Intel wants you to believe that its newest processors usher in a new era of personal computing.
In the chipmaker’s words, this is the era of “AI Computing,” where specialized neural processors help run large language models, create works of art, and perform many other machine learning tasks. Intel believes that by powering on-device AI, we will see new applications that are either too costly or too privacy nightmare to run in the cloud.
The company may be right, but it is not the first company to make this feature possible. Much of what Intel unveiled at this month’s CES trade show was to capture what’s already available in Apple chips, which include neural processing units found in iPhones since 2017 and Macs since 2020
INTEL’S AI APPLICATION
One of Intel’s most notable CES demos came from Rewind, a software that aims to record almost everything you do on your computer, including what you type, look at, and listen to. It then uses AI to search through all of these events and surface relevant snippets and summaries.
On-device rendering is crucial for the Rewind service, as users may not want to send their entire computing history to the cloud. But Intel isn’t the only chipmaker making this possible. Rewind first launched on MacOS over a year ago, and CEO Dan Siroker has cited Apple’s M-series processors as a key enabler. (Rewind for Windows says it’s developing a separate app called “Superpower” that will technically also be released on MacOS, but it’s unclear how that will differ from the existing Rewind app.
Echoes of what Apple already offers were visible elsewhere in Intel’s demos. Apple offers similar features such as Center Stage and Portrait Mode on Macs with M series processors. Another demo of Audacity, which separates vocal and instrumental tracks, is reminiscent of how Apple Music Sing uses on-device AI to enable karaoke on any song.
IDC’s computer industry analyst Linn Huang states that Intel is not alone in developing artificial intelligence computers. AMD is making similar moves with its latest Ryzen chips, and Qualcomm is talking up AI capabilities in its upcoming Snapdragon X Elite chips for PCs. However, he says that as of now, it is too early to distinguish between them in terms of abilities.
“The silicone is just the enabling part,” Huang says via email. “How AI works and how to use it is a matter of software.”
INTEL’S ANSWER
Robert Hallock, Intel’s senior director of technical marketing, argues via email that the company’s AI offerings are not just about what the hardware can do, but also about the work Intel is doing on the subject. He points to the company’s AI PC Accelerator Program, which connects hardware and software vendors looking to develop AI features, and notes that Intel is currently working with more than 100 software companies.
Hallock also points to OpenVINO, Intel’s deep learning toolset developed over the past eight years that now helps enable on-device AI applications. For example, open source image editor GIMP uses it for AI features like Stable Diffusion rendering and super-resolution upscaling.
“Ultimately, it takes more than hardware (software, partnerships, ecosystem enablement) to ensure the success of AI on a given platform,” says Hallock.
Preaching the Gospel of AI
Therein lies the true purpose of Intel’s AI PC push: It’s about communicating that the company is all-in on productive AI and its role in the future of personal computing, and rallying the rest of the PC industry behind it.
Naturally, Microsoft also shares this view, and computer manufacturers seem to have caught on to this idea as well. We’re at the beginning of a turning point for PCs, and a wave of AI applications will soon follow, says Tom Butler, Lenovo’s general manager of worldwide commercial portfolio and product management.
“All software companies are now rapidly looking at what they can do to leverage this additional engine,” he says. “This is the beginning of this wave, and over time everything will become an AI computer.”