Bodily AI is shifting from labs to large-scale use in self-driving automobiles, supply drones, and manufacturing robots. This shift brings a brand new knowledge problem for AI groups: managing complicated real-world multimodal knowledge. In contrast to massive language fashions educated on web knowledge, bodily AI requires coaching on sensor-rich datasets similar to video, audio, LiDAR, and telemetry, which older platforms wrestle to deal with.
San Francisco-based Encord needs to repair this bottleneck. The information infrastructure firm just lately raised $60 million in a Sequence C spherical led by Wellington Management, with assist from Y Combinator, CRV, N47, Crane Venture Partners, Harpoon Ventures, and new buyers Bright Pixel Capital and Isomer Capital.
This brings Encord’s whole funding to $110 million. Ulrik Stig Hansen, co-founder and co-CEO of
Encord shares with TFN, “The corporate’s valuation following its Sequence C is $550M.”
Encord’s platform helps AI groups handle, curate, annotate, and align multimodal knowledge at scale, preserving knowledge prepared for bodily AI improvement. Over the previous yr, knowledge on its platform grew from 1 to five petabytes, and income from bodily AI clients elevated tenfold. Its purchasers embody Woven by Toyota, Zipline, Skydio, and AXA Monetary.
Aiming to be the common knowledge layer for AI
Encord was based by co-CEOs Ulrik Stig Hansen and Eric Landau, who noticed a rising hole between superior AI fashions and the fragmented, handbook knowledge infrastructure groups trusted.
Hansen shares with us, “Eric and I labored on massive knowledge programs and deep studying analysis
and noticed firsthand how ML groups spent 80%+ of their time on knowledge, not fashions.
Your complete business was obsessing over mannequin structure whereas the true
bottleneck – knowledge high quality, curation, and analysis – was utterly underserved.
We began Encord to repair that.”
At its core, Encord is constructing AI-native knowledge infrastructure, purpose-built for the sensor-heavy, multimodal world of bodily AI. Their platform covers the complete knowledge lifecycle: capturing, organising, labelling, aligning with human suggestions, and redeploying for retraining.
In contrast to generalist instruments like Labelbox, Scale AI, or SuperAnnotate, Encord is all-in on automation and steady studying. That’s a game-changer for robotics, automotive, and drone firms, the place knowledge is at all times in flux.
“We deal with the complete lifecycle at petabyte scale, with an AI-native structure that makes use of fashions to enhance the information that trains them. That knowledge flywheel and suggestions loop is the core differentiator,” elaborates Hansen.
Encord’s benefit lies in a unified knowledge layer that mixes curation, annotation, and analysis workflows to keep away from software fragmentation. Its AI-native structure is constructed particularly for video, LiDAR, sensor, and 3D knowledge. It additionally gives scalable automation with lively studying and human-in-the-loop programs to enhance knowledge high quality over time.
What about range?
On range, Hansen notes, “We don’t share particular breakdowns publicly, however we’re a staff of 150+ throughout two nations (US, UK) with 40+ nationalities represented. Variety of background and thought is one thing we actively spend money on.”
What’s subsequent?
With recent funding in hand, Encord is gearing as much as speed up product improvement, develop into new markets, and scale up globally as bodily AI adoption takes off.
