Movidius raises $40M to put the "eye" in IoT

Movidius, which designs “vision processing” chips, announced $40 million in new investment this week. The round was led by Summit Bridge Capital, which is co-managed by Atlantic Bridge Capital andWestSummit Capital. WestSummit’s David Lam will join Movidius’ board of directors. Also joining the round were new investors ARCH Venture Partners and Sunny Optical Technology Group, and previous Movidius investors Atlantic Bridge Capital, AIB Seed Capital Fund, Capital-E, DFJ Esprit andRobert Bosch Venture Capital.

“Machine vision” technologies allow camera-equipped devices to make sense of the images they’re capturing, just as a humans and other animal understands and navigates the world by interpreting signal from the eyes in the visual cortex. The Internet of things offers plenty of opportunities for low-power, high-performance vision processing, from helping robots find their way, to enabling connected devices to recognize and adapt to their users. (Think of an ATM or vending machine that knows when there’s a person facing it, and acts accordingly -- maybe by lighting up a screen, or saying “Hello”.)

Recognizing objects and faces, building 3D models from several images taken at different angles, and creating dynamic maps to stay oriented in unfamiliar spaces are just a few functions that Movidius’ chips can add to cell phones, robots, and IoT devices. The company made a splash in early 2014 when it’s first-generation Myriad 1 processor was selected to power Google’s Project Tango, a cell phone-like device and software kit created to promote advancement and experimentation in machine vision.

Myriad 2, released in July, improved on Movidius’ initial design by making it 20 times more power efficient -- on half a watt of power, it can handle 1080p HD feeds at 60 frames per second from six cameras simultaneously. Just half a centimeter square, it offers dedicated functions for basic machine vision as well as programmable elements so that developers can use it create new vision-processing applications. As a full system-on-a-chip, Myriad 2 can be used either as a specialist component to offload machine vision tasks from a device’s main processor, or as a standalone processor in a wearable or embedded computing device.

This week’s funding -- which follows a year that included the Myriad 2 release and the Google partnership, as well as expansions in R&D -- indicates investors’ confidence that Movidius will be supplying the eyes (or, more properly, the visual cortices) for many of the billions of devices that will be joining the IoT in years to come.

Related: OpenMV Cam