NVIDIA Data Center GPU A30
The NVIDIA A30 is a versatile dual-slot enterprise accelerator built on the Ampere GA100 architecture, optimized for mainstream enterprise AI inference, conversational AI natural language processing, and high-density virtualization. Equipped with 24GB of high-bandwidth HBM2 memory delivering 933 GB/s throughput, it operates within an economical 165W power envelope. It supports Multi-Instance GPU (MIG) partitioning into up to 4 hardware-isolated instances with dedicated memory and compute resources. Because it delivers high HBM2 bandwidth at half the power of an A100, it is widely used in mainstream 1U and 2U enterprise servers. Second-hand buyers should check HBM2 memory health and verify passive airflow.
Also written as: NVIDIA A30 PCIe, A30 24GB, Tesla A30
Specifications
| Brand | NVIDIA |
|---|---|
| Family | Data Center GPU |
| Form factor | PCIe dual-slot full-height full-length |
| Released | Apr 2021 |
| End of sale | Mar 2024 |
| Gpu memory (GB) | 24 |
| Tdp (W) | 165 |
Generations
- Previous generation: NVIDIA Data Center GPU T4
- Next generation: NVIDIA Data Center GPU L4
Reference prices (last 14 days, updated 2026-10-27)
Refurbished: USD 3,400 (1 sighting, Oct 2026)
Used: USD 2,400 (1 sighting, Oct 2026)
All sightings
| Seller | Condition | Price | Note | Seen |
|---|---|---|---|---|
| serversupply.com | refurbished | USD 3,400 | Refurbished passive PCIe card with 90-day warranty | 2026-10-27 |
| ebay.com | used | USD 2,400 | Clean pull tested with 100% health in Lenovo SR650 | 2026-10-27 |
Prices seen on public listings on the dates shown, before tax and shipping. Not an offer.
Buying a used A30: what to check
Inspect HBM2 memory with DCGM diagnostics, verify passive server cooling compliance, check that the 8-pin auxiliary power connector and cable harness are intact, and verify MIG partition functionality.
FAQ
What memory type does the NVIDIA A30 use?
It uses 24GB of high-speed HBM2 memory with 933 GB/s bandwidth.
What is the TDP of the NVIDIA A30?
It operates at a power-efficient 165W TDP, making it compatible with dense server chassis.
Does the A30 support Multi-Instance GPU (MIG)?
Yes, it can be partitioned into up to 4 independent hardware MIG instances.
Buying or selling A30?
New, refurbished or used. Tell us the quantity and condition and we connect you with a vetted reseller in your region, usually within one working day.
Sources
- NVIDIA A30 Tensor Core GPU Product Brief, checked 2026-10-27