Broadcom Ships Tomahawk Ultra Ethernet Switch, Redefining AI and HPC Networking

AVGO
September 06, 2025
Broadcom Inc. announced on July 15, 2025, the shipment of its breakthrough Ethernet switch, the Tomahawk Ultra. This new switch is engineered to transform high-performance computing (HPC) and AI workloads by delivering industry-leading ultra-low latency, massive throughput, and lossless networking. The Tomahawk Ultra is optimized for tightly coupled, low-latency communication patterns found in HPC systems and AI clusters. When deployed with Scale-Up Ethernet (SUE), it enables sub-400ns XPU-to-XPU communication latency, setting a new benchmark for synchronized AI compute at scale. It also dramatically improves network efficiency by reducing Ethernet header overhead from 46 bytes to 10 bytes. This switch incorporates lossless fabric technology, utilizing Link Layer Retry (LLR) and Credit-Based Flow Control (CBFC) to eliminate packet drops during high-volume data transfer. Furthermore, Tomahawk Ultra accelerates performance through In-Network Collectives, executing operations like AllReduce directly within the switch chip to reduce job completion time and improve XPU utilization. The chip is 100% pin-compatible with Tomahawk 5, ensuring a fast time-to-market. The content on BeyondSPX is for informational purposes only and should not be construed as financial or investment advice. We are not financial advisors. Consult with a qualified professional before making any investment decisions. Any actions you take based on information from this site are solely at your own risk.