Untether AI was founded in 2018 with a mission to break AI out of data centers and make it available all the way to the network edge. We’ve developed a unique at-memory architecture for deep learning that places the processing next to the memory – minimizing data movement and power consumption. Using our hardware, enterprises become untethered from the data center, free to run AI workloads nearly anywhere. Other AI inference accelerator companies make promises; Untether AI makes products.

Solution Briefs

  • thumbnail: Faster, Smarter, More Efficient AI in the Real World
    Faster, Smarter, More Efficient AI in the Real World

    The power and energy efficiency of Arm Cortex-A CPUs, coupled with Untether AI at-memory compute provides an industry-leading combination for energy efficient AI inference across a wide variety of workloads.

    Learn More