Microsoft’s next HoloLens will use dedicated AI chip to recognize speech and images

Windows giant Microsoft will outfit the next version of its mixed reality HoloLens goggles with an extra AI processor that will let the device recognize speech and images in real time, without having to send the data back to the cloud for analysis.

Bloomberg reported Monday that the new in-house designed chip will improve the user experience without draining the battery. Designing chips in-house is expensive, but Microsoft has no choice because the technology is changing so fast it’s easy to get left behind.

“We really do need custom silicon to help power some of the scenarios and applications that we are building,” says Kevin Scott, Microsoft’s Chief Technology Officer.

An excerpt form the article:

In a May speech, Microsoft CEO Satya Nadella touted the idea of using AI to track industrial equipment, telling the user things like where to find a jackhammer, how to use it and generating a warning in case of unauthorized use or a chemical spill.

The new HoloLens chip will make that and much more possible.

A version of Microsoft’s existing Holographic Processing Unit, the new HoloLens processor was unveiled at an event in Honolulu, Hawaii, on Sunday. The chip is under development and will be included in the next version of HoloLens, but no timeframe was provided.

Microsoft claims this is the first chip of its kind designed for a mobile device. Apple is also thought to be working on a dedicated AI chip for future iPhones and Google is on the second version of its own AI chips, called Tensor Processing Units.

A report Monday from DigiTimes said that semiconductor foundry Taiwan Semiconductor Manufacturing Company is planning to expand its Chip on Wafer on Substrate packaging and testing capacity to fill increasing orders from Google, Nvidia and others for dedicated AI chips.

Virtual/augmented/mixed reality headsets require lots of processing power.

In order to provide satisfactory user experience, such headsets must be able to analyze, render and refresh what the user sees at least 96 times per second to prevent them from experiencing nausea and motion sickness while wearing the device.

By 2025, says Jim McGregor, an analyst at Tirias Research, every device people interact with will have AI built in. The general-purpose chips made by Intel for PCs and servers are unable to rapidly process multiple things at once—hence the need for dedicated AI processors.

Microsoft is no stranger to building custom chips, having designed a motion-tracking processor for the Kinect and customizable chips for servers. The company reportedly buys the chips from Altera, a subsidiary of Intel, and adapts them for its own purposes using software, an ability that’s unique to that type of chip.