robotic execution are converging to enable systems that design, run, and learn from their own experiments.
Directly after the talk, NVIDIA representatives will join Opentrons at Booth #917 to answer any attendee questions, with opportunities for scientists to sign up for a new program enabling the contribution of real experimental execution data to the training of AI systems for science.
About Opentrons
Opentrons builds the execution layer for autonomous science. Its open, API-driven laboratory robots standardize biological experimentation for drug discovery, genomics, and diagnostics, enabling continuous learning between AI models and physical laboratories. Backed by SoftBank and Khosla Ventures, Opentrons is headquartered in New York. For more information, visit opentrons.com.
Contacts
Media Contact
Patrick Schmidt
[email protected]
