Arrcus has introduced TGAX, a new Telco-Grade AI Ethernet Switch designed to help network operators unlock new revenue streams by integrating AI-driven networking services. Built on the NVIDIA Spectrum-4 platform and powered by Arrcus’ ArcOS network operating system (NOS), TGAX provides high-performance, programmable networking capabilities that enable multi-cloud networking, VPN services, and AI-RAN convergence on a single infrastructure. The solution is aimed at telecom service providers looking to monetize distributed AI workloads while optimizing RAN performance.
As a key component of the broader AI-RAN architecture, TGAX allows telcos to scale AI and networking applications from the edge to the core and across cloud environments. By leveraging a common software and hardware platform, network operators can reduce complexity, improve cost efficiency, and seamlessly deploy AI and RAN workloads. The switch supports SRv6-based traffic engineering for low-latency, high-bandwidth AI applications, cloud bandwidth cost management via Egress Cost Control, and advanced QoS mechanisms to prioritize and manage network traffic.
Arrcus’ TGAX represents a future-proof platform for telcos, delivering scalability from small deployments to thousands of nodes, ensuring AI-ready edge networking. The company will showcase TGAX at Mobile World Congress 2025 in Barcelona (Hall 2, Booth 2D41). Industry leaders, including NVIDIA, recognize the switch as a critical building block for distributed AI infrastructure and AI-RAN adoption.
• Arrcus launches TGAX, a Telco-Grade AI Ethernet Switch based on NVIDIA Spectrum-4 and ArcOS NOS.
• Supports multi-cloud networking, VPN services, and AI-RAN convergence on a single infrastructure.
• SRv6-based traffic engineering ensures low-latency, high-bandwidth AI applications.
• Egress Cost Control optimizes cloud bandwidth costs for network operators.
• Scalability from edge to core, supporting thousands of nodes and AI deployments.
• Showcased at MWC 2025, with NVIDIA highlighting its role in AI-driven telecom evolution.
“The TGAX solution represents the next generation of flexible, future-proof platforms for Telcos, enabling the delivery of new revenue streams while driving operational efficiencies in a phased approach,” said Shekar Ayyar, Chairman and CEO of Arrcus.
In addition, Arrcus Expands AI Networking Capabilities with New Partnerships in 5G, Edge Computing, and AI Model Training
BARCELONA, Spain – March 3, 2025 – Arrcus has announced multiple collaborations to enhance AI-driven networking across mobile edge computing, AI model training, and telco infrastructure. Partnering with Liberty Global, Fujitsu, Eviden, and Philips, Arrcus is deploying AI-powered 5G solutions to improve public safety, industrial automation, and immersive entertainment. The initiative leverages real-time AI video analytics, ultra-low latency networking, and scalable infrastructure to support smart city applications and automated manufacturing processes. Arrcus’ ACE platform provides mobile network operators (MNOs), service providers (SPs), and enterprises with a high-performance, secure, and cost-efficient networking solution for AI workloads at the edge.
In the AI data center space, Arrcus has teamed up with Actapio, a subsidiary of LY Corporation, to optimize AI model training using its ACE-AI networking solution. This deployment integrates open networking hardware from UfiSpace and Broadcom Tomahawk 5 silicon to enhance connectivity for Nvidia GPU clusters used in large language model (LLM) training. The solution addresses key challenges such as network congestion, energy efficiency, and scalability, providing up to 51.2 Tbps non-blocking switching capacity and support for 512 x 400G ports. By modernizing its infrastructure, Actapio can now accelerate AI workloads while achieving 40%+ TCO savings through operational automation and network efficiency.
Arrcus has also partnered with Lanner Electronics to enhance AI inferencing and AI-driven RAN (AI-RAN) at the telco edge. Integrating Arrcus’ ACE networking stack with Lanner’s MGX Edge AI platform, this collaboration enables telecom providers to manage AI workloads more efficiently while reducing CAPEX and OPEX. By offloading networking functions to DPUs and leveraging SRv6-based intelligent traffic management, the joint solution enhances real-time automation, reduces latency, and improves energy efficiency. “The convergence of AI and 5G is reshaping how telco networks are built,” said Shekar Ayyar, Chairman and CEO of Arrcus. “Our partnerships enable enterprises and service providers to deploy AI-powered infrastructure with scale, efficiency, and cost-effectiveness.”
