Encord Raises $60M for 'Physical AI' Data
Encord, a startup building data infrastructure for real-world AI applications like robotics and perception, has closed a $60M Series C. The round signals strong investor belief in the "picks and shovels" companies that enable the training and validation of embodied AI systems at enterprise scale.
Wellington Management led the round, bringing Encord's total funding to $110M and its valuation to $550M. Existing investors, including Y Combinator and CRV, also participated, signaling strong continued belief in the company's direction. Encord is tackling the data bottleneck in "physical AI," where models interact with the real world. Unlike LLMs trained on internet text, robotics and autonomous systems rely on complex, proprietary sensor data like LiDAR, video, and audio, which legacy platforms struggle to handle. Encord's platform is designed to manage, curate, and annotate these multimodal datasets. The company was co-founded by Ulrik Stig Hansen and Eric Landau, who met in London's entrepreneur network. Hansen has a background in computer science from Imperial College London and finance at J.P. Morgan, while Landau brought experience in high-frequency trading and big data systems. They identified that the critical bottleneck in AI development was shifting from compute or model size to data readiness and quality. Encord's client roster includes prominent names in autonomy and robotics like Woven by Toyota, drone delivery company Zipline, and drone manufacturer Skydio. The company also serves defense customers, including the Royal Navy, highlighting the dual-use applicability of its data infrastructure in both commercial and military domains. This traction has led to a tenfold increase in revenue from physical AI customers in the last year. The platform consolidates the entire data workflow—from management and curation to annotation and model evaluation—into a single system. This provides a clear audit trail for why a model makes certain decisions, a critical capability for debugging and improving performance in safety-critical applications like autonomous driving and defense systems. As AI moves from labs into production, the demand for robust data infrastructure is surging. Encord has seen the volume of data managed on its platform grow from one to over five petabytes in the past year. This investment aims to scale the platform to meet the needs of an industry projected to deploy over 400 million AI-powered robots in the next four years.
