A group of leading technology companies including AMD, Broadcom, Cisco, Google, Hewlett Packard Enterprise (HPE), Intel, Meta, and Microsoft have joined forces to develop an open industry standard for AI chip technology known as Ultra Accelerator Link (UALink). This technology aims to enhance high-speed, low latency communication between AI accelerator chips in data centers.
The establishment of an open standard for AI cluster performance will ensure that no single company gains an unfair advantage in the rapidly growing market for AI/ML, high-performance computing, and cloud applications. Notably absent from the group are NVIDIA and Amazon Web Services, with the intention of challenging the dominance of these two companies in AI hardware and the cloud market.
The UALink Promoter Group plans to create a consortium of companies to oversee the development of the UALink standard in Q3 of 2024. The first version, UALink 1.0, will enable up to 1,024 GPU AI accelerators to be connected to a single Ultra Accelerator Switch, improving speed, reducing latency, and simplifying scalability.
The UALink standard will benefit system OEMs, IT professionals, and system integrators looking to build data centers that support high speeds, low latency, and scalability. The collaboration of industry leaders in developing this standard will prevent any single company from monopolizing the AI hardware market, promoting innovation and advancing AI technology on an industry-wide scale.
