Encord Raises $60M for AI Data Infrastructure
London-based Encord has secured $60 million in Series C funding to scale its AI data infrastructure platform. The company's managed data volume grew from 1 to 5 petabytes in the last year, highlighting surging enterprise demand for robust data operations to productionize agentic and multimodal models.
Wellington Management led the Series C round, with notable participation from existing investors like Y Combinator and CRV, as well as new investors including Bright Pixel Capital and Isomer Capital. This latest capital injection brings Encord's total funding to approximately $110 million and boosts its valuation to $550 million. Encord's founders, Ulrik Stig Hansen and Eric Landau, bring a unique blend of experience to the AI infrastructure space. With backgrounds in quantitative finance and physics, they identified that the critical bottleneck in AI development wasn't the models themselves, but the quality and preparation of data. Their journey through the Y Combinator accelerator program in Winter 2021 helped sharpen their focus on solving this data-centric challenge. The company's go-to-market strategy is centered around a direct sales model targeting enterprise clients in the "physical AI" sector, which includes industries like autonomous vehicles, robotics, and drone technology. Key customers include major players such as Woven by Toyota, Zipline, and Skydio, who use Encord's platform to manage complex multimodal data like LiDAR, video, and other sensor information. For AI chip companies, it's significant that Encord's platform is NVIDIA GPU-driven. The company is a member of NVIDIA's Inception program and leverages the NVIDIA Metropolis framework. This partnership indicates a focus on high-performance computing and suggests that their customers are likely operating within the NVIDIA ecosystem, a crucial piece of intelligence for go-to-market teams in the semiconductor space. Encord's platform is specifically designed to handle the petabyte-scale datasets required for training models that interact with the physical world. This involves managing a full data lifecycle, from curation and annotation to model evaluation, with a strong emphasis on automation and active learning to improve data quality over time. Their AI-native architecture is a key differentiator from more generalized data labeling tools. The focus on "physical AI" positions Encord in a rapidly growing market, with projections of over 400 million AI-powered robots coming online in the near future. This trend highlights the increasing demand for specialized infrastructure that can handle the complexities of real-world sensor data, a departure from the text and image datasets that dominated earlier AI development. From a GTM perspective, Encord's approach is typical of deep-tech companies that need to educate the market on a new category of infrastructure. Their strategy involves demonstrating clear ROI through case studies with companies like OnsiteIQ and CONXAI, showcasing significant improvements in data throughput and labeling speed. This focus on quantifiable business outcomes is essential for selling into technical and enterprise buyers. For those on a technical founder track, Encord's story underscores the importance of identifying a critical, underserved niche within a burgeoning industry. By focusing on the foundational data layer for physical AI, Hansen and Landau have carved out a defensible position in the market, attracting significant venture capital and a roster of industry-leading customers.
