Genesis AI raises $500M to develop general-purpose robot Eno
The funding round values the company at $3.5B and aims to scale a robot that uses human-like hand movements to perform complex tasks.
The short version
- Genesis AI raised $500 million at a $3.5 billion valuation from investors including Nvidia, AMD, and Samsung.
- The robot, Eno, uses a wheeled base with a hinge mechanism to adjust its height for different tasks.
- Eno is powered by GENE, an AI model designed to be easier to train than traditional robotics-optimized networks.
- The robot's hands have 20 degrees of freedom and are covered in a soft material to mimic human skin.
- The company plans to use the new capital to hire more researchers and expand its training data collection.
Why it matters
This investment signals a shift toward general-purpose robots that can operate in both homes and factories. By mimicking human hand movements, the company is trying to solve the 'data bottleneck' that makes training robots for specific tasks so difficult.
A major funding milestone
Genesis AI SAS has successfully raised $500 million in a Series A funding round. This investment places the company's valuation at $3.5 billion. The round was led by a group of prominent investors including Premji Invest, Khosla Ventures, and Eclipse.
The funding list includes a heavy-hitting lineup of technology giants and executives. Venture capital arms from Nvidia, Advanced Micro Devices, Zoom Communications, Samsung Electronics, and Ericsson all participated in the round. Former Google CEO Eric Schmidt also contributed to the investment.
The company intends to use this new capital to scale its operations significantly. They plan to hire more AI researchers and roboticists to refine their technology. Additionally, they will increase their efforts to collect training data to expand what their AI model can do.
The design of Eno
The company is developing a general-purpose robot named Eno. Unlike many humanoid robots that use legs to move, Eno utilizes a wheeled base. This design choice allows it to navigate environments while maintaining a stable platform for various tasks.
To adapt to different heights and settings, the robot features a unique mechanical configuration. It uses two large panels connected by a hinge that allows the wheeled base to adjust Eno's height. This modularity helps the robot change its shape to fit the specific requirements of a job.
Eno is also designed with a focus on tactile interaction and dexterity. Its palms and fingers are modeled closely after the human hand, which has 27 degrees of freedom. Eno's hands feature 20 degrees of freedom and are covered in a soft material designed to mimic human skin.
Simplifying AI training
Eno is powered by a custom artificial intelligence model called GENE. Genesis claims that this algorithm is easier to train than traditional robotics-optimized neural networks. This ease of training is intended to result in a higher quality of output for the robot's movements.
Training AI for robots usually requires very specific datasets tailored to the robot's physical form. For example, a model for a four-legged robot requires footage of four-legged robots performing actions. The more specialized a robot is, the harder it becomes to gather the necessary data for its unique movements.
Genesis found a workaround by designing Eno's hands to be human-like. This allows the robot to be trained on demonstrations where humans perform tasks with their own hands. Because human movement data is much easier to produce than robot footage, this approach streamlines the development of complex behaviors.
Versatility in action
The GENE foundation model allows Eno to perform tasks that involve dozens of individual steps. The company has already released demonstration videos showcasing the robot's capabilities. These videos show Eno successfully playing the piano, cooking, and solving a Rubik’s Cube.
This level of versatility is a key goal for the company's general-purpose vision. By mastering these diverse tasks, the robot is intended to be useful in multiple settings. Genesis states that this capability allows Eno to operate in industrial environments as well as inside private homes.
CEO Zhou Xian noted that the company wants to build the world's best general-purpose robot. By building the entire system in-house, the team can focus on every detail of the hardware and software. This closed loop is intended to help the company improve the robot faster than competitors who might rely on outside components.
Scaling for the real world
The primary goal of this funding is to move the technology out of the lab and into the real world. Genesis wants to deploy its robots at a larger scale to test their utility in diverse environments. This requires a significant increase in both human talent and data infrastructure.
By hiring more researchers and roboticists, the company aims to accelerate the refinement of the GENE model. They need to ensure the AI can handle the unpredictability of real-world tasks. Expanding the training data collection is a critical part of making the robot more reliable for everyday use.
As the company moves forward, the focus will remain on perfecting the interaction between the wheeled base and the dexterous hands. Success will depend on whether the human-mimicry approach can truly generalize across different industries. The next phase involves proving that Eno can handle complex, multi-step jobs consistently in non-laboratory settings.
What to watch
- How the robot performs in non-laboratory, real-world home environments.
- Whether the human-mimicry training data approach scales to more complex industrial tasks.
Try it
Look up the demo videos for the Eno robot to see how it handles multi-step tasks like cooking or solving a Rubik's Cube.
Sources and links
- OfficialGenesis AI genesis.ai
- Officialgenesis.ai/blog/gene-26-5-advancing-robotic-manipulation-to-human-leve genesis.ai
- ReportingSiliconANGLE: the article we sourced
First reported by SiliconANGLE. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.