Apple’s recent acquisitionOpens a new window of the artificial intelligence start-up Xnor.ai is the latest move by a big technology company to boost edge computing.
Seattle-based Xnor.ai, founded in 2017, specializes in making low-powered chips that carry out machine learning processes on devices rather than uploading those tasks to the cloud.
This system, known as edge computingOpens a new window , holds the key to spread of artificial intelligence (AI) and the vaunted dream of “AI everywhere.â€
By processing data locally, edge computing reduces latency, the time taken to complete a task, even if only by seconds. It also solves the issue of network overload — if every device were to run its machine-learning tasks via the cloud, the tidal wave of connections would require a huge expansion of servers in cloud data centers.
1 start-up = $200 million
Start-ups such as Xnor.aiOpens a new window are developing chips that connect to the cloud to download machine learning programs. But they carry out the processing themselves and provide the results directly to the device.
Although Apple hasn’t provided details about the Xnor.ai acquisition, press reports say it paid $200 million after a bidding war with Amazon, Microsoft and Intel.
Apple is said to be looking to shore up its position in AI after falling behind its biggest rivals.
Xnor’s technology could allow Apple to create a variety of AI tasks on its devices, for instance using image recognition so iPhones could automatically tag users or edit photos rather than having to upload the data to the cloud for processing.
The technology could allow users to put a picture of themselves in another background or even create deepfake videos.
Edge computing will also be vital for another of Apple’s big bets for the future: autonomous vehicles. Latency in the cloud is a huge issue here, as decisions about steering and avoiding collisions on the road must be made in microseconds.
With hundreds of millions of autonomous vehicles set to hit the highway, data center capacity would be unable to cope with the AI processing required.
A boost of privacy
One advantage of edge computing is that it plays to Apple’s strategy of respecting user privacy as the data isn’t sent to the cloud. Apple claims the high ground on privacy against rivals such as Facebook and Google, both of which it accuses of playing fast and loose with users’ data.
Some claimOpens a new window that Apple’s personal assistant Siri lags behind Amazon’s Alexa and Google Assistant because Apple is careful not to collect as much data about users as its rivals for privacy reasons, which makes Siri less responsive.
By that analysis, edge computing offers Apple an avenue to improve the performance of Siri while maintaining privacy as all data is stored locally.
An important feature of Xnor’s chips is that they use such low amounts of energy that they can be run on solar power, helping preserve the precious battery life of Apple devices.
The development of low-energy chips fueled by movement, light and other local power sources allows the chips to run for decades without being replaced, a technology that will be vital for the spread of the Internet of Things.
Apple’s buying spree
The Seattle-based start-up’s low-power edge technology has already been used commercially to power facial recognition AI for the Wyze webcam service, which at $20 is priced far below rivals such as Google Nest and Ring. In November, Wyze announced that it will no longer use Xnor’s technology, though whether this was related to the Apple acquisition is unknown.
Apple has a history of buying start-up technology firms and has purchased dozens in recent years. Chief executive Tim Cook says Apple acquires a company “on average every two to three weeks.â€
Will Xnor.ai’s technology become a vital ingredient in future Apple products by boosting edge computing and enhancing the privacy of its AI services? Stay tuned.