Apple has allied with the Ultra Accelerator Link Consortium, aiming to elevate AI data center technology. The consortium, now counting Apple among its board members, is shaping a standard known as UALink. This is designed to connect AI accelerator chips widely deployed in server farms.
Becky Loop, Apple’s director of platform architecture, has touted UALink’s potential. She believes it could answer the central challenge of improving connectivity whilst paving the way for AI growth.
This is a series of progressive steps for Apple. The tech giant continues to demonstrate its commitment to spearheading and participating in industry-shifting innovations. By becoming part of the UALink board of directors, Apple extends its influence in the realm of AI tech development.
The main goal of UALink is to expedite the process surrounding AI models—from training and fine-tuning to operation. Utilizing open standards like AMD’s Infinity Fabric, the consortium anticipates that the first wave of UALink products will be launched in the upcoming years.
Notably absent from the lineup of UALink consortium members, which include heavyweights like Intel, AMD, Google, AWS, Microsoft, and Meta, is Nvidia—despite being the dominant player in AI accelerator production. This is possibly due to Nvidia’s proprietary interconnection technology, NVLink, used within their data center clusters.
Apple’s involvement in UALink aligns with its ongoing investment to bolster the infrastructure that supports Apple Intelligence—their suite of AI product features. As reported by The Wall Street Journal, Apple is allegedly developing a new server chip to enhance the operational effectiveness of its AI data centers.
Original source: Read the full article on TechCrunch
