AI / GPU

NVIDIA GPU servers, ready for AI.

Delivered with Ubuntu 24.04, CUDA, cuDNN and the major machine-learning frameworks preinstalled, with full SSH access.

PyTorchTensorFlowJAXCUDA 12.2cuDNNJupyterLab

GPU A4000

1 × RTX A4000

16 GB VRAM

CPU: 16 cores · AMD EPYC
Memory: 128 GB DDR5
Storage: 2 TB NVMe

⁦$299.99⁩/mo

GPU A100NVLINK

2 × A100

2 × 80 GB VRAM

CPU: 32 cores · AMD EPYC
Memory: 256 GB DDR5
Storage: 4 TB NVMe

⁦$1,199.99⁩/mo

GPU H100NVLINK

4 × H100

4 × 80 GB VRAM

CPU: 64 cores · AMD EPYC
Memory: 512 GB DDR5
Storage: 8 TB NVMe

⁦$2,899.99⁩/mo

Ready on delivery

Start working right away.

You get full SSH access with nothing else installed on your server. Check the GPU with two commands and start training.

verify.sh

$ nvidia-smi --query-gpu=name,memory.total --format=csv
name, memory.total
NVIDIA H100 80GB HBM3, 81559 MiB
$ python -c "import torch; print(torch.cuda.is_available())"
True