
Enterprise GPU Dedicated Server - A100
Best price now
Compared across 4 stores. Prices are checked against each store and labelled with when.
Ask our agent about this product
It checks today's price at every store, compares the alternatives and answers in about a minute.
Price facts
- Last seen at $8,075.00 refurbished at ServerOrbit.com on Nov 15.
- Checked at 4 stores.
Our take on the Enterprise GPU Dedicated Server - A100
A dedicated bare-metal server from Database Mart built around an Nvidia A100 GPU with 40GB of HBM2 memory, paired with a dual 18-core Xeon E5-2697 v4 setup and 256GB of RAM. Storage is split across a 240GB SSD, 2TB of NVMe and 8TB of SATA, and the machine comes with one dedicated IP and a 100Mbps to 1Gbps uplink. It suits teams running AI training, inference or heavy data workloads who want a whole A100 to themselves rather than a shared slice. Against cloud GPU instances, the appeal is full-machine isolation and large local storage; the trade-off is older Xeon cores and a single A100 instead of the multi-GPU nodes larger providers offer.
- Full Nvidia A100 with 40GB HBM2, not a shared slice
- 256GB RAM supports large models and datasets
- 2TB NVMe plus 8TB SATA for fast scratch and bulk storage
- Dual 18-core Xeon E5-2697 v4 gives 36 cores and 72 threads
- Dedicated IP included
- Uplink scales from 100Mbps to 1Gbps
Enterprise GPU Dedicated Server - A100 prices at every store
- ServerOrbit.comRefurbished
In stock
$8,075.00Free shippingChecked 8 h ago - Hard Disk DirectRefurbished
In stock
$9,258.00Free shippingChecked 8 h ago - CompSourceNew
Out of stock
$11,601.56Free shippingChecked 8 h ago - ServerMonkey.comRefurbished
Out of stock
$12,000.00+ $117.00 shippingChecked 8 h ago
Enterprise GPU Dedicated Server - A100 price history
Lowest price we've seen: $11,601.56 on Nov 15 at CompSource. Tracked across 1 store since Nov 15.
About the Enterprise GPU Dedicated Server - A100
This is a dedicated GPU server built around an Nvidia A100 graphics card with 40GB of HBM2 memory. It is powered by dual 18-core Intel E5-2697v4 processors and has 256GB of RAM.
Storage combines a 240GB SSD, a 2TB NVMe drive and an 8TB SATA drive. The server includes one dedicated IP address and a network connection of 100Mbps to 1Gbps.
- GPU
- Nvidia A100, dedicated, 40GB HBM2
- CPU
- Dual 18-core Intel E5-2697v4
- RAM
- 256GB
- Storage
- 240GB SSD + 2TB NVMe + 8TB SATA
- Bandwidth
- 100Mbps to 1Gbps
- IP address
- 1 dedicated IP included
Questions and answers
- Is one A100 enough for AI training?
- For fine-tuning mid-sized models, inference and single-GPU training runs, yes. Training large models from scratch usually needs several GPUs, and this server holds one A100.
- How does the A100 here compare with an H100?
- The A100 with 40GB HBM2 is the previous generation. An H100 offers more memory and faster compute, but costs considerably more per GPU.
- What is the 40GB HBM2 for?
- It is the GPU's own high-bandwidth memory. It holds model weights and activations, so 40GB caps the size of the models and batches you can fit on the card.
- What storage does the server include?
- A 240GB SSD, a 2TB NVMe drive and an 8TB SATA drive. Use the NVMe for datasets and checkpoints and the SATA disk for bulk storage.
- Is the bandwidth enough for large datasets?
- The uplink runs from 100Mbps to 1Gbps. At the low end, moving multi-terabyte datasets takes a long time, so confirm the speed you get before relying on frequent large transfers.
- Is it a shared or dedicated machine?
- It is a dedicated server with a dedicated GPU and one dedicated IP, so no other customer shares the hardware.
- Where is the Enterprise GPU Dedicated Server - A100 cheapest?


