Multiply Labs uses NVIDIA platforms to scale robotics in advanced biomanufacturing

Robotic biomanufacturing company Multiply Labs announced a milestone in its mission to scale production of cell and gene therapies.

The company is now utilizing NVIDIA’s open Isaac and GR00T technologies, including advanced robotics simulation and perception. Multiply Labs said this marks a turning point for an industry that has historically relied on manual, "artisanal" processes.

Robots for cell and gene therapies

Multiply Labs said that cell and gene therapies are life-changing treatments that hold the potential to address cancer and autoimmune diseases, but manufacturing remains difficult to scale with time-intensive processes and variability that can impact throughput and cost.

The San Francisco-based company said it is developing robotics-first biomanufacturing systems intended to bring greater consistency, traceability and operational efficiency to advanced therapy production - so personalized treatments can be more broadly accessible.

“Advanced biomanufacturing is one of the highest value applications for robots. That puts us in a fortunate position to be able to invest in the most cutting-edge robotic technologies that exist,” said Fred Parietti, co-founder and CEO at Multiply Labs. “By combining our robotic approach to biomanufacturing with NVIDIA’s state-of-the-art simulation, perception and foundation model technologies, we accelerate development and unlock the next level of scalability for hardware and software systems, driving our robots towards broader patient impact.”

Multiply Labs’ robot arm operations

Multiply Labs said its systems use four robotic arms operating in parallel to maximize output in existing facilities, targeting up to 100x more patient doses per square foot of cleanroom space compared to traditional manual processes.

To accelerate development, Multiply Labs said its integration of NVIDIA robotics and AI infrastructure covers three core areas:

1. Digital biomanufacturing twins
   Multiply Labs said it is building high-fidelity robotic digital twins with NVIDIA Isaac Sim to simulate new hardware, test automation strategies and perform software-in-the-loop validation before deploying to physical systems. The company said that this approach is designed to reduce trial-and-error time on robotic systems, increasing confidence in changes prior to on-site execution.
2. Training and generalizing manipulation skills with foundation models
   Multiply Labs said it is exploring how NVIDIA Isaac GR00T open robot foundation models can learn manipulation behaviors relevant to biomanufacturing - targeting tasks that require robust material handling, general assembly, and adaptation to real-world variability.
3. Perception pipelines to convert expert demonstrations into training data
   Multiply Labs is building perception workflows with NVIDIA FoundationPose and NVIDIA FoundationStereo to learn advanced biomanufacturing steps from expert demonstrations, targeting non-intrusive data collection methods so scientists can perform at their best in true GMP (Good Manufacturing Practices) environments.

“Advanced biomanufacturing is a powerful frontier for physical AI, where robotics and AI can help scale the manufacturing of therapies that can help patients across the world,” said Stacie Calad-Thomson, North America business development lead, healthcare and life sciences, NVIDIA. “Multiply Labs, leveraging NVIDIA AI infrastructure, is helping accelerate biomanufacturing automation designed to increase reliability and scalability for advanced therapies - and translate advances in physical AI into meaningful patient impact.”