Visionbay.ai has selected Netris to run network automation for its GPU cluster and AI supercomputing centre in Taiwan, a project the companies describe as the country's largest GPU cluster.
Visionbay, backed by Foxconn and operating as the group's unit for AI supercomputing and cloud operations, has also standardised on the Netris platform across its broader GPU cluster roadmap. The deployment centres on Netris's NAAM platform, short for Network Automation, Abstraction, and Multi-Tenancy.
The decision puts Netris at the core of Visionbay's effort to manage the networking demands of large-scale AI infrastructure. GPU clusters for AI workloads rely on multiple networking layers, and operators must coordinate changes across them when customers are added, resources are reassigned, or capacity is resized.
Visionbay reviewed technical and procurement options before choosing Netris for its first cluster and for future systems based on newer GPU generations. The selection was driven by the need for automated configuration, hardware-enforced separation between tenants, support for different networking fabrics, and the ability to scale operations as the business grows.
Visionbay is an NVIDIA Cloud Partner and one of Asia's larger GPU cloud providers. Its role within Foxconn connects AI computing infrastructure with the parent group's manufacturing and supply chain operations as it builds what it calls AI Factory services for enterprise users.
Network challenge
Traditional data centre methods were not suitable for the architecture Visionbay is building. In a GPU cluster, each server can connect across north-south Ethernet, east-west networking, NVL72 systems, edge infrastructure, host networking, and data processing units, creating a more complex environment than a standard enterprise network.
That complexity makes manual configuration difficult at scale. Changing settings by hand across hundreds or thousands of switches can slow deployments and increase the risk of outages or traffic leakage between customers.
Visionbay also ruled out building its own network management software. According to the announcement, shifts in reference architectures, GPU generations, and workload types would require frequent redevelopment of internal tools.
The Netris system is designed to automate changes across these layers and translate cloud-style functions such as virtual private clouds, peering, elastic IPs, load balancers, and network address translation rules into network configurations. The platform also supports hard multi-tenancy enforced in networking hardware, which matters for customers that need workload separation for compliance or data sovereignty reasons.
Broader roadmap
The standardisation decision goes beyond a single installation. Visionbay plans to use Netris as the network automation foundation for future GPU cluster deployments, keeping the software in place as newer hardware generations are introduced.
That approach reflects a wider issue in AI infrastructure, as operators try to avoid rebuilding operational software whenever chip and system designs change. Networking has become a critical part of that challenge because utilisation rates and customer provisioning depend on how quickly operators can reassign compute resources without interrupting services.
Netris said its recent deployments have been concentrated among neocloud providers, AI factories, and sovereign AI cloud operators. The company positions NAAM as a category focused on automating network operations, abstracting network complexity, and separating customer environments across physical infrastructure.
For Visionbay, the appeal also included local support in Asia and the ability to work in local languages, according to the announcement. Those factors can matter in long deployment cycles that require on-site engineering and close coordination with hardware suppliers and cloud teams.
"Netris NAAM is essential infrastructure for any GPU cluster at AI factory scale," said Neo Yao, Chief Executive Officer, Visionbay.ai.
"Our collaboration with Netris supports the operational scalability and flexibility required for next-generation AI Factory infrastructure," said Yao.
The deal also highlights Taiwan's role in the regional buildout of AI computing capacity. The island already holds a central position in semiconductor manufacturing, and cloud and infrastructure groups are now investing more heavily in local compute facilities tied to AI model development, training, and inference.
Netris described the Visionbay project as part of that trend, particularly in sovereign AI deployments where governments and enterprises want data and computing resources to remain within specific jurisdictions. In those settings, the ability to separate tenants at the hardware layer has become a selling point for infrastructure operators.
"Asia is leading the sovereign AI buildout, and Visionbay is at the center of it with Taiwan's largest GPU cluster and AI supercomputing center," said Alex Saroyan, Chief Executive Officer and Co-Founder, Netris.
"Visionbay is helping define what next-generation AI infrastructure looks like in the region, and we are honored that the Netris NAAM platform is the network automation foundation behind their AI cloud," said Saroyan.