Intel: The AI PC generation is here

• The AI PC is upon us as Intel unveiled its first-ever dedicated neural processing unit to accelerate PC AI workloads.
• Intel said a large AI supercomputer will be built with StabilityAI on Intel Xeon processors and Intel Gaudi 2 AI hardware accelerators. 
• The shift into home laptops running generative AI could well be a pivot point for computing.

When Pat Gelsiner took the helm at Intel in February 2021, he came with a mission to restore the chipmaker to its former glory. He set the path for the company to make the world’s most advanced semiconductors by 2024. Then, when artificial intelligence began taking center stage in the technology industry, Gelsinger assured investors that his team would “build AI into every product that we build.” Gelsinger meant it — and now Intel has plans to add AI chips to every PC.

In a keynote presentation at the Intel Innovation event, Gelsinger showed how Intel will bring AI capabilities to customers across its hardware product range, and makes the technology accessible through open, multi-architecture software solutions. He also explained that he sees AI becoming a significant driver for the PC market. The company’s upcoming ‘Meteor Lake’ chips will help usher in the new era.

Meteor Lake is the umbrella name for the upcoming Intel Core Ultra processors, the chipmaker’s first consumer chip with a built-in neural processor for machine learning tasks. The chip, due in December, will be able to run ChatGPT-style technologies on a laptop rather than having to tap into cloud data centers for computing power. 

The capability, which Intel showed off during the software developer conference held in Silicon Valley, is made possible by new AI data-crunching features built into the forthcoming Meteor Lake laptop chip and from new software tools the company is releasing.

At the conference, Intel demonstrated laptops that could generate a song in the style of Taylor Swift and answer questions in a conversational style, all while disconnected from the internet, according to Reuters. Gelsinger even said Microsoft’s Copilot AI assistant can run on Intel-based PCs.

“We see the AI PC as a sea change moment in tech innovation,” Gelsinger said, adding that AI is about to get more personal. “AI will fundamentally transform, reshape, and restructure the PC experience – unleashing personal productivity and creativity through the power of the cloud and the PC working together,” he added. 

We are ushering in a new age – the age of the AI PC.

Here’s what we know about the AI PC

The new PC experience arrives with the upcoming Intel Core Ultra processors, code-named Meteor Lake, and it will also feature Intel’s first-ever integrated neural processing unit (NPU) for power-efficient AI acceleration and local inference on the PC. 

According to Intel executive vice president Michelle Johnston Holthaus, training large language models is still mostly done on cloud data centers, but an increasing amount of the decision-making workload done by AI – generating an answer to a query, for example – is being carried out by devices like a PC or smartphone. 

“More than 50% of enterprise-managed data will be created and processed outside the data center or cloud by 2025,” she said. The company confirmed Core Ultra will launch on December 14. “We will ship tens of millions of AI-enabled PCs into the market throughout 2024,” Holthaus said. “It’s the inflection point that will drive power-efficient AI at scale. Intel is the only company in the world to deliver that scale.”

Meteor Lake is also the first product made using the Intel 4 process node, and Intel 4, in turn, is the company’s first chip production tech to use cutting-edge extreme ultraviolet (EUV) lithography, said Intel vice president William Grimm, director of product engineering for logic tech and development

For Intel, adopting EUV lithography is critical if it’s to regain the lead in semiconductor manufacturing by 2025. The Meteor Lake chipset also comes with what Intel describes as “the most fundamental system-on-chip architecture design shift” in the past four decades. Intel said it was mainly achieved due to advanced chip packaging technology, the 3D Foveros packaging technology.

That means that instead of combining all the computing features onto a single chip, Intel separated them into “four disaggregated tiles” and then used its in-house 3D packaging technologies to connect them in one chip system. That means Intel can outsource the production of some of the individual chips and then package them in-house.

Intel has previously shared its aim to quadruple its 3D chip packaging capacity by 2023, with Malaysia eventually becoming Intel’s most extensive production base. Intel also said the company will deliver Meteor Lake’s successor chip, “Arrow Lake,” next year. 

The chip giant believes that when introduced later this decade, glass substrates –a breakthrough Intel announced this week– will allow for continued scaling of transistors on a package to help meet the need for data-intensive, high-performance workloads like AI, and keep Moore’s Law going well beyond 2030.

An AI supercomputer

At the event, Gelsinger teased what he claimed would be the “world’s largest AI supercomputer in Europe” and one of the “top 15 AI supercomputers worldwide.” Built-in collaboration with Stability AI, the company behind the popular Stable Diffusion image generation model, the all-Intel system will reportedly use a combination of Xeon processors and the 4,000 Gaudi2 AI accelerator

Besides Dell, Alibaba Cloud’s chief technology officer, Zhou Jingren, explained how the Chinese tech giant applies 4th Gen Intel Xeon processors with built-in AI acceleration to its generative AI and large language model, Alibaba Cloud’s Tongyi Foundation Models. Intel’s technology, he said, results in “remarkable improvements in response times, averaging a 3x acceleration.”