Broadcom starts shipping new Tomahawk Ultra networking chip

Broadcom (NASDAQ:AVGO) said it has started the shipment of its breakthrough Ethernet switch, Tomahawk Ultra, for high-performance computing and AI workloads.

The company said Tomahawk Ultra is optimized for the tightly coupled, low-latency communication patterns found in both high-performance computing, or HPC, systems and AI clusters.

With ultra-low latency switching and adaptable optimized Ethernet headers, it provides predictable, high-efficiency performance for large-scale simulations, scientific computing, and synchronized AI model training and inference, according to the company.

The chip intends to compete with Nvidia’s (NASDAQ:NVDA) NVLink Switch chip which has a similar purpose, but the Tomahawk Ultra can tie together four times the number of chips, said Ram Velaga, a Broadcom senior vice president, in an interview to Reuters. And instead of a proprietary protocol to move the data, it utilizes a boosted-for-speed version of ethernet, the report added.

Both companies’ chips help data center makers and others tie together as many chips as possible within a few feet of each other, a technique the industry calls “scale-up” computing, according to the report.

Velaga noted that Taiwan Semiconductor Manufacturing (NYSE:TSM) will make the Ultra line of processors with its five nanometer process, the report noted.

TSM and Broadcom did not immediately respond to a request for comment from Seeking Alpha.

In addition, Broadcom said that as part of its Ethernet-forward strategy for AI scale-up, the company has introduced SUE-Lite — an optimized version of the SUE specification tailored for power and area-sensitive accelerator applications.

The company noted that Tomahawk Ultra is 100% pin-compatible with Tomahawk 5, ensuring a very fast time-to-market. It is shipping now for deployment in rack-scale AI training clusters and supercomputing environments.