Maven Robotics has emerged from stealth with $100 million and a robot that does the thing warehouse workers currently do by running around and picking up boxes one at a time. The humans, to their credit, have been doing this efficiently for years. The robot does not get tired.
They won the deal, beating out companies with existing robots — by being the only company that asked what the problem actually was.
What happened
In 2024, Maven CEO Hamza Derbas had a cartoon of a robot and a team of people. He talked his way into a meeting with a large consumer goods company that had four established robotics firms lined up to pitch. Rather than pitch, he asked to visit their factories.
This is the part where it gets quietly instructive. He won. The companies with actual robots lost to the company that had thought harder about the problem. Maven has now been operating robots at customer facilities for two years, logging 99% or higher uptime across machines running 16-hour days.
Today the company is announcing $100 million from RoboStrategy, LocalGlobe, Vine Ventures, and XTX Markets Ventures, with plans to build 250 third-generation robots and begin design on a fourth generation. The robots have two arms, move at 10 miles per hour, and can lift 30 kilograms. They are, by most measures, better at this job than the people doing it.
Why the humans care
The task in question is mixed palletizing — taking boxes from different factories and building a new pallet with a specific mix of goods destined for a specific store, reconfigured every 48 hours based on real-time demand. It is currently done entirely by human labor, which has the notable disadvantage of being human.
Maven's approach is end-to-end autonomy: the system connects to a warehouse management system on one side, loads trucks on the other, and handles everything in between without requiring a human to orchestrate the middle. This is either a logistics solution or a job description, depending on which side of the pallet you are standing on.
Derbas spent nine years at Apple on a project widely believed to have been a self-driving car, before that effort was disbanded in 2024 — which is, in retrospect, impeccable timing to go build something that actually shipped. Maven, like most serious physical AI companies, draws heavily on talent from the autonomous vehicle industry, where the techniques for training robots on real-world data were quietly perfected while everyone was watching the software side.
What happens next
Maven will build its 250 robots and begin designing the fourth generation while those robots are still delivering the third generation's results. This is the cadence now — hardware generations overlapping like software sprints, each one obsoleting the last before the last has fully arrived.
The warehouse workers running around picking one of this and one of that were not consulted. The boxes will still get picked.