ARM Makes Foray Into Machine Learning With Two New AI Chip Designs

ARM on Tuesday lifted the curtain on two new artificial intelligence chip designs as part of a new initiative called Project Trillium, which will increase the company's investments in machine learning.

"We are launching Project Trillium to kickstart a new wave of invention in the world of artificial intelligence (AI), of which machine learning is a key part," said Jem Davies, vice president, fellow and general manager of Machine Learning at ARM in a statement. "Getting to this point is the result of significant and prolonged investment from ARM to enable the kind of future devices we and our partners see on the horizon."

The two new designs include the ARM Machine Learning and Object Detection designs. The Machine Learning design is targeted for devices ranging from smart speakers to connected cars and servers at the high end. This design features machine learning tasks such as keyword spotting.

[Related: Intel Releases New Spectre Microcode Updates For Skylake Processors]

"We now have an ML processor architecture that is versatile enough to scale to any device, so it is more about giving markets what they need when they need it. This gives us, and our ecosystem partners, the speed and agility to react to any opportunity," said Davies.

Meanwhile, the Object Detection model is designed for processing visual data and object detection. This design can detect objects from a size of at least 50-by-60 pixels, and can process full HD at 60 frames-per-second in real time.

Similar to ARM's mobile and IoT architectures, these two designs will not go into processors made by ARM, but will be licensed to other third-party manufacturers.

ARM said that it would release its AI design to partners in the middle of 2018, and it expects the first ARM-powered devices to arrive nine months after that.

ARM's announcement comes as both chip competitors Intel and Nvidia have made significant investments in the artificial intelligence market.

Nvidia is investing heavily in deep learning, which often requires the powerful and efficient parallel computing capabilities of GPUs to teach machines how to process text, voice and other types of data as part of artificial intelligence.

Other chip companies have been looking at AI as a potential market opportunity, including Intel, which has been working up to build out its own AI portfolio, punctuated by acquiring AI startup Nervana and developing the company's technology into an application-specific integrated circuit called Nervana Engine.

"I see everyone interested in artificial intelligence," said Mike Goldstein, president and CEO of Fort Lauderdale, Fla.-based LAN Infotech. "We see AI have an impact on the medical industry. There is so much information streaming nowadays; consumers need ways to sift through that information and AI is perfect for that."