Overview

AMD logo AMD MI350X

The AMD MI350X is a data center GPU on the CDNA 4 architecture, released Q2 2025, with 288 GB HBM3e and 8,000 GB/s memory bandwidth.

High Performance Datacenter

Nvidia logo Nvidia A40

The Nvidia A40 is a data center GPU on the Ampere architecture, released Q4 2020, with 48 GB GDDR6 and 696 GB/s memory bandwidth.

Mid-Range Datacenter

Memory per GPU

6x more on MI350X
AMD logo MI350X
288 GB HBM3e
Nvidia logo A40
48 GB GDDR6

Memory bandwidth

11x higher on MI350X
AMD logo MI350X
8,000 GB/s
Nvidia logo A40
696 GB/s

GPU interconnect

AMD logo MI350X
Infinity Fabric 1,075 GB/s
Nvidia logo A40
PCIe only

Peak performance (FP16)

15x higher on MI350X
AMD logo MI350X
2,300 TFLOPS
Nvidia logo A40
150 TFLOPS

Vendor peak figures. Sparsity and tensor-core conventions differ between vendors.

Precision support

AMD logo MI350X
FP4 FP6 FP8 FP16 BF16 FP32 FP64 INT8
Nvidia logo A40
FP16 BF16 TF32 FP32 INT4 INT8

Highlighted formats are supported by only one of the two.

Architecture

AMD logo MI350X
CDNA 4
Nvidia logo A40
Ampere

Released

5 years newer on MI350X
AMD logo MI350X
Q2 2025
Nvidia logo A40
Q4 2020

Price comparison

Cheapest on-demand

AMD logo MI350X
Not sold on demand
Nvidia logo A40
$0.34 / GPU / hr

Typical rate

10x higher on MI350X
AMD logo MI350X
$4.38 / GPU / hr
Nvidia logo A40
$0.44 / GPU / hr

Median across providers, one vote each, so a large catalog does not outweigh a small one.

Spot floor

12x higher on MI350X
AMD logo MI350X
$4.00 / GPU / hr
Nvidia logo A40
$0.32 / GPU / hr

Full price range

AMD logo MI350X
$4.00 - $4.76 / GPU / hr
Nvidia logo A40
$0.32 - $2.05 / GPU / hr

Lowest to highest listing per GPU-hour, across on-demand, spot, reserved and custom.

Providers listing it

5 more on A40
AMD logo MI350X
1
Nvidia logo A40
6

Heads up: Prices are estimates from published rates for common setups and vary by region and usage; verify with the provider before provisioning.

Offers and availability

How this list works

Order

Each provider appears once per GPU. We pick the one configuration that speaks for that provider, and rank on it in this order:

  1. Availability: in stock, then waitlist, then not reported, then out of stock. A price you can act on outranks a cheaper one you cannot.
  2. Billing type: on-demand first, then reserved, then spot, then quote-only. On-demand leads because it is the only rate you can hold without committing to anything.
  3. Price: the lowest per-GPU hourly rate the provider lists, converted to USD.

The configuration we pick is chosen the same way, availability before billing before price. The rest of that provider's configurations are on the GPU's own page.

Ties sort alphabetically. Reserved rates are not separated by term length, so a three-year commitment can sit above a one-month one: the label beneath each price says which you are looking at. Providers who publish no rate are listed last within their group.

Transparency and funding

  • Ads and sponsors: Paid placements sit at the top of the list and are always labeled as sponsored content. Sponsorship never influences the organic ranking itself.
  • Affiliates: Affiliate links are marked. We may earn a commission if you click them, but commissions never affect the order.
  • Prices: Shown in USD, converted at daily reference rates where a provider publishes in another currency. A month means 720 hours. How we estimate costs.

Similar GPUs

Other comparisons in the same class:

View all GPUs

Our data for AMD MI350X was last updated on Aug. 28, 2026, and for Nvidia A40 on Aug. 26, 2026.