#Data Infrastructure #Physical AI #Encord

Encord Secures Series C Funding for Innovative AI Data Infrastructure

Encord, a leading data infrastructure company for physical artificial intelligence (AI), has successfully raised $60 million in a Series C funding round led by Wellington Management. This new investment, announced on February 27, 2026, brings the company’s total financing to $110 million, reflecting its rapid growth and rising demand in the AI sector.

The capital infusion will expedite Encord’s efforts to develop and enhance its platform dedicated to native data infrastructure for AI, enabling teams to manage, store, annotate, and align complex multimodal data essential for physical AI systems. These systems encompass a wide range of data types, including audio, video, imagery, sensor data, and 3D point clouds—formats that traditional data platforms often struggle to handle.

With over 300 AI teams globally, including industry giants such as Woven by Toyota, Skydio, and AXA Financial, Encord has witnessed a tenfold increase in its revenue over the past year, largely driven by the boom in physical AI technologies.

The Surge of Physical AI

As Encord moves forward, the physical AI sector is poised for explosive growth. After years spent in the experimental phase, applications powered by physical AI—such as robots, autonomous vehicles, and drones—are now transitioning into full-scale production. Analysts predict that over 400 million AI robots will enter service within the next four years, contributing to an industry projected to exceed $30 billion in value during the same timeframe.

Unlike large language models trained on openly available data, physical AI systems require proprietary data to function efficiently. This includes inputs from sensors, video feeds, telemetry, and real-world scenario cases, which all necessitate substantial computational power for storage and processing.

Addressing the Data Dilemma

Commenting on these needs, Ulrik Stig Hansen, co-founder and co-CEO of Encord, emphasized that the bottleneck in physical AI isn't the size of the models but rather the accessibility and quality of the data they rely on. He stated, “You can have the most sophisticated model in the world; it will still fail if the underlying data is inadequate or misaligned with real-world conditions.