Control versus Liquidity in Robot Data

This split reveals two different bets on what makes robot data valuable, control or liquidity. Config is built like a factory, it owns the operators, the capture workflow, and the step that turns human demonstrations into robot ready trajectories and model updates through Loop and CFG-1. Maniformer is built more like an exchange plus tooling vendor, using Hive to attract many outside suppliers while selling MEgo devices and custom data services into that network.1 2 3 4 5

• Config keeps the critical handoff in house. Its product flow covers planning a task, recording demonstrations with Data Studio operators or a customer team, adapting CFG-1 or another model, and deploying the result. That makes quality control tighter because the same system owns collection, conversion, and deployment.2 3 4
• Maniformer is trying to scale supply faster than a single operator team could. Hive is described as a shared network that connects providers and buyers across 15 plus countries, while MEgo gives partners standardized grippers and cameras to collect synchronized first person data in a consistent format.1 5
• A close comparable is Lightwheel. It also offers data and tooling, but leans into simulation, with SimReadyGen for building simulation assets and RoboFinals on NVIDIA Isaac Lab for evaluation. That shows the broader market is fragmenting into three layers, real world capture, marketplace supply, and simulation based testing.6 7 8

The likely direction is convergence. Marketplace models will add more standards and QA so buyers trust outside data, while vertically integrated players will expose more of their workflow to partners to grow supply. The winners will be the ones that can show a straight line from raw demonstrations to better performance on a customer robot in a real task.1 2 6 7