Case Study: CyberAgent achieves 4x cost-performance and accelerated AI training with NetApp (AFF A800 & Trident)

CyberAgent accelerates AI with a hybrid cloud

CyberAgent, a major online advertising, gaming, and media company known for its AI-driven services (including its Kiwami Prediction AI for ad creatives), faced rapidly growing AI development demand and escalating cloud costs. The team needed a multi-tenant, container- and Kubernetes-friendly AI learning environment that could fully leverage the latest GPUs and deliver high-performance persistent storage for concurrent training workloads.

They built a hybrid on‑premises AI platform using NVIDIA DGX A100 servers and NetApp AFF A800 storage with Trident (CSI) to provide dynamic persistent volumes and seamless Kubernetes integration, while linking to public clouds via Cloud Volumes ONTAP. The result: roughly 2x GPU performance compared with prior cloud setups at half the projected five‑year cost—delivering about 4x the cost performance—while improving developer productivity and data mobility across cloud and on‑prem environments.

CyberAgent

Masaya Aoyama

Infrastructure Engineer/Developer Expert