Meet Aviz Networks at NVIDIA GTC Berlin 2026: Building AI Networks That Keep Pace with GPU Innovation
July 24, 2026
AI Factory
TL;DR
Meet Aviz Networks at NVIDIA GTC Berlin 2026 NVIDIA’s largest event in Europe, Aviz Networks will be demonstrating how you can operationalize AI fabric with ease, enabled by open networking, unparalleled RDMA observability, and smart lifecycle management. Aviz Network’s 2-in-1 solution works seamlessly on NVIDIA Spectrum-X orchestration all the way down to the packet-level view enabled by NVIDIA BlueField-3 DPUs for NeoCloud and sovereign AI projects and enterprises building hyperscale AI fabric.
Introduction
Generative AI has prompted a rethinking of what really matters in infrastructure. GPUs continue to grow in performance but what’s often being discovered is that the network can just as easily be the bottleneck. Getting data from point to point among accelerators in low latency becomes the core challenge. It's easy for the network fabric to hit its stride. Congestion becomes common. Traffic patterns moving in one direction have an unexpected ripple effect on the fabric in other directions. RDMA traffic visibility is limited. And vendor lock-in becomes a drag on costs and choices. The result? GPU utilization suffers, and model training runs take much longer. Rethinking the Future of AI Networking - Meet Aviz Networks at NVIDIA GTC 2026 Meet the Aviz Networks team, who is coming up with a totally different solution to AI networking over here in Berlin! As an NVIDIA ecosystem partner, Aviz helps organizations deploy, run and fine tune their AI fabrics through the perfect combination of open networking standards, full-stack traffic visibility, and workload-aware automation. It doesn’t matter if you are building an enterprise AI factory, establishing a sovereign AI capability or managing a large GPU cloud, you have one common mission: ensure your network supports your AI aspirations.
Why Has Networking Become the Critical Infrastructure Layer for AI?
Simply purchasing a massive number of GPUs and plugging them in does not build an AI factory. Instead, you need an architecture where thousands of GPUs can communicate with minimal latency, packet loss, and consistent utilization. Once AI clusters reach a certain scale, the network becomes the bottleneck. Many data center monitoring tools are unaware ofeast-west traffic or RDMA communication, making it impossible to diagnose network issues and optimize job placement when congestion arises or performance degradation occurs. Troubleshooting becomes guesswork.
Continuous visibility and workload-aware automation are the solutions to this problem. They have become prerequisites for any organization serious about AI infrastructure.
How Aviz simplifies operations for NVIDIA AI factories
NVIDIA Spectrum-X is only the beginning of deploying an AI factory. The ongoing operational activities - management of configuration consistency, lifecycle operations, software updates, monitoring, troubleshooting, and optimization - can be complex in large GPU deployments.
This reduces the need to stitch together complex workflows across multiple, disjointed network toolkits, Aviz ONES for NVIDIA AI Factory does this for you. Offering complete lifecycle management, multi-vendor orchestration, and uniform telemetry across the network, Aviz gives the network operations team the tools they need to control AI Fabrics from start to finish. Eliminate costly and time-consuming manual processes, achieve higher network uptime and enable your engineers to prioritize GPU utilization and network innovation.
Why is RDMA observability indispensable for AI workloads?
The performance of many high-performance AI workloads relies heavily on RDMA and RoCE traffic, enabling GPUs to communicate with each other with low latency. Unfortunately, almost all of the networking tools out there are completely blind to this traffic. As a result, if you're seeing congestion or dropped packets on your east-west fabric, you'll typically only notice a drop in GPU utilization or slower training times, before you figure out the problem is with the network. With Aviz's Deep Network Observability Suite and the power of Service Nodes running on NVIDIA BlueField-3 DPUs, we provide deep, packet-level visibility into your AI fabrics without any hardware changes or rewiring, and reveal the patterns that enable you to find bottlenecks in hours instead of days.
How open networking contributes to the growth of the AI ecosystem in Europe?
Building out their AI infrastructure, Europe’s fast-growing AI ecosystem - comprising companies, research facilities, government bodies and cloud providers - has a clear and present need for a network that not only delivers but is also elastic enough to meet the changing demands over time. The use of open networking technologies like SONiC provides a heterogeneous choice of hardware vendor options and negates vendor lock-in, allowing companies to switch out hardware at will, avoiding supply chain delays and setting their own roadmap. Given that AI will be employed over the long term, dependency on a vendor stack can present inflexibility. Open networking allows enterprises to select the optimum hardware, drive costs down, and transform network architecture as needed, without vendor lock-in.
How can Aviz help NeoCloud and GPU cloud providers scale effectively?
Multiple GPU customers are competing on the same infrastructure. GPU cloud provider challenges include meeting customers' requirements for predictable performance. Visibility in real-time on the network where the bottlenecks of congestion, lack of resources or workload patterns lie becomes indispensable.
But all without impact on current usage, without compromising the current experience, and without slowing anything down.
Aviz enables IT infrastructure teams to get full visibility of the AI fabric, providing real time, multi-tenant monitoring. With such monitoring capabilities, infrastructure teams are able to pinpoint emerging bottlenecks and monitor how workloads are performing under stress, and base decisions regarding optimization of utilization without building fixed infrastructure constraints.
Who should meet Aviz at NVIDIA GTC Berlin 2026?
NVIDIA GTC Berlin is the premier venue for architects, engineers, and business leaders who are at the forefront of building AI infrastructure in Europe. It’s the place where decisions are made and strategies are shaped.
Architecture that optimizes for AI infrastructure; Architect the NVIDIA Spectrum-X fabrics and understand scaling-out, traffic flow tuning and acceleration for GPU clusters. Network Ops team get practical, hands-on advice for using observability, RDMA telemetry and automation for transitioning from reactive issue management to proactive monitoring. NeoCloud and GPU cloud service providers can share experience and best practices in multi-tenant observability, operational agility, and open networking's ability to positively affect infrastructure costs. CIOs of enterprise IT organizations, telecom providers and sovereign AI providers in search of networking from a non-proprietary vendor, will find a willing partner in Aviz, a team focused on enabling flexible and controlled infrastructure investment.
Conclusion
With the explosion of AI infrastructure across geographies and with demands ever increasing on the AI performance front, the network has morphed into a key part of the AI Factory equation. Simply providing faster networking speeds, though a prerequisite, will not be enough. Businesses need the operational visibility, automation capabilities, processes that can be managed, and flexibility to manage shifting AI workloads as they scale
.
Visit Aviz Networks at NVIDIA GTC Berlin 2026 and learn how our combination of open networking, Deep Network Observability, and AI fabric lifecycle management will empower you to get the most from your GPUs, while controlling operational expenses. From your first AI cluster to a nationwide AI deployment, the architecture choices that you make today will define the success of your AI efforts for years to come.
Frequently Asked Questions
What is Aviz Networks demonstrating at NVIDIA GTC Berlin 2026?
Aviz Networks is demonstrating how organizations can operationalize AI fabric using open networking, RDMA observability, and smart lifecycle management. Their 2-in-1 solution works across NVIDIA Spectrum-X orchestration down to packet-level visibility enabled by NVIDIA BlueField-3 DPUs.
Why does the network become a bottleneck in large AI deployments?
As AI clusters scale, congestion becomes common and RDMA traffic patterns can have unexpected ripple effects across the fabric. Most monitoring tools are blind to east-west and RDMA traffic, turning troubleshooting into guesswork and causing GPU utilization to suffer.
What does Aviz ONES offer for NVIDIA AI Factory operations?
Aviz ONES provides complete lifecycle management, multi-vendor orchestration, and uniform telemetry across the network, eliminating the need to stitch together multiple disjointed toolkits. It helps network operations teams manage AI fabrics from Day-0 design through ongoing optimization.
How does Aviz provide RDMA observability without hardware changes?
Aviz’s Deep Network Observability Suite uses Service Nodes running on NVIDIA BlueField-3 DPUs to deliver packet-level visibility into AI fabrics without any hardware changes or rewiring. This allows teams to identify bottlenecks in hours rather than days.
How does open networking benefit Europe’s growing AI ecosystem?
Open networking technologies like SONiC give organizations a choice of hardware vendors, eliminating lock-in and allowing them to swap hardware, avoid supply chain delays, and evolve their architecture over time. This flexibility is especially important for long-term AI deployments across enterprises, research institutions, and government bodies.
How does Aviz support GPU cloud and NeoCloud providers?
Aviz provides real-time, multi-tenant monitoring that gives infrastructure teams full visibility into the AI fabric without impacting current workloads. This allows providers to pinpoint emerging bottlenecks and optimize utilization across multiple GPU customers sharing the same infrastructure.
Who is the target audience for Aviz at NVIDIA GTC Berlin 2026?
Aviz is focused on connecting with infrastructure architects, network operations teams, NeoCloud and GPU cloud providers, and enterprise or sovereign AI leaders looking for flexible, non-proprietary networking solutions. The event is a key venue where AI infrastructure decisions and strategies are shaped.
Meeting the Aviz Networks team at NVIDIA GTC Berlin 2026 is an opportunity to discuss AI fabric design, open networking, RDMA observability, and operational automation with experts focused on modern AI infrastructure. Schedule a meeting to explore how Aviz can help your organization build scalable, high-performance AI networks that are ready for the next generation of AI workloads.Click the link below to register..
