June 9, 2025 /SemiMedia/ — AMD has acquired the engineering team of Untether AI, a Toronto-based startup known for developing power-efficient AI inference accelerators for edge and data center applications. The deal marks another strategic step by AMD to expand its AI computing capabilities and challenge NVIDIA’s dominance in the field.
In a statement, AMD said the newly integrated team will enhance its AI compiler development, core design, and SoC integration. Financial terms of the agreement were not disclosed.
“We are excited to welcome this group of world-class engineers to AMD,” the company stated. “Their expertise will significantly enhance our AI hardware and software development, including digital design, verification, and product integration.”
As part of the transition, Untether AI will discontinue support for its speedAI hardware products and imAIgine SDK, according to company executive Bob Beachler.
This move follows AMD’s recent acquisition of Brium, a software startup focused on AI inference and compiler optimization, to accelerate the performance of its Instinct GPU portfolio for data centers.
Founded in 2018, Untether AI had introduced several AI acceleration products based on its proprietary near-memory compute architecture. In October 2024, the company launched the speedAI240 Slim PCIe card, which reportedly offered triple the power efficiency of rival chips in MLPerf closed data center tests, and six times greater efficiency in edge categories. It also claimed best-in-class ResNet-50 performance among single-card solutions.
The 75W PCIe card was adopted by embedded systems firm J-Squared Technologies in the U.S. and Indian AI cloud provider Ola-Krutrim. Untether AI also partnered with chipmakers such as Ampere Computing, Arm, and NeuReality, and worked with solution providers including Boston and Asa Computers.
Earlier this year, the company reported strong demand for its chips, driven by a market need for lower-power alternatives to power-hungry GPUs that can push rack power consumption up to 120kW.
All Comments (0)