We’ve released GPU Virtual Machine with a fully self-service experience, making it easier for developers to access on-demand AI cloud infrastructure powered by NVIDIA HGX H100 and HGX H200.
Designed for speed and control, GPU Virtual Machine enables users to get started in minutes and customize their environment for AI experimentation and pre-production workloads, while delivering enterprise-grade performance and flexibility without long provisioning cycles or upfront commitments.
- On-demand access to GPU resources, with Pay-As-You-Go pricing, second-level billing for precise cost management.
- High-performance, stable infrastructure optimized for AI training and inference, provided through a bundled all-in-one configuration of GPU, CPU, RAM, Storage*, and Public IP. No hidden cost.
- Full VM-level control, including custom networking and root access. NVIDIA GPU drivers are pre-installed in the FPT public image for faster environment setup.
*GPU Virtual Machine currently includes local NVMe storage with each instance. Support for Block storage (Persistent disk) will be released soon.
GPU Virtual Machine is an ideal choice for developers and small AI teams who need immediate GPU access for:
- Long-running experiments
- Multi-GPU LLM training
- Fine-tuning models with custom libraries
- Short-term inference and evaluation
Start GPU workloads in 2 minutes, scale when needed, and stop when the job is done—all through a simple, self-service workflow.
CTA:
