NomadicML raises $8.4 million to automate AV data processing

NomadicML, a startup building tools for artificial intelligence, announced on Tuesday that it has raised an $8.4 million seed round. The funding round, which values the company at a post-money valuation of $50 million, follows the startup winning first prize at Nvidia’s developer conference, GTC, last month.

The investment round was backed by:

• TQ Ventures (lead investor)
• Pear VC
• Jeff Dean

The startup, founded by CEO Mustafa Bal and CTO Varun Krishnan, targets a bottleneck for developers of self-driving cars and robotics: the volume of video data. According to the company, some customers have 95% of their fleet data sitting unused in archives. This video data is difficult to search, especially when developers are looking for edge cases to train their models. Bal noted that the company aims to provide developers with insights from their own footage to drive their autonomous vehicles and robots, arguing that this is what advances autonomous systems builders rather than random data.

To solve this, NomadicML provides a platform that converts raw video footage into structured, searchable datasets. While data labeling firms rely on humans or automated labeling, Krishnan argues that NomadicML’s tool is an agentic reasoning system. The system uses multiple models to understand the actions taking place in a video and put them into context, allowing users to describe what they need so the system can find it. The platform is already used by customers including Mitsubishi Electric, Natix Network, and Zendar. Antonio Puglielli, the VP of Engineering at Zendar, stated that the tool allowed his company to scale its work much faster than outsourcing.

The startup’s backers argue that building this kind of data infrastructure in-house is a distraction for robotics companies. Schuster Tanger, a partner at TQ Ventures who led the seed round, compared the infrastructure to other utilities. “It’s the same reason Salesforce doesn’t build its own cloud and Netflix doesn’t build its own [content distribution facilities],” Tanger said. “The second an autonomous vehicle company tries to build Nomadic internally, they’re distracted from what makes them win, which is the robot itself.”

Why it matters

NomadicML’s platform addresses the massive bottleneck of raw video data in autonomous systems, allowing companies to focus on robotics rather than building internal data infrastructure.