At Intel Innovation, the company’s annual event, last month, CEO Pat Gelsinger proclaimed: “We are ushering in a new age of AI PC.”
Gelsinger was talking about the next generation of Intel processors, code- named Meteor Lake, which will be launched as Intel Core Ultra in December. Chip designing, manufacturing and packaging advancements aside, one of the highlights of the new consumer chip is that it incorporates an integrated neural processing unit, or NPU.
To put it simply, an NPU is a dedicated AI (Artificial Intelligence) processor, specifically designed to take some of the load off your computer’s CPU (primary processor) and GPU (graphics processor) when running AI-related workload. An NPU enables AI inference locally on the PC—essentially allowing one to run generative AI solutions on a laptop efficiently instead of tapping into the cloud.
The shift towards introducing AI capabilities into personal laptops and desktops will be a pivotal point for personal computing. The very first slide of Gelsinger’s keynote presentation at Intel Innovation talked of “bringing AI everywhere”—personal computers, apart from the edge paradigm and cloud.
This year, AI has captured the imagination of not just large enterprises and businesses of all kinds but students and professionals as well—from chatbots to re-imagined chat experiences, to innovative design and writing tools.
A chip solely purposed for handling such tasks, therefore, make sense. NPUs are capable of handling machine learning workloads as much as 10,000 times faster than a standard GPU, and they are also more power efficient.
With the addition of an NPU, you can expect new local AI capabilities on PCs and a more personal PC. The local processing also opens a range of new experiences and helps with stronger data privacy.
For a demo at the event, Intel showed how Audacity, a popular audio software, could generate a Taylor Swift-like song in a matter of seconds while running the AI workload locally without connecting to a cloud-based service.
We have seen a similar journey in smartphones. Google’s latest Pixel 8 series smartphones, for example, pack in a slew of AI-powered features that run locally for various reasons like efficiency, performance, or data privacy via the company’s third-generation Tensor G3 chip, custom-designed to run Google’s AI models.
During a Q&A session at the event, Gelsinger called AI PCs the “Centrino moment for Intel”.
Intel’s Centrino platform—a combination of a chipset and wireless network interface—enabled broader wireless connectivity for laptops.
While the first Wi-Fi specifications came out in 1997, it was Centrino in 2003 that spawned the wireless world we are used to today and drove the pivot from desktop computers to the now dominant laptop computers.
The AI PCs, according to Intel, will drive a similar revolution.
Unlike the Centrino moment, though, it’s apparent that things are moving quite fast here. From the public availability of ChatGPT towards the end of last year to hardware and software makers building AI platforms and solutions, to Intel’s Core Ultra processors dropping later this year, generative AI has gone from an interesting concept to a massive market and technology driver in less than a year.
The age of AI PCs, hence, may be upon us already.
The need for an AI PC is also driven by Microsoft’s AI-powered capabilities coming to Windows and Microsoft 365. Windows Copilot and other augmented AI capabilities and experiences will imply greater reliance on neural processors going forward.
The hardware makers too are looking to get aboard the AI gravy train.
Microsoft’s new Surface Laptop Studio 2 is the first laptop to feature an NPU chip by Intel; the latest Apple MacBooks in the M-series silicon too boast of a multi-core neural processor, although it’s built on to the main processor (like integrated graphics). At the Intel event, Jerry Kao, chief operating officer of Acer, gave a sneak peek of a suite of Acer AI applications on an upcoming Acer laptop powered by Intel Core Ultra.
Yet, while Intel Core Ultra is a groundbreaking transformation for Intel’s consumer line-up, it’s certainly not alone.
AMD’s Ryzen 7040 series of notebook processors pack Ryzen AI NPUs. Qualcomm’s Snapdragon processor for Windows-on-Arm devices too includes its “Hexagon NPU”. Nvidia too is on it with its legacy in all-powerful consumer GPUs and recent success with high-performance chips for AI computing.
AI is set to fundamentally transform, reshape and restructure the PC experience and AI PCs will be an inflection point that will drive power-efficient AI on scale.
Intel expects to ship tens of millions of new AI-enabled PCs into the market in 2024, scaling thereafter to hundreds of millions of units. Gelsinger said he sees AI becoming a significant driver for the PC market. This is important since the PC market has been flagging for a while now due to the macroeconomic climate. AI PCs could be just the growth driver that the industry needs.
Abhishek Baxi is a technology journalist and digital consultant.