Maven Robotics Is Building Factories' Missing Workforce
The Santa Clara startup skipped the humanoid hype, raised $100M, and is running wheeled manipulators on factory floors right now.
By The Robotics Desk
Maven Robotics did not start with a robot. Founded in 2024, the Santa Clara company has since built four iterations of its mobile manipulator, raised $100 million, and landed a production contract with its first customer — all before most of its competitors have left the lab. [1]
The pitch is blunt: American manufacturing is short over half a million workers today, and analysts expect that gap to widen to millions within a few years. [1] CEO Hamza Derbas frames the situation plainly: the world demands more industrial output while the available workforce shrinks. Legs-first humanoids, he argues, are solving the wrong problem first — designing a form factor and then hunting for work it can do.
Maven's counter-move is morphology derived from task, not aspiration. The result can reach three metres high and carry 30-kilogram payloads while fitting through a standard doorway — a constraint that eliminates a surprising number of competitor platforms. [1] Current deployments run 16 hours a day, five days a week, at up to 99% uptime. [1]
The gripper story is a quiet lesson in why off-the-shelf rarely works at scale. Maven tried five-fingered, four-fingered, and three-fingered tooling before recognising that the arm's seven axes of freedom could substitute for finger articulation. [1] A modular end-of-arm interface lets the robot swap tools as tasks change.
The broader tension in industrial automation is not technological — it is epistemological. Knowing what a robot can do in a controlled environment tells you almost nothing about what it will do when dust, vibration, and shift changes enter the picture. The companies gaining ground are the ones that treat the factory floor as the only laboratory that matters.
Maven's commercial model reflects that stance. Rather than selling iron, the company offers robots as a service — deploying one workflow, mastering it, scaling it, then moving to the next. Mixed-case palletizing and tote handling were the first two tasks. Assembly and fabrication of durable goods — cars, ships, planes, and eventually other robots — are next on the roadmap. [1] Derbas describes the progression as a data flywheel: real-world deployment generates training signal that makes subsequent models more capable.
Shared learning across the fleet is baked into what Maven calls the Maven Network. If a robot makes a mistake in California, a robot in Japan can learn how to deal with that mistake the next day. [1] The company plans to extend operating hours to 24 hours a day, seven days a week before year-end, and targets several hundred robots on production contracts within a year, scaling toward thousands by 2028 as it prepares for mass production. [1]
Getting there means cutting costs at three points: mechanical and structural manufacturing, actuator pricing negotiated with suppliers, and redundancy elimination in electronics.
What Maven is quietly demonstrating is that the path to general-purpose industrial automation may run through patience rather than ambition — task by task, skill by skill, trust built in increments on floors where failure has a real cost.
Derbas puts the stakes with characteristic economy: "Industrial work is not for the faint of heart. The bar for reliability is very high. There's less demand on wanting a robot to be perfectly generalizable in its intelligence on Day 1. Generalization is a journey; reliability is not. It's required from Day 1." [1] For a sector that has spent years debating what the ideal robot should look like, Maven's answer is to let the work decide.