RoboBoom
Dr. Know-it-all Knows it all· March 19, 2025· 13:07

Mind-Blowing Bot Is Here! Interview with 1X CEO + Neo Robot at Nvidia GTC

Bernt Børnich, CEO of 1X, discusses Neo’s design, safety, autonomy and learning strategy, and 1X’s work with NVIDIA, with Dr. Know-it-all at NVIDIA GTC.

In this talk

  1. 1:41

    Neo’s movement and safety

    Børnich says Neo’s tendon drive uses strong motors to move tendons in a way that resembles muscle, giving the robot fluid movement. He describes Neo as lightweight, quiet and designed with soft exterior materials to reduce injury and pinch risks. He argues that safety matters because robots need to learn around people.

  2. 4:38

    Tendon drives and reliability

    Børnich says 1X uses proprietary tendons and has spent about ten years addressing stretching, wear, cycle life and system complexity. He characterizes the work as a materials science challenge and says the resulting tendon system is now more reliable than many competitors’ off-the-shelf harmonic drives.

  3. 6:06

    Autonomy grows from real-world experience

    Neo handles balance, walking and force control autonomously, while a human currently supplies higher-level directions such as where to move or what to touch. Børnich says 1X combines internet data, simulation and some teleoperation to get started, then trains on outcomes from autonomous attempts. He gives a Coke-from-the-fridge example: both a successful delivery and Neo getting distracted provide useful training data.

  4. 8:39

    NVIDIA collaboration and hardware

    Børnich describes a close working relationship with NVIDIA, including exchanging some data, compute and technical notes while 1X develops its own foundation models and reinforcement learning. Neo uses an NVIDIA Orin for onboard computing, with a planned transition to Thor. The companies also work together on low-level software for humanoids in homes.

  5. 9:30

    Home data and the case for first mover advantage

    Børnich says 1X’s goal is to reduce household chores and return 2.3 hours a day to people, starting in homes because they provide diverse data. He argues that access to large-scale, varied robot experience could create a durable advantage and help models improve their physical and spatial reasoning. He sees applications extending to industrial settings, retail, hospitals and elder care.

Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.

More talks