Untether AI develops high-density AI accelerators that utilize at-memory computing to enhance the speed and energy efficiency of AI inference tasks. Their technology enables real-world applications, such as autonomous vehicles and smart cities, to operate more effectively and affordably.
Funding
$125M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
TCFounders
Product
Problem
AI inference workloads are increasingly demanding, requiring significant computational power and energy efficiency, especially for edge applications like autonomous vehicles and smart cities. Traditional computing architectures often struggle to meet these demands due to memory access bottlenecks and high power consumption associated with data movement. This limits the feasibility and cost-effectiveness of deploying AI in real-world scenarios.
Solution
Untether AI offers high-performance AI inference accelerators based on its unique at-memory compute architecture, designed to overcome the limitations of traditional von Neumann architectures. By placing the compute element directly adjacent to the memory cells, Untether AI significantly reduces data movement, leading to improved energy efficiency and compute density. Their solutions enable faster, cooler, and more cost-effective AI inference, making it practical to deploy AI in a wide range of applications from cloud to edge. The company's products include accelerator cards and devices, along with a software development kit (SDK) for model deployment and optimization.
Target Audience
Untether AI's primary customers include organizations in sectors such as vision AI, automotive, AgTech, government, financial services, and data centers that require high-performance, energy-efficient AI inference solutions.
Features
- At-memory computing architecture that minimizes data movement and maximizes compute density.
- High energy efficiency, delivering TeraFlops per watt performance.
- Support for standard AI frameworks like TensorFlow, PyTorch, and ONNX.
- runAI® architecture designed for deep neural networks.
- imAIgine® SDK for multi-chip partitioning, enabling larger networks to run across multiple devices.
- speedAI240 Slim Accelerator Card with a low-profile, 75-watt TDP PCIe design.
- Devices featuring up to 1,400 custom RISC-V processors.