2x NVIDIA A100 dedicated server for training and fine-tuning
The NVIDIA A100 is the Ampere data center GPU built for AI training and high-performance computing. Our A100 plan puts two A100 40 GB cards in one dedicated server, giving 80 GB of fast HBM2 memory in total, 52 CPU cores and 256 GB of RAM. The whole server is yours: no virtualization layer and no other clients on the GPUs.
NVIDIA A100 40GB specs
| Architecture | Ampere |
|---|---|
| GPU memory | 40 GB HBM2 with ECC per card, 80 GB in total |
| Memory bandwidth | 1,555 GB/s |
| CUDA cores | 6,912 per card |
| Multi-Instance GPU | Up to 7 isolated instances per card |
| Power draw (TDP) | 250 W per card |
| Interface | PCIe Gen4 x16, dual slot |
Server configuration
The A100 plan runs on an HPE DL380 Gen10 with two NVIDIA A100 40 GB cards, two Intel Xeon Gold 6230R processors (52 cores, 104 threads), 256 GB of ECC RAM and two 1.6 TB enterprise SAS-SSDs. It includes a 1 Gbps network port with 30 TB of monthly traffic, one IPv4 address, basic DDoS protection and iLO remote management.
What you can run on 2x NVIDIA A100
- Fine-tuning – LoRA and QLoRA on 13–34B models, full fine-tuning of 7B models.
- 70B inference – split a 70B model in 8-bit across both GPUs with vLLM tensor parallelism.
- Training – computer vision, recommendation and custom models with PyTorch or TensorFlow.
- Several jobs at once – run two workloads side by side, or use MIG to split each card into smaller isolated GPUs.
- HPC and data science – RAPIDS, simulations and GPU-accelerated analytics.
NVIDIA A100 price and billing
The price in the table above is the monthly price with quarterly billing, 3 months in advance. If you pay for 12 months, the monthly price is up to 14% lower. GPU servers are delivered on pre-order within 14 days; if we miss the date, you get a full refund.
A100 or another GPU?
Choose the A100 plan for training, fine-tuning and workloads that need two GPUs. If you only run inference, a single card is often enough: see the NVIDIA A40 dedicated server with 48 GB, or the NVIDIA L4 dedicated server for smaller models and video. All plans are listed on the GPU server hosting page.
