Simulating a 50,000-GPU cluster before a single rack is installed marks a shift in how AI infrastructure is built — and who can build it.
IREN Ltd. is partnering with BE Networks to use NVIDIA's DSX Air simulation platform for validating the network architecture of its upcoming deployment of more than 50,000 NVIDIA Blackwell Ultra GPUs, the companies said Monday. The initiative creates a production-representative digital twin of IREN's AI cloud environment before physical infrastructure is deployed, allowing the companies to model cluster behavior, validate network topologies and test automation workflows — addressing what has become a bottleneck in large-scale AI deployments.
"AI cloud infrastructure at this scale requires extreme precision," said Denis Skrinnikoff, chief technology officer at IREN. "By combining NVIDIA DSX Air with BE Networks' automation expertise, we can validate critical design and operational decisions before deployment, reduce integration risk and bring customer capacity online with greater confidence."
The simulation environment covers NVIDIA's full AI factory stack — Blackwell Ultra compute, Spectrum-X Ethernet fabric and NVLink scale-up networking — alongside storage, orchestration and security layers. BE Networks will support the rollout with its Verity automation platform, translating validated designs into repeatable workflows across Day 0 design, Day 1 turn-up and Day 2 operations. The approach aims to compress validation timelines from months to weeks, according to Gilad Shainer, senior vice president of networking at NVIDIA.
"AI factories are among the most complex systems ever built, and simulation is becoming essential to deploying them at speed and scale," Shainer said.
DSX platform expands across the ecosystem
The deployment comes as NVIDIA expands its DSX platform, announced at GTC Taipei on May 31, which bundles open-source software libraries, reference designs and partner technologies into a unified framework for AI factory design and operations. Cloud partners including CoreWeave, Crusoe, Firmus, Lambda, Nebius, Nscale and Yotta Data Services are already deploying DSX components — DSX Sim, DSX MaxLPS and DSX OS — to reduce risk and improve GPU utilization.
On the hardware side, Dell Technologies, Hewlett Packard Enterprise, Lenovo and Supermicro are building DSX-ready systems alongside Taiwan-based manufacturers ASUS, Foxconn, GIGABYTE, Pegatron, Quanta Cloud Technology, Wistron and Wiwynn. System integrators Quanta Cloud Technology and Pegatron are working with Dassault Systèmes to create a live AI factory digital twin configurator that automates rack-to-facility design.
What simulation means for the AI infrastructure race
For IREN, the ability to simulate before building reduces integration risk at a time when AI cloud providers face pressure to bring capacity online faster while maintaining reliability. The company's vertically integrated model — owning both data center infrastructure and GPU clusters across renewable-rich regions in North America, Europe and Asia Pacific — positions it to compete with larger hyperscalers for AI training and inference workloads.
The approach also lowers the barrier for smaller cloud providers to deploy at hyperscale complexity. By validating network designs in a virtual environment, companies can avoid costly rework during physical deployment — a risk that has delayed AI factory projects across the industry. NVIDIA's Shainer said DSX enables organizations to validate infrastructure designs in weeks instead of months and deploy software in days instead of weeks.
For investors, the partnership signals that AI infrastructure buildout remains on an aggressive trajectory. IREN's 50,000-GPU cluster, if deployed successfully, would rank among the larger single-site AI supercomputers, competing with clusters operated by CoreWeave and Lambda. The company did not disclose the total capital expenditure for the deployment or a timeline for when the cluster will come online.
This article is for informational purposes only and does not constitute investment advice.