Encord Bags $60M to Build the Data Engine for the Physical AI Boom

SAN FRANCISCO, CA – February 26, 2026 – Encord, a data infrastructure company at the forefront of the burgeoning physical AI sector, has secured $60 million in a Series C funding round led by the prominent global investment manager Wellington Management. The investment, which brings the company's total funding to $110 million and reportedly sets its valuation at $550 million, signals a significant vote of confidence in the foundational technology required to power the next wave of intelligent machines.

Existing investors including Y Combinator, CRV, and Crane Venture Partners also participated, joined by newcomers Bright Pixel Capital and Isomer Capital. The capital infusion is earmarked to scale Encord's platform, which is purpose-built to manage the complex, real-world data that fuels physical AI systems like autonomous vehicles, drones, and advanced robotics. This move comes as the industry reaches a widely recognized inflection point, transitioning from research and development into large-scale production.

The 'Picks and Shovels' of the AI Robot Gold Rush

While headlines often focus on the AI models themselves, savvy investors are increasingly betting on the critical infrastructure that makes them possible. The investment in Encord represents a strategic play on the 'picks and shovels' of the physical AI gold rush—the essential tools required to enable the entire industry. Wellington Management, known for its deep sector expertise and support of late-stage growth companies, is leading a charge that recognizes a fundamental truth: without robust data pipelines, the promise of physical AI remains just a promise.

Industry analysts have identified physical AI as a top strategic technology trend, with projections suggesting over 400 million AI-powered robots will come online in the next four years, creating an industry projected to eclipse $30 billion annually. This explosive growth is not built on code alone; it is built on data. Unlike large language models trained on the vast, text-based expanse of the open internet, physical AI models must learn from messy, proprietary, and sensor-rich data captured in the real world.

This is where Encord has found its niche and its surging demand. The company's revenue from physical AI customers has skyrocketed, growing 10x over the last twelve months. In the same period, the volume of data managed on its platform has swelled from one to over five petabytes—a volume more than three times larger than the dataset used to train the landmark GPT-4 model. This exponential growth underscores the market's urgent need for specialized data solutions.

Beyond Code: The Data Bottleneck in Physical AI

The central challenge facing the deployment of autonomous systems is no longer just about building bigger models. As Encord's Co-Founder and Co-CEO, Ulrik Stig Hansen, stated, the real bottleneck has shifted. "Everyone is focused on building bigger models," said Hansen. "But for physical AI, the bottleneck isn't model size. It's data readiness. You can have the most sophisticated model in the world, and it will still fail if the data feeding it is incomplete, inconsistent, or misaligned with real-world conditions. That's the problem we solve."

This 'data readiness' problem stems from the unique nature of physical AI's sensory inputs. These systems rely on a symphony of multimodal data streams, including high-resolution video, audio, LiDAR, radar, 3D point clouds, and robotic telemetry. Legacy data platforms, designed for simpler and more structured information, are ill-equipped to manage, curate, and annotate this torrent of complex, synchronized data. The result is often tool fragmentation and immense operational friction for AI development teams.

Encord's platform is designed to be the unified data layer that resolves this friction. It provides a comprehensive suite of tools to handle the entire data lifecycle, from initial capture and organization to annotation, curation, and alignment with human feedback. This allows companies to create a virtuous cycle where models are continuously improved with high-quality, relevant data.

An AI-Native Platform for a Physical World

Encord differentiates itself in a competitive landscape that includes major players like Scale AI and Labelbox by branding its solution as an 'AI-native' data infrastructure. This means the platform is not just a tool for labeling data but an intelligent system in its own right, built from the ground up to handle the specific demands of physical AI. It leverages automation and machine learning to create what the company calls a 'data flywheel,' where the AI models themselves are used to improve the quality and efficiency of the data preparation process.

This approach is critical for managing the dynamic and unpredictable nature of real-world environments. For an autonomous vehicle or a delivery drone, the ability to quickly identify, label, and learn from rare 'edge cases'—like an unusual obstacle in the road or unexpected weather conditions—is paramount for safety and reliability. Encord's system is designed to facilitate this rapid, continuous learning loop, ensuring that models can adapt and improve as they encounter new scenarios.

This deep specialization in physical AI and its focus on automated, continuous learning sets it apart from more generalist platforms. It provides a cohesive environment that scales with the complexity of geospatial and sensor-fusion workflows, which is a key requirement for companies operating at the frontier of robotics and automation.

From Labs to Highways: Real-World Deployment and Validation

The most compelling evidence of Encord's impact lies in its customer base, which includes over 300 leading AI teams. The company works with industry giants like Woven by Toyota, a key player in the autonomous vehicle space, and Skydio and Zipline, leaders in the drone manufacturing and delivery sectors. These partnerships validate the platform's ability to meet the stringent demands of high-stakes, real-world applications.

The testimony of its clients speaks volumes about the platform's value. Bill Tinney, Senior Director of AI Product Management and Partnerships at Vantor, a company that builds AI for critical infrastructure and national security, highlighted the platform's strategic importance. "At Vantor, we build AI for critical infrastructure and national security - we needed a data platform that could match our ambitions," said Tinney. "Encord gives us a unified data layer that scales with the complexity of our geospatial workflows, from curation to annotation to evaluation, without tool fragmentation. For production AI teams, how you operationalize your data is a core competitive advantage."

With the new funding, Encord plans to accelerate product development and expand into new markets. Eric Landau, Co-Founder and Co-CEO, emphasized the foundational role of data in the industry's future. "The companies winning in physical AI understand something that others are just beginning to realize: the model is only as good as the data behind it," Landau said. "We're building the infrastructure that makes that data usable—not just once, but continuously, as these systems learn and improve in the real world."