⬡viabandwidthGPU Compute
MI350X · GPU model hub

MI350X GPU cloud providers

Which GPU cloud providers offer MI350X?

19 providers in viabandwidth's directory advertise MI350X availability. The list below sorts direct operators first (own ASN, network-verified), then marketplaces and resellers. Each provider page shows claimed locations, pricing where published and any third-party hyperscaler capacity source.

Data updated September 2026. We list a provider here only if its own website advertises MI350X availability. We do not infer model offerings.

$0.15/hrlowest published entry rate
$0.53/hrmedian entry rate
5of 19 publish a rate

An entry rate is the lowest hourly price a provider publishes across its whole GPU range. It is not necessarily the MI350X rate, because operators publish one rate card for every accelerator they sell. Rows marked MI350X only list no other accelerator, so for those the entry rate is the MI350X rate.

ProviderEntry rateScopePublished
Thunder Compute$0.15/hracross its GPU rangeQ2 2026 · not current
cloudrift.ai$0.28/hracross its GPU rangeQ3 2026 · not current
GPU Cloud Providers in Asia-Pacific 2026: H100, H200, and B200 ...$0.53/hracross its GPU rangeQ3 2026 · not current
spheron.ai$0.53/hracross its GPU rangeQ3 2026 · not current
gpu.fm$2.04/hracross its GPU rangeQ3 2026 · not current

19 providers

Showing the top 3
FAQ

Frequently asked questions

How much does MI350X cost per hour?
Most providers do not publish a per-accelerator price, so an exact MI350X rate is only available from the ones that sell nothing else. Of the 19 providers offering MI350X, 5 publish an hourly rate at all, and their entry rates, meaning the lowest hourly rate each one publishes across its whole GPU range, start at $0.15/hr with a median of $0.53/hr. Where a provider lists MI350X and nothing else, that entry rate is the MI350X rate.
How many providers offer MI350X GPU compute?
19 providers in viabandwidth's directory currently advertise MI350X availability, a mix of direct operators (own metal), marketplaces and resellers.
Where can I rent MI350X GPUs?
Provider-claimed geography is sparse for MI350X. Check individual provider pages for pricing and regional availability.
What's the difference between a direct operator and a marketplace?
A direct operator runs its own GPUs on its own ASN, so capacity is theirs. A marketplace surfaces capacity from many independent operators. A reseller fronts hyperscaler capacity (AWS/GCP/Azure). All three are legitimate sourcing paths. The right pick depends on workload, price and reliability needs.