Apple has joined a consortium developing the Ultra Accelerator Link (UALink) standard, aiming to link chips within AI data centers. The UALink consortium, consisting of tech giants like Alibaba, AMD, Google, and Meta, is creating an open-standard solution for connecting GPUs and custom-designed solutions to speed up AI model training and deployment.
As a new board member, Apple’s director of platform architecture Becky Loop hailed the standard as showing “great promise” in addressing connectivity challenges. The first UALink products are expected to launch within the next couple of years, based on open standards like AMD’s Infinity Fabric.
The consortium includes Intel, AMD, Google, AWS, Microsoft, and Meta, but Nvidia, the largest AI accelerator producer, is not participating due to its proprietary NVLink technology. Apple’s participation in UALink coincides with increased investments in infrastructure for Apple Intelligence, a suite of AI product features.
UALink aims to enhance efficiency in AI data centers, which are essential for Apple’s growing AI capabilities. This move underscores Apple’s commitment to pushing innovation and collaboration within the tech industry.
Source: https://techcrunch.com/2025/01/14/apple-joins-consortium-to-help-develop-next-gen-ai-data-center-tech