Americas

  • United States

Asia

Jon Gold
Senior Writer

AMD’s new mobile and desktop chips push hard into AI

news
Apr 16, 20244 mins
AMDCPUs and ProcessorsGenerative AI

The chipmaker has become the latest to jump into the burgeoning AI PC market.

Chips Semiconductor
Credit: Shutterstock

AMD’s latest desktop and laptop processors put the company squarely in the emerging AI PC market with the integration of generative-AI (genAI) technology into some of its most popular chip lines.

The idea of an AI PC is relatively straightforward, with hardware makers like Intel, Nvidia and AMD building dedicated AI chips into processors for desktop and laptop use, delivering substantive improvements on AI-specific tasks such as text-to-image generation, content creation, and image and video processing.

The first new chips, AMD said in a briefing last week, are part of the popular Ryzen line of desktop processors, with updates in the Ryzen 7 PRO, 5 PRO and 3 PRO designs designated as the 8000 series. The larger 8-core Ryzen 7 PRO, as well as one variant of the Ryzen 5 PRO lines, will now feature dedicated neural processing units (NPUs).

Similar AI-focused upgrades are coming to AMD’s main laptop chipsets in the new 8040 series processors. Available in several configurations, with varying numbers of cores and clock speeds, all but the 8540U variant will get access to Ryzen AI.

AMD said the basic idea is to increase productivity by enabling AI to help in four key enterprise areas:

  • Automation and efficiency, abstracting away repetitive tasks, summarizing information quickly and providing translation and transcription for video calling. 
  • Optimization and predictive maintenance, with genAI tools being able to understand potential device failure and catching performance drop-offs early. 
  • NPU integration to help content creators with common tasks and the localized ability to do faster video editing, optimization and “stable diffusion” text-to-image tasks. 
  • And advantages in security, including automated threat detection and faster diagnosis and resolution of security issues.

AMD is not alone in moving to integrate NPUs, and their associated AI capabilities, into endpoint chipsets. Dell announced in February that it would partner with Intel to use the latter’s Core Ultra NPU-equipped chipsets, saying it hoped to bring energy efficiency to its desktops and laptops, as well as local AI performance improvements.

The hype around the AI PC movement is rising, but some experts say the reality of what customers can expect to see is limited by several factors. First, most popular genAI applications are likely to run almost exclusively in large public clouds for the foreseeable future. GenAI designed to run on a new generation of NPU-equipped endpoints is much more likely to be limited in scope, and feature more feature-light versions of popular tools. Another limiting factor is Microsoft, or more specifically, Windows. The company is more focused on providing its own AI features via the cloud, as mentioned, and hasn’t really outlined a cohesive vision for local AI.

Nevertheless, the AI PC concept clearly has believers, and it seems evident that there is a demand for the technology, according to TIRIAS Research principal analyst Jim McGregor. He sees several AI-based improvements coming to desktops and laptops.

“I think the best thing is going to be enhanced capabilities in the productivity apps, and I think we’re going to see digital systems,” he said. “I’d want it scanning my emails and being able to summarize them and tell you what’s important.”

The true holy grail of the technology, McGregor said, is in personalization — making AI use seamless and tailored to individual users, rather than a relatively anonymous interaction with a chatbot.

“The real pot of gold is the application that can really personalize it,” he said. “If AI knows the type of information that you track or the sources you trust, it’s going to be more intelligent around you.”

The new Ryzen chips will cost between $200 and $360 each, depending on their specific capabilities, and will be available via OEMs such as HP and Lenovo within the next couple of months, AMD said.