DriveNets Expands AI Networking Portfolio with High-Capacity AI Fabric Platforms

Lanham, Israel, July 1, 2026 – DriveNets, a leader in large-scale networking, today announced the expansion of its AI Fabric product portfolio with the introduction of the DriveNets 2600SL and DriveNets 2601S, high-performance networking platforms designed for large-scale AI infrastructure equipped with hundreds of thousands of XPUs. These high-performance platforms are based on Broadcom's Tomahawk 6 (TH6) ASIC and deliver a total capacity of 102.4 Tbps with a configuration of 64 ports × 1.6 Tbps. Industry analysts predict that the 1.6 Tbps port will dominate the majority of newly deployed AI ports by 2027. The platforms will begin shipping in the third quarter of 2026, offering both air-cooled (DriveNets 2601S) and liquid-cooled (DriveNets 2600SL) configurations to support diverse data center environments. Unlike other solutions, the DriveNets 2600SL adopts a true liquid-cooled design, with the entire system cooled 100% by liquid, maximizing power efficiency through consistent thermal performance across the entire system.

DriveNets 2601S with Broadcom Tomahawk 6 (TH6) ASIC

DriveNets 2600SL with Broadcom Tomahawk 6 (TH6) ASIC

Ido Susan, Co-Founder and CEO of DriveNets, stated: "The most expensive idle asset in the world today is an XPU waiting for the network. These new additions to our AI portfolio deliver industry-leading performance in hyperscale clusters, enabling our customers to maximize infrastructure utilization and power efficiency across any AI accelerator. As the industry moves toward heterogeneous AI architectures, these capabilities become increasingly critical."

Maximizing Performance and Economics

The new platforms expand DriveNets' AI Fabric portfolio, integrating scale-up, scale-out, and scale-across networking with front-end and storage networking to deliver rack-scale performance and high reliability for hyperscale clusters. With flexible radix (port configurations: 64 × 1,600 Gbps, 128 × 800 Gbps, 256 × 400 Gbps, 512 × 200 Gbps) and low latency, they meet the most demanding performance and scalability requirements.

Key features of the new platforms:

Support for current and next-generation XPU data rates and clusters: Utilizing 1.6 Tbps or 2 × 800 Gbps NIC connections, they enable clusters ranging from hundreds to hundreds of thousands of XPUs within a single configuration through diverse network architectures (2-stage Fat Tree, 3-stage Fat Tree, 2-stage Fat Tree multi-plane).

High performance at lower cost compared to single-vendor solutions: DriveNets' AI Fabric portfolio delivers performance comparable to single-vendor clusters even in open, multi-vendor AI clusters, while reducing infrastructure costs.

Improved token economics: DriveNets' solution provides full-stack performance optimization across the AI cluster, including optimization of NIC drivers, kernel, and Collective Communications Library (CCL). Combined with high capacity and radix (high port density), it reduces the number of required platforms and improves hardware efficiency.

Support for a wide range of AI use cases: DriveNets' AI solutions support diverse AI use cases including training and inference, serving a broad range of customers from hyperscalers and foundational model developers to NeoCloud providers and enterprises.

Charlie Kawwas, President of Broadcom's Semiconductor Solutions Group, stated: "As AI systems scale to hundreds of thousands of XPUs, the network fabric has become a critical factor determining AI performance, infrastructure utilization, and overall token economics. The combination of Broadcom's Tomahawk 6 silicon and DriveNets' high-performance AI fabric delivers unprecedented performance and reliability in hyperscale clusters. This collaboration demonstrates that open Ethernet is becoming the foundation for next-generation AI data centers."

Performance Optimization Tools and Services

The software running on the new platforms benefits from end-to-end network performance fine-tuning through full-stack optimization, from NICs through drivers, kernel, and CCL optimization, up to network protocols. Complementing this tightly integrated stack is the DriveNets AI Cluster Orchestrator, which manages the entire lifecycle of clusters, including provisioning, benchmarking, and large-scale Day-N operations. This enables rapid and efficient deployment with built-in performance validation and tuning capabilities. Furthermore, DriveNets Infrastructure Services provides professional services including cluster onboarding and advanced software optimization, ensuring each deployment quickly and reliably achieves optimal performance.

"What DriveNets is delivering validates a shift already observed in the market: Ethernet is becoming the dominant fabric for AI infrastructure in scale-out, scale-across, and increasingly scale-up scenarios, surpassing proprietary alternatives that once existed," said Alan Weckel, Founder and Technology Analyst at 650 Group. "At the same time, the rack itself is rapidly evolving. Liquid cooling, massive AI bandwidth, and a new class of high-performance systems are emerging. This is no longer a gradual opportunity in networking—it's an entirely new market that didn't exist two years ago and is now expanding toward a TAM (Total Addressable Market) exceeding $100 billion."

About DriveNets

DriveNets is a leader in large-scale networking solutions for AI infrastructure and service providers. The company leverages a disaggregated and distributed network architecture