Untether AI develops high-performance AI chips intended for edge and embedded applications, including autonomous-vehicle perception, natural language processing, smart cities, banking, and retail. The company is partnering with General Motors on a two-year project to demonstrate an AV perception system that accelerates neural networks, improves vision response, reduces power draw, and helps enable battery downsizing. The GM collaboration will combine Untether’s computation technology with GM’s guidance and access to its current AV neural-network technology. The two-year project is funded with a $1 million OVIN R&D Partnership Fund award from the Ontario government plus $2.09 million in in-kind contributions from GM and Untether (people and salaries). Untether has an investor base that includes GM Ventures as a strategic investor and broader participation from VCs and industry investors. The company was founded in 2019 and is based in Toronto; with the recent grant and prior financings it has secured around $155 million in total funding to date. Untether AI develops at-memory compute architecture and ultra-efficient AI inference accelerators. Its tsunAImi accelerator cards, powered by runAI devices, are designed to eliminate the data-movement bottleneck and deliver industry-leading energy efficiency and compute density for inference. The company says its architecture targets cloud, edge, and embedded inference workloads across industries such as telecom, financial services, retail, autonomous vehicles, NLP, and smart cities. The new oversubscribed $125 million funding will enable Untether AI to expand current product reach, enhance its software offering, and accelerate development of next-generation products. Untether AI was founded in Toronto in 2018 and is backed by a group of strategic and institutional investors. The company emphasizes its scalable architecture and power efficiency as differentiators for high-throughput, low-latency AI acceleration. Untether AI develops ultra-efficient, high-performance AI chips that use at-memory computation combined with digital processing to eliminate the data-movement bottleneck in neural net inference. The company says its chip architecture enables higher energy efficiency and performance for AI inference workloads. Untether announced it will bring its chips to production and go to market this year. It raised additional funding to support that commercialization effort and has incumbent investors backing the company. Untether strengthened its management team with hires for VP of Product and VP of Software Engineering to bolster go-to-market and software optimization efforts. Founded in Toronto in 2018, the company is funded by Radical Ventures and Intel Capital. Untether AI has developed a new chip architecture for neural-net inference that combines near-memory design with digital processing to eliminate the data-movement bottleneck that costs energy and performance in traditional architectures. The company built a successful prototype with an initial small seed investment and says it has made significant technical progress. Untether plans to bring its chip innovations to markets including data centers, autonomous vehicles, vision processing and other embedded applications. Leadership changes accompanying the next stage of growth include naming Arun Iyengar as CEO and moving co-founder Martin Snelgrove to CTO. Financially, the company disclosed a $20 million Series A round that closed in May 2019 to fund product and go-to-market roadmaps. The team comprises founders Snelgrove, Darrick Wiebe, and Raymond Chik alongside new board representation from investors.

Total raised

$153M

Funding rounds

4

Latest round

Equity

Latest activity

May 2022