Apple seems to be making it clear that, unlike its big tech rivals, it really has no interest in working with the U.S. government on military applications of its technology. While Apple is undoubtedly happy to sell iPhones, iPads, and Macs to military organizations, and even work with the FBI (up to a point) the company appears to be considerably less enthusiastic about getting involved in more shadowy projects.
Earlier this month, it was revealed that Apple had acquired Xnor.ai, a startup company that specialized in “edge-based” artificial intelligence, which had cracked a method of making complex machine learning algorithms work on relatively low-power hardware. While Xnor.ai was best known for powering facial and pet recognition technology in Wyze’s home security cameras, the company’s expertise ran much deeper.
In fact, as The Information has discovered, it turns out that Xnor.ai had been working on the rather controversial Project Maven, a Pentagon initiative to leverage AI software for analyzing images collected by military drones. While it’s not surprising that Xnor.ai would be involved in a project like this considering its expertise in image recognition technology, what’s interesting is that since acquiring Xnor.ai, Apple has reportedly terminated the work, according to sources familiar with the matter.
While Apple’s motivations for doing this aren’t entirely clear — it’s possible Apple simply wants to focus Xnor.ai’s talents entirely on its own projects — the move may turn out to be a popular one for both Apple employees and customers. When it was revealed last year that Google had been participating in Project Maven, thousands of its employees stood up in protest, creating pressure that ultimately forced it to withdraw from the project.
So it’s understandable how Apple might want to head this controversy off at the pass, since Apple’s employees are just as likely to be unhappy with the company being in bed with the U.S. military as Google staffers were. However, Apple is also under no obligation to maintain any existing contracts when acquiring a company, especially since its normal acquisition process is to dissolve the company’s resources into its own corporate structure. So for all intents and purposes, Xnor.ai will cease to exist as a separate entity, and it’s a safe bet that the customers of Apple’s other many acquisitions have been similarly left out in the cold in the past.
No Military Contracts
According to an investigation by The Information, however, Apple doesn’t appear to have any active military contracts at all. It sells equipment to a number of U.S. government agencies, but the U.S. military’s relationship with Apple appears to be that of a customer and nothing else.
By contrast, even after pulling out of Project Maven, Google still has several other projects on the go with the Department of Defense and the U.S. Air Force as part of its Google Cloud platform, while both Amazon and Microsoft provide cloud computing services. Microsoft recently won a $10 billion deal to provide cloud computing services to the Pentagon, and Amazon, which is protesting that deal, also has a $600 million cloud computing contract with the CIA. Further, Microsoft has a $479 million deal to supply its HoloLens AR headset to the U.S. Army, and the Department of Defense has paid Microsoft $50 million over the past two years on various information technology contracts.
Of course, to be fair Apple has no cloud computing services to offer, nor does it provide the kind of business-class platforms that Google’s G Suite and Microsoft’s Exchange and Office 365 do, so other than direct involvement in initiatives like Project Maven , Apple has a lot less that the U.S. military would be interested in.
Apple of course hasn’t commented on why it acquired Xnor.ai — it almost never talks about its acquisitions — but it’s easy to see how the highly efficient machine learning technology has the potential to not only improve Apple’s computational photography efforts, but could also help to bolster Siri, since the smaller AI company’s claim to fame is the machine-learning algorithms that can run on lower-powered devices.
In the case of Project Maven, this probably meant being able to install them on military drones, but now that it’s in Apple’s hands, we’ll likely see Xnor.ai’s technology rolled into the Neural Engine in its future A-series CPUs, where it could someday completely eliminate the need for cloud-based servers for things like Siri — something that Apple is almost certainly trying to move away from as quickly as possible in order to emphasize its stance on user privacy.