AI teams building physical-world systems pay for tooling that turns messy sensor data into training-ready datasets, with governance and quality controls that reduce rework. Encord sits in that workflow, selling the infrastructure layer that helps robotics, autonomous systems and computer vision teams label, curate and manage data at scale.
London-based Encord has raised EUR 50 million in Series C funding, led by Wellington Management, with participation from existing investors Y Combinator, CRV, N47, Crane Venture Partners and Harpoon Ventures, alongside new investors Bright Pixel Capital and Isomer Capital. The round was announced recently. Encord’s total funding now stands at EUR 93 million.
A funding signal: investors are backing AI infrastructure, not just models
The round lands amid a clear shift in European and global AI financing. While model developers continue to attract large cheques, investors are increasingly underwriting the picks-and-shovels layer: data pipelines, compute, cloud capacity and tooling that makes deployment repeatable. Encord’s Series C fits that pattern, alongside large rounds into French model development, UK cloud and data centre capacity, and Berlin-based autonomy platforms.
The scale of capital flowing into the space is substantial. More than EUR 3 billion has been directed towards AI models, compute infrastructure and physical-world applications in Europe, underlining how quickly the AI stack is industrialising.
Why Encord’s wedge matters: data scale becomes a switching cost
Infrastructure businesses win when they embed into production workflows and become hard to replace. Encord’s traction points to that kind of operational dependency.
The company said its platform footprint grew from 1 petabyte to over 5 petabytes in twelve months, and that revenue from physical AI customers increased 10x over the same period. It serves over 300 AI teams globally, including Woven by Toyota, Zipline and Skydio.
For physical AI teams, the pain is not just “labeling”. It is dataset versioning, auditability, quality assurance, edge-case management and keeping humans-in-the-loop without ballooning costs. As volumes climb into petabytes and models iterate faster, the data layer becomes a gating function for time-to-train and time-to-deploy.
That dynamic typically supports retention. Once a company’s data governance, annotation workflows and QA rules are wired into how engineering teams ship, switching becomes risky. The value is less about a single feature and more about operational continuity.
Investor mix: institutional capital meets repeat backers
Wellington Management’s lead is notable because it brings heavyweight institutional capital into a category historically funded by specialist venture. Wellington manages more than $1.4 trillion in assets for institutional clients worldwide.
The participation of established investors such as Y Combinator and CRV alongside new names signals continued conviction that Encord’s position in the physical AI toolchain can expand. In infrastructure markets, follow-on participation often reflects confidence in enterprise adoption trends and the ability to build a durable distribution engine.
Likely focus areas after the raise
Encord has not detailed a full use-of-proceeds plan in the announcement. Based on where infrastructure platforms typically invest at this stage, likely priorities include expanding enterprise sales capacity, deepening product capabilities for larger-scale deployments, and supporting more customer segments within physical AI (inference: these are common scaling levers for data infrastructure businesses).
What is clear is the underlying demand driver: physical AI teams are moving from experimentation to deployment, and they need repeatable data operations to keep model performance improving over time.
What this enables
- More capacity to serve physical AI teams managing petabyte-scale datasets
- Faster product iteration around dataset governance, QA and production workflows
- Stronger go-to-market execution as infrastructure buyers standardise tooling
What to watch
- Whether Encord can keep expanding beyond early adopter teams into broader enterprise procurement cycles
- Competitive pressure from adjacent MLOps and data tooling vendors that may bundle similar functionality
- Signals of pricing power as data volumes grow and the platform becomes more deeply embedded
- Pace of international expansion as physical AI deployment ramps across industries