DriveNets Extends AI Networking portfolio with High-Capacity AI Fabric Platforms
Powered by Broadcom’s Tomahawk 6 (TH6) ASIC, the new 1.6 Tbps platforms maximize the performance of very large AI clusters with hundreds of thousands of XPUs
Raanana, Israel – July 1, 2026 – DriveNets, a leader in large-scale networking solutions, today announced the expansion of its AI Fabric portfolio, introducing DriveNets 2600SL and 2601S, networking platforms designed for large-scale AI infrastructures with hundreds of thousands of XPUs. These high-performance platforms are based on Broadcom’s Tomahawk 6 (TH6) ASIC with a total capacity of 102.4Tbps, across 64 ports of 1.6Tbps – which industry analysts expect to make up the majority of new AI ports deployed by 2027. The platforms will ship in Q3 2026, in air-cooled (DriveNets 2601S) and liquid-cooled (DriveNets 2600SL) configurations to support diverse data center environments. Unlike solutions from other vendors, DriveNets 2600SL delivers a true 100% liquid-cooled design, ensuring consistent thermal efficiency across the entire system to maximize power efficiency.
“The most expensive idle asset in the world today is an XPU waiting on the network,” said DriveNets co-founder and CEO Ido Susan. “The new additions to our AI portfolio deliver industry-leading performance at massive cluster scale and let our customers maximize infrastructure utilization and power efficiency on any AI accelerator. These capabilities will only become more critical as the industry moves to Heterogeneous AI architectures.”
Maximized performance and economics
The new platform extends DriveNets’ AI Fabric portfolio supporting scale-up, scale-out, scale-across, and integrated front-end and storage networking at rack-scale performance and high reliability for massive cluster sizes. The platforms support the most demanding performance and scale requirements with flexible radix (64x1600Gbps, 128x800Gbps or 256x400Gbps, and 512x200Gbps) and low latency.
The new platforms:
- Support both current and next-generation XPU data rates and clusters – utilizing 1.6Tbps or 2x800Gbps NIC connectivity, via various network architectures (2-level fat tree, 3-level fat tree, and 2-level fat tree multi-plane), enabling hundreds to hundreds of thousands of XPUs in a cluster deployment.
- Provide high-performance alternative on par with a single-vendor solution, at a lower cost – DriveNets’ AI Fabric portfolio enables open, multi-vendor AI clusters to achieve performance comparable to single-vendor clusters while reducing infrastructure costs.
- Improve token economics – DriveNets’ solution provides full-stack performance optimizations across the entire AI cluster, including Network Interface Card (NIC) drivers, kernel, and Collective Communications Library (CCL) optimizations, as well as hardware efficiency (due to the high capacity and radix fewer platforms are needed).
- Address a wide range of AI use cases – DriveNets’ AI solutions address a wide set of AI use cases, including training and inference, and for customers ranging from hyperscalers and foundation model developers to NeoCloud providers and enterprises.
“As AI systems scale to hundreds of thousands of XPUs, the network fabric has become a critical factor in AI performance, infrastructure utilization, and overall token economics,” said Charlie Kawwas, President, Semiconductor Solutions Group, Broadcom. “Broadcom’s Tomahawk 6 silicon, paired with DriveNets’ high-performance AI Fabric, delivers unprecedented performance and reliability at massive cluster scale. This collaboration reflects how open Ethernet is becoming the foundation of the next-generation AI data center.”
Performance optimization tools and services
The software running on the new platforms benefits from full-stack optimization that fine-tunes network performance end-to-end, from the NIC through drivers, kernel and Collective Communications Library (CCL) optimizations, up to the network protocols. This tightly integrated stack is complemented by the DriveNets AI Cluster Orchestrator which manages the entire cluster lifecycle, including provisioning, benchmarking, and Day-N operations at scale. It enables fast and efficient deployment with built-in performance validation and tuning. In addition, DriveNets Infrastructure Services provides expert professional services, including cluster bring-up and advanced software optimization services, ensuring each deployment reaches optimal performance quickly and reliably.
“What DriveNets is delivering reinforces what we’re already seeing in the market: Ethernet is the dominant fabric for AI infrastructure across scale-out, scale-across, and increasingly scale-up, overtaking the proprietary alternatives that came before it,” said Alan Weckel, Founder and Technology Analyst, 650 Group. “At the same time, the rack itself is changing fast: liquid cooling, tremendous AI bandwidth, and a new class of high-performance systems. This is no longer an incremental networking opportunity — it’s a brand-new market that didn’t exist two years ago and is now heading toward a TAM of more than $100 billion.”
About DriveNets
DriveNets is a leader in large-scale networking solutions for AI infrastructure and service providers. The company’s disaggregated networking architecture transforms the economics of large-scale infrastructures while maximizing performance, utilization, and operational efficiency. Its high-performance AI fabric maximizes GPU utilization and accelerates deployments by optimizing the AI stack end-to-end, resulting in higher tokens-per-second and lower cost-per-token. DriveNets’ solutions power production networks for global tier-1 operators like AT&T and Comcast, and scale multi-vendor AI infrastructures at foundation model labs, NeoClouds, and enterprises. Learn more at https://www.drivenets.com
