Major Tech Firms Collaborate to Challenge Nvidia with New AI Data Center Standard

Leading technology companies, including Meta, Microsoft, Advanced Micro Devices (AMD), and Broadcom, announced the development of a new industry standard for networking in AI data centers.
Eaton's Brightlayer Data Centers suite
This initiative, named the “Ultra Accelerator Link,” aims to create an open standard for communication between artificial intelligence accelerators, crucial for handling the vast data involved in AI tasks.

Other prominent members of this coalition include Alphabet-owned Google, Cisco Systems, Hewlett Packard Enterprise (HPE), and Intel.

Notably absent from this group are Nvidia, the current market leader in AI chips with an estimated 80 percent market share, and Marvell Technologies, key competitor of Broadcom in the networking and custom chip market.

The move is significant as companies like Google and Meta seek to reduce their reliance on Nvidia, whose comprehensive networking solutions have been integral to its dominance in the AI sector.

By establishing a standardized interface for AI and machine learning, high-performance computing (HPC), and cloud applications, the consortium aims to foster greater innovation and competition in the next generation of AI data centers, Reuters news report said.

“An industry specification becomes critical to standardize the interface for AI and Machine Learning, HPC (high-performance computing), and Cloud applications for the next generation of AI data centers and implementations,” the consortium stated.

The push for the Ultra Accelerator Link comes at a time when tech giants are heavily investing in the infrastructure needed to support AI applications, driving a surge in demand for AI data centers and their associated hardware.

The specifications for the Ultra Accelerator Link are set to be available to companies that join the Ultra Accelerator Link (UALink) Consortium by the third quarter of 2024.

Related News

Latest News

Latest News