TAIPEI (Taiwan News) — Foxconn launched Friday the first phase of its Visionbay.ai supercomputing center, bringing Taiwan’s largest Nvidia HGX B300 AI cluster online.
The 5 MW cluster is built on Nvidia HGX B300 systems and has more than 90% of its capacity already leased, per CNA. Foxconn said the facility is designed to support large-scale AI model training and inference as enterprise demand for AI computing grows.
Visionbay.ai — Foxconn’s subsidiary focused on AI infrastructure and cloud platforms — said the cluster uses liquid cooling, a 2N high-availability architecture, and Nvidia Quantum-X800 InfiniBand networking, per UDN. It has also received Nvidia Exemplar Cloud validation for HGX B300 training workloads, making it one of the first 10 platforms worldwide and the second in Asia to earn the certification.
The company said the certification confirms the platform is ready for production-scale AI deployment, helping enterprise customers reduce implementation risks and accelerate AI adoption. It added that demand has shifted beyond GPU ownership to reliable infrastructure capable of supporting AI workloads.
Unlike conventional GPU cloud services, Visionbay.ai said its platform combines AI infrastructure, GPU cloud services, an AI operations platform, and a token factory platform into a single ecosystem. The integrated platform supports model training, fine-tuning, inference, and AI agent deployment.
Visionbay.ai said it plans to expand its AI infrastructure across the Asia-Pacific region using Nvidia Blackwell and Vera Rubin platforms. Leveraging Foxconn’s strengths in hardware, key components, and software platforms, the company said it aims to provide AI infrastructure solutions for enterprise customers.
