Broadcom has introduced Tomahawk 6, the first Ethernet switching silicon to deliver over 100 Tbps of bandwidth. In this interview, Pete Del Vecchio, Product Line Manager at Broadcom, discusses the architecture and design goals behind the chip, which is optimized for AI, machine learning, and large-scale cloud networks. With support for 100G and 200G SerDes, as well as future co-packaged optics (CPO), Tomahawk 6 enables new scale-up and scale-out topologies for hyperscale GPU clusters.
Topics covered include Cognitive Routing 2.0, Broadcom’s role in the Ultra Ethernet Consortium, and the shift toward Ethernet as the unified fabric for both front-end and back-end AI networks. The discussion also addresses market readiness, power and density considerations, and performance challenges such as link flaps. This video provides technical insights into how Tomahawk 6 is positioned to support next-generation data center infrastructure at scale.
Have a tech update that you want to brief us on? Contact info@nextgeninfra.io!
