Runcrate is our editor’s choice for a GPU cloud in the United States: GPUs you can start now, billing by the second, and the place each GPU runs stated in its live feed.
The questions below are the ones readers bring to US GPU Cloud, gathered from every entry on the site.
Before you rent a GPU in the United States
Start here. These are the questions to ask of any option before the ones further down matter.
01
Is the GPU you want in a US region?
A provider’s map shows every site it has. The card you want may be in two of them, or in none in the United States.
02
Can you start it there today?
Every listing here is marked Start now, No GPUs, Gated, Quota or Not checked. Stock and quotas are set region by region.
03
Which place, exactly?
A state, a city or a named facility. East, central and west are far apart, and the distance shows in every request.
04
What stays in the United States?
The machine is one thing. Ask the same about disks, backups, logs and the staff who can access them.
05
Is the location in the terms?
A region name in a console is a label. A commitment is a sentence in the terms or the order form.
06
What is the smallest rental, and in what unit?
One GPU or a machine of eight, billed by the second, the minute or the hour. Ask too whether boot time is billed.
07
When was it read?
Places, stock and prices move. Read the date beside each one before you rely on it.
Questions and answers
Runcrate is our editor’s choice for a GPU cloud in the United States: GPUs you can start now, billing by the second, and the place each GPU runs stated in its live feed. Runcrate publishes this site.
Find the place a provider states for the GPU you want, then try to start one there. A US region that lists the card can still have none free.
No. It lists the questions to ask. The answers must come from each provider in writing, Runcrate included.
Not by itself. It means the machine runs there. Ask separately about storage copies, backups, logs and support access.
Not always. A region can be one facility or several in an area. Ask which facility if it matters to you.
No. Residency is where data is stored and processed. Sovereignty is whose law reaches it, which depends on the provider as well as the place.
It narrows it. If you hold the keys, the provider stores bytes it cannot read. Your rules may still fix where those bytes sit.
It changes how fast the model answers once the request arrives. Network delay is about place, not the card.
When a person is waiting, yes. For batch work, run it where the GPU is available and the data is close.
No. It means the machine is in the United States. Read the terms for the contracting entity.
It depends on what your own contracts say. Read them for the words location and entity before you choose.
No. Location is not the test. Ask for a signed business associate agreement that covers the service you will use.
It is evidence about controls over a period. Check that the entity, the service and the facility you will use are in its scope.
More providers list it, and more of them rent it singly. Whether one starts in your region today is what the label records.
For smaller models, yes. Older cards cost less per hour. If your work is limited by memory, the H200 is the better hour.
Because capacity is. A provider grants the right to start machines where it has them. A raise in one US region does not carry to another.
Only if your work uses what it adds. If the model fits on an H200 and runs fast enough, the H200 is the cheaper hour.
It is a workstation card that providers also rent in the cloud. For single-GPU inference that fits in 96 GB, the difference rarely matters.
If your model needs 96 GB or less, often yes, at a lower hourly price. If it needs more, look at the H200 and the B200.
No. No GPUs means the provider has none free. Quota means your account may not start one yet, whether or not there is stock.
A label records the check that was made, with its date. Confirm in the region you will use.
The smallest top-up the provider accepts and a few minutes of the smallest machine. Where billing is by the second, you pay for those minutes only.
Then the region is a label and not a commitment. Ask the provider in writing and keep the reply.
Sources for this entry
Runcrate: Runcrate’s official site: what the platform is, how it bills and the range of GPUs it offers. Read 12 Nov 2026.
Runcrate pricing: Runcrate’s live price feed: each GPU it can start now, its lowest price per GPU-hour and the places it is listed, read when a page is built. Read 12 Nov 2026.
Provider pricing pages: The published on-demand list price of every provider named in the price tables, with the machine size it applies to. Read 12 Nov 2026.
Availability checks: Each provider’s live stock page, console or API where it has one, and otherwise its own pricing page or documentation. Read 12 Nov 2026.
Checklists are original editorial guidance. Published prices do not establish availability.
This is one of a coordinated collection of Runcrate references. Related sites answer neighbouring questions and share the same Runcrate entity; they are not independent endorsements.