AI is making its way to PCs, with Apple and NVIDIA currently dominating the creative and gaming user market. This leaves the volume enterprise market and the developer space as the main opportunities for new technology from companies like Intel. Omdia’s report, AI PCs: An Emerging Device Class, highlights two technology options for delivering AI inference on PCs – one using a high-performance GPU with a traditional PC architecture, and the other using a smartphone-style system-on-chip with dedicated AI acceleration. PC manufacturers have already started shipping NVIDIA GPUs in 2019, while Apple has set the standard with its Apple Silicon MacBooks.
Researchers at Omdia have created a range of user personas to understand the different needs of PC users. The “Corporate Warrior” persona, representing business users who heavily use collaboration tools and travel frequently, is the largest group not well-served by the dominant players. Intel is aiming to break into this market by targeting these users with their Core Ultra CPUs. However, Arm-based systems like Apple’s have an advantage with their unified memory architecture, dating back to the 1980s, which is crucial for AI on PCs.
Omdia’s Principal Analyst for Advanced Computing, Alexander Harrowell, emphasizes the importance of unified memory to AI on PCs. The ability to run AI models is limited by VRAM, with popular 7B parameter class models requiring at least 8GB, and performance being limited by memory bandwidth. Arm’s investment in unified memory architecture gives them a significant advantage in this space. The report is now available in Omdia’s Advanced Computing Intelligence Service, covering technology, user personas, major silicon vendors, and PC OEMs.
Omdia, a part of Informa Tech, is a technology research and advisory group providing deep knowledge of tech markets and actionable insights to help organizations make informed growth decisions. For more information, contact Fasiha Khan at [email protected]