Race for Robotic Task Data
Dyna Robotics
Capital is rushing into robotics companies that look like data platforms, not just hardware vendors. Dyna moved from a $23.5 million seed in March 2025 to a $120 million Series A in September 2025 because investors now see embodied AI as a race to get robots into real workplaces fast, collect task data, and turn that data into better action models that can spread across more jobs and more robot types.
-
The pattern across the category is speed and scale. Figure had raised about $1.75 billion by September 2025, Agility about $641.3 million by November 2025, and Skild about $300 million by July 2024. Dyna’s rapid jump still stands out because it happened within six months of the seed, showing investors are funding early claims on training data, not waiting for mature revenue.
-
What investors are buying is the data flywheel. In modern robotics, the hard part is no longer just motors, cameras, or arms. The bottleneck is collecting real world examples of robots failing, being corrected, and improving. Dyna’s system, two arms on a wheeled base that can swap grippers and fit into existing workstations, is built to start gathering that data without customers rebuilding the whole site.
-
The investor list also matters. NVentures, Amazon Industrial Innovation Fund, Samsung Next, LG Technology Ventures, Salesforce Ventures, CRV, and First Round combine compute, distribution, and industrial channel access. That mix suggests Dyna is being financed as a full stack robotics company that could sell hardware, software, and eventually the model layer, not as a single product point solution.
The next phase is a land grab for deployed robots and the proprietary task data they create. Companies that can place systems into factories, kitchens, warehouses, and other repetitive work settings fastest will train better action models, expand into adjacent workflows, and capture more of the value as robotics shifts from one off machines toward reusable foundation model software.