Zurich-based Flexion, a robotics intelligence startup creating a reinforcement studying platform for humanoid robots, has raised €43 million ($50 million) in Collection A funding.
The spherical was led by DST International Companions with participation from NVentures (NVIDIA’s enterprise capital arm), redalpine, Prosus Ventures, and Moonfire. This follows a €6.3 million ($7.35 million) Seed spherical secured simply months earlier.
“Our mission is easy however bold: to energy the intelligence stack for humanoid robots, to allow them to work alongside people, not depend upon them. We’re not constructing the physique. We’re constructing the mind. It is a exhausting downside. However the proper of exhausting,” stated the Flexion staff.
Throughout Europe, 2025 has seen a gentle stream of capital into robotics and autonomous methods, illustrated by raises from firms akin to NEURA Robotics in Germany, which secured €120 million for its cognitive humanoid platform; Unchained Robotics, additionally in Germany, which added €8.5 million to scale its automation-matching software program; and sensmore, with €6.5 million to advance “Bodily AI” for autonomous heavy equipment.
Additional exercise consists of Power Robotics, which raised €11.5 million for autonomous inspection methods; Greece’s Progressive Robotics, attracting €1.55 million to broaden no-code automation for SMEs; Italy’s Adaptronics, which secured €3.15 million for adaptive gripper expertise; and notably Switzerland’s mimic, elevating €13.8 million for dexterous robotic palms – making it immediately related as a fellow Zurich-based robotics startup.
Collectively these offers quantity to roughly €165 million in disclosed funding, situating Flexion’s €43 million Collection A as one of many extra substantial mid-range raises in a yr marked by diversified funding throughout cognitive robotics, autonomous inspection, dexterous manipulation, and industrial automation.
Based in 2024, Flexion is on a mission to construct the intelligence layer for robotics – focusing not on mechanical innovation, however on the adaptive autonomy wanted to make humanoid robots practical in dynamic, real-world settings.
The founding staff of Flexion contains engineers and scientists with backgrounds in reinforcement studying, management methods, mechatronics, and notion – hailing from establishments akin to ETH Zurich, and corporations together with NVIDIA, Meta, Google, Tesla, and Amazon.
Its platform combines pure language job comprehension, artificial vision-language-action coaching, and transformer-based full-body management to create robots that may function with minimal human oversight.
These capabilities goal to free robotics from reliance on pre-scripted behaviours or fixed teleoperation.
The newly raised capital will help enlargement of the corporate’s R&D operations in Zurich, scale up its compute and robotic infrastructure, set up a presence in the USA, and drive the industrial rollout of its autonomy stack.
Flexion is already collaborating with main OEM companions, and the extra sources are anticipated to assist deepen and globalise these industrial relationships.
Whereas many robotics firms have made headlines for creating lifelike humanoid machines, few have managed to show scalable, helpful functions exterior managed lab settings. The problem isn’t mechanical – it’s cognitive.
Flexion’s strategy addresses this by decoupling robotics intelligence from {hardware} design, enabling robots to interpret, plan, and act in unpredictable environments. Its software program structure is meant to work throughout numerous robotic morphologies, unlocking use circumstances in manufacturing, logistics, industrial automation, catastrophe response, and even extraterrestrial missions.
The startup’s expertise stack is structured in three layers:
- A Command Layer that makes use of massive language fashions for job breakdown and contextual grounding
- A Movement Layer skilled totally on artificial information however tuned for real-world exceptions
- A Management Layer that allows fast behaviour synthesis from a modular talent library.
Collectively, these elements allow Flexion-powered robots to exhibit studying and flexibility with out counting on human enter or brittle automation logic.
In a world dealing with vital demographic shifts and rising labour shortages – particularly in industrial manufacturing – Flexion sees humanoid robotics not simply as a technological risk, however as an financial necessity.
With a 3rd of the developed world projected to be over 60 by 2050, and mounting strain on productiveness, adaptable autonomous methods are quick changing into indispensable.
With the convergence of LLM-scale compute and new simulation environments, the startup is poised to rework transfer and assume.
