How Encord has raised $60M in Series C funding? Data Infrastructure Emerges as the New Battleground in the Physical AI Economy

Artificial intelligence is rapidly moving beyond text-based interfaces and into machines that interact with the physical world. Autonomous vehicles, robotics, drones, and medical automation systems are increasingly powered by AI models that must interpret complex real-world environments in real time. This shift—often referred to as “physical AI”—is creating a new layer of infrastructure demand around the data that trains and operates these systems.

While large language models have dominated headlines in recent years, industry analysts increasingly point to data quality as the critical bottleneck in real-world AI deployment. Physical AI systems must process multiple data types simultaneously, including images, video, audio, LiDAR, and sensor streams. Unlike text models trained on web-scale datasets, robots and autonomous machines require structured, highly curated data reflecting real-world conditions.

The economic stakes are substantial. The global robotics industry alone is projected to grow significantly in the coming decades, with the humanoid robotics market potentially reaching trillions of dollars in long-term value as automation spreads across manufacturing, logistics, and healthcare.

As physical AI systems move from research labs to production environments, investors are increasingly focusing on companies that provide the underlying data infrastructure required to train and maintain these systems. This emerging segment includes platforms that manage data annotation, model evaluation, dataset governance, and simulation feedback loops.

The shift reflects a broader realization in the AI industry: the performance gap between competing models is narrowing, while the quality and usability of training data are becoming the defining factor in real-world deployment success.

Within this emerging category, companies building multimodal data infrastructure platforms are attracting new waves of venture funding as enterprises seek tools capable of managing the complexity of physical AI data pipelines.

The Funding Announcement

Against this backdrop, AI infrastructure startup Encord has secured $60 million in Series C funding, bringing the company’s total capital raised to approximately $110 million.

The round was led by global investment firm Wellington Management, with participation from both existing and new investors. Returning backers include Y Combinator, CRV, N47, Crane Venture Partners, and Harpoon Ventures. New investors joining the round include Bright Pixel Capital and Isomer Capital.

The capital injection arrives at a moment when demand for infrastructure supporting robotics and autonomous systems is accelerating globally. Encord’s platform helps organizations manage and prepare the large-scale multimodal datasets required to train physical AI systems, including sensor data, images, video streams, and 3D point clouds.

According to the company, more than 300 AI teams currently use its platform. Customers include major industrial and technology organizations such as Woven by Toyota and AXA, alongside robotics developers and research labs building real-world AI systems.

Encord reports that revenue tied to physical AI deployments has increased significantly over the past year as companies move from experimental pilots to operational deployments. The company says data volume on its platform has grown rapidly as AI developers begin training models with petabyte-scale datasets drawn from real-world environments.

The new funding will be used to accelerate product development, expand into additional global markets, and further scale the company’s AI-native data infrastructure platform.

For investors, the bet is that the next phase of the AI economy will not be defined solely by model breakthroughs but by the infrastructure enabling those models to operate reliably in real-world environments.

Business Model Deep Dive

Encord operates within the rapidly growing category of AI data infrastructure platforms. Its core offering provides tools that help organizations prepare and manage the complex datasets required to train and deploy AI systems operating outside controlled digital environments.

The platform supports what engineers describe as “multimodal data pipelines.” Instead of focusing solely on text or images, it enables teams to ingest and organize multiple data formats simultaneously—including video feeds, LiDAR scans, medical imaging, geospatial datasets, and telemetry from sensors.

This capability is particularly important for industries such as autonomous driving, robotics, and healthcare technology, where models must integrate multiple signals to make accurate decisions.

Encord’s business model follows a software-as-a-service (SaaS) structure. Customers pay for access to its cloud-based platform, which supports the entire AI lifecycle—from dataset creation and annotation to model evaluation and deployment monitoring. The company also integrates human-in-the-loop workflows that allow teams to refine training data continuously as models learn from real-world operations.

A key differentiator lies in the platform’s ability to manage extremely large datasets. Physical AI systems generate massive volumes of raw data. Autonomous vehicles alone can produce terabytes of sensor data per day. Without automated tools for data cleaning, annotation, and alignment, training models at scale becomes prohibitively complex.

Encord’s system addresses this by automating data labeling workflows while enabling AI-assisted annotation and quality verification. This helps reduce the time required to prepare datasets while improving model accuracy.

The company positions its platform as a “universal data layer” for physical AI development—essentially acting as the data backbone connecting raw data sources, machine learning models, and operational deployment pipelines.

As industries increasingly deploy AI systems in environments where errors carry safety or financial risks—such as hospitals, transportation networks, or defense applications—the need for reliable data infrastructure becomes more pronounced.

Competitive Landscape

The AI data infrastructure market has grown increasingly competitive as demand for training data explodes.

One of the most prominent competitors in this space is Scale AI, which provides data labeling and model evaluation services used by major technology firms and government agencies. Scale AI has raised billions in funding and built partnerships with leading AI labs.

Another fast-growing player is Surge AI, which focuses on high-quality human-generated training data used in reinforcement learning and specialized AI workflows. The company has explored large capital raises amid strong demand for expert-labeled datasets used to train advanced AI models.

Meanwhile, platforms such as Labelbox and Snorkel AI are building tools aimed at improving dataset creation and governance across enterprise AI systems.

Encord’s positioning differs slightly from many of these companies by focusing specifically on multimodal datasets used in robotics and physical-world AI systems rather than purely digital AI models.

Geographically, the competitive landscape also reflects broader AI investment patterns. The United States remains the largest market for AI infrastructure startups, benefiting from strong venture capital ecosystems and close relationships with leading AI labs. Europe, however, has produced several emerging infrastructure companies seeking to build specialized tools for industrial automation and robotics.

In Asia, particularly China, government support and manufacturing scale have accelerated robotics adoption, potentially creating demand for large-scale AI training infrastructure across industrial sectors.

For companies like Encord, the opportunity lies in becoming the default data infrastructure layer for organizations building real-world AI systems.

Strategic Implications

The funding round reflects a broader shift in how investors evaluate the AI opportunity.

During the first wave of the generative AI boom, capital flowed heavily into model developers and foundation model providers. However, as the technology matures, investors are increasingly turning their attention to infrastructure layers that enable AI systems to operate reliably at scale.

Data infrastructure is emerging as one of those critical layers.

Physical AI systems—from delivery drones to surgical robots—require significantly more complex data management than text-based AI models. The datasets must be continuously updated, labeled, validated, and aligned with real-world operating conditions.

This creates a recurring demand for platforms that can manage data pipelines throughout the entire lifecycle of an AI system.

For venture investors, companies providing this infrastructure may offer more durable business models than those focused solely on building new models. Infrastructure platforms can serve multiple AI developers simultaneously and integrate deeply into enterprise workflows.

The trend also signals a broader maturation of the AI industry. As AI systems move from experimental deployments to operational infrastructure, reliability, safety, and governance become increasingly important.

That shift favors companies building the operational backbone of the AI economy.

For Encord, the Series C funding provides the resources to expand its platform at a moment when physical AI appears to be approaching a commercial inflection point. As robots, autonomous vehicles, and AI-driven medical systems move into mainstream use, the underlying data pipelines supporting them may become one of the most valuable segments of the broader AI ecosystem.