#USA #San Francisco #AI Infrastructure #Physical AI #Encord

Encord's Major Funding Move to Boost Physical AI Data Systems

In a groundbreaking development, Encord, renowned for its role in data infrastructure for physical AI, has successfully raised $60 million in a Series C funding round led by Wellington Management. This significant influx of capital brings the total funding for Encord to an impressive $110 million. It is important to note that existing investors such as Y Combinator, CRV, N47, Crane Venture Partners, and Harpoon Ventures, were also part of this latest funding event, alongside newcomers like Bright Pixel Capital and Isomer Capital.

Accelerating AI-Native Data Management

The funding is aimed at enhancing Encord's AI-native data infrastructure platform, a critical tool that aids AI teams in managing, curating, annotating, and aligning the diverse multimodal data that physical AI systems rely upon—such as audio, video, images, sensor data, 3D point clouds, and more. Traditional data platforms often struggle to accommodate such variety.

With over 300 AI teams globally leaning on their support, Encord is well-positioned in the market. Notable partnerships include major names like Woven by Toyota, Skydio, and AXA Financial. The last twelve months have proven exceptional for Encord, marked by a tenfold increase in revenue as physical AI solutions surge.

Entering a New Era of Physical AI

Recent trends indicate that physical AI, powering technologies from robots to autonomous vehicles and drones, is advancing into a significant growth phase. After years of experimentation and pilot programs, many systems are now transitioning into full-scale production. Analysts predict that more than 400 million AI robots will become operational in the next four years, with the physical AI sector potentially exceeding $30 billion in size by then.

Unlike models fueled by extensive internet data, physical AI requires proprietary datasets collected from various real-world scenarios, including sensor data and video footage. The complexity and volume of this data necessitate robust computational resources and specialized AI infrastructure designed for effective data handling.

Data Readiness and Its Implications

Ulrik Stig Hansen, Co-Founder and Co-CEO of Encord, emphasizes that while the focus might be on constructing larger AI models, the true barrier in physical AI lies in the readiness and quality of the data used. He articulates, "You can have the most sophisticated model in the world, and it will still fail if the data feeding it is incomplete or misaligned with real-world conditions. That's the problem we solve."

In the wake of heightened demand, Encord has witnessed its data volume explode from 1 petabyte to over 5 petabytes in just one year—illustrating a data growth that is three times that of what was used to train models like GPT-4. Furthermore, revenue from physical AI clients has experienced a staggering tenfold increase.

Comprehensive Data Handling throughout the AI Lifecycle

Encord's platform equips leading AI companies and teams with the capacity to gather, organize, and redeploy data throughout various phases of the model lifecycle. Whether it’s data generation during pre-training or refining models based on human feedback, Encord’s solutions are tailored to address every challenge that comes with data automation and processing for physical AI.

For instance, Bill Tinney, Senior Director of AI Product Management at Vantor, one of Encord's customers, praised the company for its ability to meet specific demands for critical infrastructure and national security. He noted, "Encord provides a unified data layer that scales according to our needs, ensuring we maintain seamless operations from curation to evaluation without experiencing tool fragmentation."

Future Growth Trajectory

Looking forward, Eric Landau, Co-Founder and Co-CEO of Encord, expressed optimism about the future, emphasizing that the fresh funding will propel product development and market expansion. He stated, "The companies excelling in physical AI recognize that the model's success hinges on the quality of the data supporting it. We are creating the necessary infrastructure to ensure that this data remains usable throughout the ongoing learning and improvement phases of physical AI systems."

Encord's advancements promise to significantly influence the capabilities and efficiency of AI technologies that operate in real-world contexts, making them a key player in the industry moving ahead.