Untether AI®, the leader in energy-centric AI acceleration technology, has officially announced the launch of its edge-optimized solutions that are all designed to address critical challenges like inefficient resource use and labor shortages.
You see, even though AI has shown substantial potential in the context of AgTech, traditional GPU-based solutions remain expensive for a larger chunk of the farmers. Not just that, they are also costly to run from an operational standpoint.
In response to that, Untether AI brings forth a speedAI® 240 Slim technology which effectively breaks these barriers by offering edge-optimized AI acceleration in practical and scalable form. By doing so, it allows for farmers to embrace the future of farming with efficient AI solutions.
Furthermore, Untether AI’s state-of-the-art speedAI240 Slim AI accelerator card employs one novel At-Memory compute architecture which delivers superior performance and low latency at a fraction of the cost and energy, making AI-powered AgTech viable and efficient. The significance of such a feature can be understood once you take into account how most AgTech solutions rely on GPU-based solutions that, on their part, were originally designed for costly, inefficient, and power-hungry use case of gaming. An example relaying the same is put on display an unsuccessful bid to get laser weeders intended to replace harmful chemicals. The whole effort turned out to be futile because of the laser weeder’s need to use multiple GPUs, thus pushing farmers back to chemical solutions.
As for the solutions that are set to hit the deck here, they will help users with tasks such as crop monitoring, weeding, identifying pests or diseases, automating harvesting processes, and optimizing irrigation.
“At Untether AI, we believe that AI is not just a technology advancement; it is a force that will redefine the agricultural landscape,” said Chris Walker, CEO Untether AI. “Our speedAI family of AI accelerators has the best performance, power and accuracy giving AgTech equipment designers the essential capabilities to maximize farm equipment efficiency and effectiveness; reducing upfront and operational costs, making sustainable practices viable.”
Untether AI is also set to unveil an open-source platform which leverages a centralized architecture to enhance reliability, reduce costs, and simplify maintenance. All in all, given this departure from power hungry, expensive GPU-centric models, Untether AI’s accelerators enable advanced AI capabilities for the purpose of driving both productivity and sustainability in agricultural operations.
We referred to the speedAI240 Slim technology, but what we haven’t mentioned yet is its powerful 2000 Tera Operations Per Second (TOPs) performance, which when placed alongside a 75-watt power consumption, drastically reduces system costs and complexity by replacing multiple GPUs with a single device, enhancing overall reliability.
Making this proposition even more important is the recent publication of MLPerf® Inference v4.1 benchmarks that have propelled Untether AI to achieve 6X improvement in energy efficiency and 3X lower latency.
“Untether AI’s speedAI240 Slim AI Inference accelerator card brings the power of data center performance to edge and near-edge applications, making it a great fit for our edge server product family,” said Andrew Woollard, CTO of J-Squared Technologies, an Untether AI partner. “We look forward to exploiting the advancements in high-speed data bandwidth and higher TOPs performance-per-watt as we continue to build state-of-the-art systems and support next-gen AI applications.”
Another detail worth a mention here is rooted in Untether AI’s focus on sustainability. In simple terms, thanks to the underlying technology, the company is now prepared to conceive efficient farming, reduced chemical use, healthier crops, and ultimately, a more profitable and sustainable future for farmers.