Dream Labs Raises $90M to Develop Robot Software

The big picture: Dream Labs, an artificial intelligence startup founded by former NVIDIA robotics researchers, has raised $90 million in seed funding to develop software that enables robots from different manufacturers to perform tasks without extensive machine-specific programming.

Why it matters:

• Hardware is ready: Declining costs of robotics hardware are creating a market opportunity for software platforms that simplify robot deployment and operation across diverse systems.
• Interoperability challenge: Current robotics often require extensive, machine-specific programming, limiting their versatility and increasing development costs. Dream Labs aims to provide a ‘brain’ that any robot can use.
• Accelerated adoption: A general-purpose AI model that can operate across different robotic systems could significantly reduce the time and specialized expertise needed to integrate robots into practical applications.

How it works:

• Dream API: The company launched Dream API, a developer platform that translates camera inputs and natural-language instructions into commands for controlling robotic movements.
• Dream-1 model: Dream Labs’ flagship AI model, Dream-1, is trained on videos of human activities and uses visual information to predict actions, controlling gripper-equipped robots on previously unseen compatible hardware without additional task-specific training.
• Ecosystem support: Dream Labs currently partners with over 20 robotics hardware manufacturers, supporting machines ranging from $2,000 to $10,000, and plans to expand support to wheeled mobile robots and advanced robotic hands.

The catch: While Dream Labs’ approach promises greater interoperability, the challenge lies in achieving robust performance across a truly diverse range of robotic hardware and complex, unstructured environments. The company’s current focus on ‘tabletop activities’ and ‘grippers’ suggests a targeted initial scope, and scaling to more advanced mobile robots and dexterous hands will require significant compute infrastructure and model generalization capabilities in a competitive market for embodied AI.

Key Facts

• Company: Dream Labs
• Amount: $90M
• Round: Seed
• Investors: B Capital and Kindred Ventures (co-lead), A.Capital, Unusual Ventures, SV Angel, Wing VC, Scribble VC, Adrian Aoun
• Founder: Joel Jang
• Sector: General-purpose Robotics AI Software
• Headquarters: San Francisco, US