Encord, a multimodal data platform for physical AI, has raised $60M in Series C funding. The platform enables AI teams to manage, curate, annotate, and align petabytes of multimodal data across text, audio, video, images, documents, LiDAR, DICOM, geospatial, and more. It incorporates AI-assisted workflows and human-in-the-loop processes for training foundation models from initial runs to deployment. The new capital brings Encord's total funding to $110M.
Physical AI Fuels Data Platform Race
Encord's Series C arrives amid surging demand for data infrastructure in physical AI domains like robotics, autonomous vehicles, and drones. The company powers 300+ top AI teams, including Synthesia for generative video AI, Woven by Toyota for ADAS, and the Royal Navy for defense applications. Recent product launches underscore this momentum: the Physical AI Suite with LiDAR and 3D point cloud support in June 2025, and SAM 3 integration for prompt-based detection in November 2025. These updates position Encord to handle complex multi-sensor fusion essential for real-world AI deployment.
Petabyte Multimodal Data Overwhelms Teams
Physical AI applications generate petabytes of diverse data types, from LiDAR scans in AVs to DICOM images in healthcare. Traditional labeling tools support only single modalities, missing edge cases in fused datasets. Robotics teams struggle with 3D scene visualization and multi-sensor alignment, slowing model training. Current solutions fail to scale annotation for foundation models, leading to inefficient human labeling at massive volumes.
World's First Fully Multimodal Platform
Encord unifies 10+ data types in one platform, claimed as the world's first fully multimodal AI data solution. Annotation tools integrate SAM 3 for object detection, tracking, and segmentation across video and images. The Encord Index enables intelligent search and edge case detection, while active learning automates labeling prioritization. Post-training features handle RLHF, preference labeling, and model evaluation for alignment.
AI-Assisted Workflows Scale HITL
Unlike siloed competitors, Encord's data orchestration ingests petabytes from cloud sources and manages end-to-end workflows. Model debugging tools identify failure modes in physical AI scenarios. The platform released E-MM1, the largest open-source multimodal dataset at 10x prior benchmarks. These capabilities support customers like Mayo Clinic in healthcare and UiPath in enterprise automation.
Cumulative Funding Signals Data Layer Bet
The $60M Series C elevates Encord's total capital to $110M, reflecting investor conviction in data as the bottleneck for physical AI scaling. Prior rounds enabled G2 leadership in data labeling categories: Best Support, Easiest to Use, and Momentum Leader. Funding aligns with enterprise traction across healthcare, retail, manufacturing, and defense. Compliance certifications including SOC 2, HIPAA, and GDPR bolster trust for regulated sectors.
300+ AI Teams Drive Adoption
Encord serves high-profile users like AXA in insurance, Standard AI in retail, and Dyna in robotics. Customer base spans frontier AI labs to Fortune 500 enterprises. Platform's scalability handles petabyte workloads for LLMs, VLMs, and diffusion models. This breadth signals a structural shift: data platforms becoming core infrastructure for applied AI.