On November 20, 2025, Flexion announced a $50 million Series A backed by DST Global Partners, NVentures, redalpine, Prosus Ventures, and Moonfire after a seed round just months earlier. The company said the capital would expand the Zurich R&D team, scale compute and robot fleets, establish a U.S. presence, and accelerate commercialization of its autonomy stack.

Flexion Robotics
PrivateFlexion is a developer of hardware-agnostic autonomy software for humanoid robots.
About Flexion Robotics
Flexion is an embodied AI company developing autonomy software for humanoid robots. Rather than building its own humanoid hardware, the company focuses on creating a hardware-agnostic intelligence platform that enables robots from different manufacturers to perceive, reason, and perform useful tasks in real-world environments. Its approach positions Flexion as an AI infrastructure provider for the broader humanoid robotics ecosystem.
Latest Updates
Flexion Robotics Team
Flexion Robotics History
- 2025
Flexion emerges from ETH and NVIDIA-linked robotics talent
Flexions public materials and investor messaging describe the company as being built by researchers and engineers from ETH Zurich and NVIDIA, with a focus on reinforcement learning, generative AI, and sim-to-real transfer. From the start, the company framed itself around the intelligence layer for humanoids rather than around manufacturing its own robot body.
ETH ZurichNVIDIA alumniAutonomy stack - November 20, 2025
Series A funding accelerates the brain-for-humanoids strategy
On November 20, 2025, Flexion announced a $50 million Series A backed by DST Global Partners, NVentures, redalpine, Prosus Ventures, and Moonfire after a seed round just months earlier. The company said the capital would expand the Zurich R&D team, scale compute and robot fleets, establish a U.S. presence, and accelerate commercialization of its autonomy stack.
Series ANVenturesDST Global - November 20, 2025
Flexion defines the product as a multi-layer autonomy stack for partner OEMs
Flexion says its platform spans a command layer for language-based task decomposition, a motion layer for vision-language-action behavior, and a control layer for low-latency whole-body control. The company also says it is already working with major OEM partners, reinforcing that its commercial model is to power many humanoids rather than ship one branded machine.
OEM partnersWhole-body controlVLA


