Ghost

Maven Robotics emerged from stealth with a $100 million Series A, but Brian McCullough focuses on how it won its first customer in 2024. CEO Hamza Derbas entered a meeting where four robotics companies were competing to automate a major consumer-goods company’s logistics operations. While the other companies had working robots, Maven had only “a cartoon of a robot and a team of people.” Derbas asked to tour the factories and warehouses, studied the workers’ flows, and pitched an end-to-end system connected to the warehouse-management system. Two years later, Maven says as many as eight robots are operating there 16 hours per day at 99% or better uptime, though the customer and performance details remain undisclosed.

The Series A was led by RoboStrategy, with LocalGlobe, Vine Ventures, and XTX Ventures also participating. Maven had previously raised about $18 million, according to PitchBook, with no valuation disclosed; the new capital will fund 250 third-generation robots and development of a fourth generation. RoboStrategy is a NASDAQ-listed, closed-end fund trading as BOT, designed to give public investors exposure to otherwise-private robotics and physical-AI companies. Maven’s initial task is mixed-case palletizing: workers build store-specific pallets from constantly changing product mixes, often within 48 hours of demand changes. Maven also targets tote handling, which it says represents an $80 billion market.

Maven’s machine uses a wheeled base, directional wheels capable of about 10 miles per hour, a telescoping central pillar, two arms, and vacuum grippers. Each robot can lift up to 30 kilograms and reach from floor level to roughly three meters. It operates for eight hours per charge and uses hot-swappable batteries for multiple shifts. Its sensor package includes 10 cameras, two lidar units, four additional sensors, and tactile feedback for situations where vision is confused. Derbas contrasts this design with humanoids such as Agility Robotics’ warehouse robots, arguing that legs are complex, unreliable, costly, and unnecessary when the job takes place on level floors.

Hamza and his brother Khalid Derbas founded Maven after Apple shut down its autonomous-car effort in early 2024. Hamza had worked on EV motors, batteries, and control software at Ford and A123 Systems, then spent nine years in Apple’s Special Projects Group, widely believed to have been the self-driving-car program. Khalid, now Maven’s CFO, came from private equity. Much of the senior team also came from Apple, alongside people from Tesla, Rivian, Cruise, Zoox, McLaren’s Formula One team, Ferrari, and Archer Aviation. Maven says the broader group has more than 200 years of combined experience.

Maven says field data returns within minutes or hours, allowing it to retrain, evaluate, run ablation studies, select new weights, and redeploy. The company aims for 100,000 autonomous operating hours by the end of this year and one million by the end of 2027. RoboStrategy says Maven moved through four hardware generations in two years: early versions proved speed and quality, while Gen 4 is intended to combine performance and cost efficiency through proprietary actuators, casting, and injection molding. Palletizing is the beachhead for a claimed manufacturing-and-assembly market exceeding $1 trillion. Maven also collects third-party data and records human manipulation using pincer-like gloves.

RoboStrategy calls Maven’s approach “deep then wide”: master one workflow against a customer’s production metrics before expanding, rather than showing many research demos and hoping reliability follows. The company points to more than 480,000 unfilled US manufacturing jobs, potentially 2.1 million by 2030, and says robotics funding reached $18.8 billion in the first half of the year. McCullough’s concerns are whether eight robots can become 250 without failures in manufacturing, parts, service, and technicians, and whether the reported 99% uptime says anything about speed, dropped boxes, or human interventions. He also questions whether a future general physical-AI model could commoditize Maven’s workflow, though he sees customer integration, safety approvals, and operational relationships as possible moats.