Intel has acquired Irish semiconductor start-up Movidius to advance its strategy for computer vision and the Internet of Things (IoT). Movidius's low-power, high-performance vision-processing chip Myriad 2 will allow Intel to make good progress in advancing its RealSense technology as well as in other visual/artificial intelligence (AI) endeavors.
Low-power, high-performance hardware will help Intel pioneer consumer IoT technology
A fundamental obstacle in the consumer application of computer vision (an AI subcategory) was rooted in the lack of low-power, high-performance microprocessors – a problem that Movidius helped solve. Movidius's product is categorized as an "AI accelerator" because it provides the hardware foundation necessary for AI to register the large amounts of data needed to structure necessary artificial neural networks – all at manageable energy levels. Moreover, it allows computer systems to learn from a steady record of external stimuli due to Myriad 2's "always on" visual processor that continually monitors visual data.
What makes Movidius an integral acquisition for Intel is the variety of deep-learning and AI algorithms that have been developed in-house at the vision-processing chip company. The combination of functional product design with effective AI computer vision formulas provides instant value. Along with Intel's RealSense, Movidius was a key component of Google Tango, which focused on using computer vision to provide smartphones with the capability to navigate and position themselves in the real world without the use of GPS or other external stimuli.
Although the initial plans for Movidius's product revolve around enhancing Intel's RealSense, the real potential of the acquisition lies in products such as drones, which rely on low heat, long battery life, and scenic/object recognition. In conjunction with other Intel acquisitions, Movidius's vision-processing chip will allow Intel to innovate around all of IoT, but especially around the Internet of Smartphone Things (IoST), in which consumer sales are projected to double to 1.4 billion units by 2020. Movidius has already had success in providing the hardware for the main components of IoST – robotics, wearables, and smart security – so the onus is on Intel to make a footprint in a variety of these spaces.
Furthermore, Intel's dedicated mobile VR – Project Alloy, which relies heavily on its two RealSense cameras – will benefit greatly from the reduction in power usage that the Movidius chip provides. Although Intel isn't planning on manufacturing Project Alloy, it will open the hardware and provide application program interfaces in 2017. Through the acquisition, Intel has retained a more versatile competitive position across a variety of industries such as VR, robotics, and computer vision. This is mainly due to the coupling of the Irish company's trailblazing product with the microprocessor company's reach and resources.
Naturally, some issues need to be overcome: Volumes of Movidius chips are still low so they won't benefit from Intel's manufacturing scale. Intel needs to court developer and platform support and, most importantly, Intel's recent track record of capitalizing on innovation and acquisition – OnCue, McAfee, x84 smartphones, and Tango – has been poor.
The Internet of Smartphone Things: Wearables, VR, 360-Degree Cameras, Robots, and Drones, TE0004-001092 (August 2016)
"Virtual reality: Intel targets all-in-one headset and Microsoft partnership to bring VR into the mainstream," TE0004-001103 (August 2016)
Rishi Kaul, Research Analyst, Consumer Technology