Apple may have bought its way closer to bigger AI capabilities on iPhones
Citing anonymous sources, a pair of technology publications are reporting that Apple Inc. has bought a startup that makes edge-based, super-efficient artificial intelligence tools for facial recognition and machine learning.
Geekwire said that multiple sources with knowledge of the rumored deal between Apple and Seattle-based Xnor.ai is valued at about $200 million. TechCrunch subsequently reported that an anonymous source close to Xnor.ai corroborated the initial report.
Apple has declined to comment on the stories other than to say it does, indeed, buy startups. Xnor.ai is a 2017 spinoff from the Allen Institute for AI, a research and engineering company created by Microsoft Corp. co-founder Paul Allen.
The move further solidifies the idea that significant artificial intelligence tasks — such as object recognition — are moving to mobile devices. The idea is to divvy up artificial intelligence functions and tasks between devices and cloud resources.
Cloud computing, for all the business innovations it has enabled, can still impose delays, or latency, between the time a command is issued to distant servers and a response is returned. Noting this, Apple has been moving toward edge computing for some time now.
Clever design by Xnor.ai promises to make it possible for phones, surveillance cameras and augmented-reality glasses, for example, to share at least some heavy-lifting computing without quickly draining batteries or having to hook up to an electrical outlet. Apple in the summer of 2018 bought Akonia Holographics, which makes augmented reality glasses.
Some machine learning can be done by the iPhone (in its camera, for example), but only to a limited extent because of the robust nature of significant artificial intelligence’s power and computing demands.
And yet, efficiency is not Apple’s only motivator here.
Moving functions out of the cloud also decreases worries about sensitive data being snared over telecommunications networks. One of Apple’s sustaining strengths in the market is its relative safety when it comes to hacking and virus attacks.
Alphabet Inc. is already on this trail. It’s edge-AI toolkit, called Coral, is aimed at fast prototyping. Coral is billed as having “industry-leading inference speeds” on embedded electronics (like those in surveillance cameras), and can run offline for some functions.
Both Apple and Alphabet no doubt have read a December report by Deloitte Touche Tohmatsu Ltd. predicting more than 750 million edge artificial intelligence chips or chip components will be sold this year, generating $2.6 billion in revenue. If true, the three-year compounded annual growth rate will be 36 percent, according to Deloitte.
The market for these chips, company analysts say, will grow faster than the overall chip market, exceeding 1.5 billion chips manufactured by 2024.