VX Compute is where the large configurations live: up to 512 GB of memory and nearly four terabytes of NVMe on a single machine. These plans exist for the work that will not fit anywhere smaller — an in-memory database, an analytical job that pages to disk on an ordinary server, a migration that has to hold the whole dataset at once.
They come in two shapes. General Purpose balances cores against memory; Memory Optimized doubles the memory for the same core count. Which one you want depends on whether the job is thinking or remembering, and the plan table shows both figures side by side so the difference is visible before you choose.
Like our GPU range, these are billed by the hour rather than the month, which is usually the right way round for a machine this size — the work is often intense and temporary. If yours is permanent, multiply the hourly rate by 730 and compare it against a monthly plan before you commit.
Large Machines, Charged by the Hour
Deploys to Newark, US — choose from every location this plan offers at checkout; some cost more
| Select | Plan | vCPU | Memory | Storage | Bandwidth | |
|---|---|---|---|---|---|---|
| 2 | 8 GB | 120 GB NVMe | 5 TB | $0.15/hour | ||
| 2 | 16 GB | 120 GB NVMe | 5 TB | $0.19/hour | ||
| 4 | 16 GB | 240 GB NVMe | 6 TB | $0.29/hour | ||
| 4 | 32 GB | 240 GB NVMe | 6 TB | $0.36/hour | ||
| 8 | 32 GB | 480 GB NVMe | 7 TB | $0.57/hour | ||
| 8 | 64 GB | 480 GB NVMe | 7 TB | $0.72/hour | ||
| 16 | 64 GB | 960 GB NVMe | 8 TB | $1.14/hour | ||
| 16 | 128 GB | 960 GB NVMe | 8 TB | $1.44/hour | ||
| 32 | 128 GB | 1920 GB NVMe | 9 TB | $2.29/hour | ||
| 32 | 256 GB | 1920 GB NVMe | 9 TB | $2.89/hour | ||
| 48 | 192 GB | 2880 GB NVMe | 9 TB | $3.43/hour | ||
| 48 | 384 GB | 2880 GB NVMe | 9 TB | $4.33/hour | ||
| 64 | 256 GB | 3840 GB NVMe | 10 TB | $4.57/hour | ||
| 64 | 512 GB | 3840 GB NVMe | 10 TB | $5.77/hour |
No plans in this family.
What VX Compute Is For
Capacity Is the Point
These are the large configurations — up to 512 GB of memory and several terabytes of NVMe on a single machine. They exist for workloads that will not fit anywhere smaller: in-memory databases, large analytical jobs, and builds that page to disk on an ordinary server.
Two Shapes
General Purpose balances cores against memory. Memory Optimized doubles the RAM for the same core count. Which you want depends on whether your workload is thinking or remembering — the plan table shows both figures side by side so the difference is visible.
Hourly, Not Monthly
A machine this size is often needed for days rather than months. You are charged for the hours it exists and invoiced afterwards, so a four-day migration or a quarterly batch job costs four days or one job.
Storage That Comes With It
Every plan listed here ships with its NVMe storage included — up to 3.8 TB on a single server, already attached and ready. Nothing extra to provision, and no separate volume to pay for or forget about.
Full Root Access
Your own machine, your own kernel modules, your own tuning. Nothing is managed on your behalf and nothing is locked down, which is usually the reason a workload needs a server this size in the first place.
Deployed in Minutes
Provisioning is automated, so a machine with half a terabyte of memory arrives in about the same time as a small one. There is no procurement step and no waiting on capacity approval.
Before You Order
Work Out the Month Before You Start
An hourly rate on a large machine adds up quickly — multiply it by 730 to see what a full month looks like before you commit to leaving one running. For short work this is much cheaper than a monthly plan; for permanent workloads it may not be, and it is better to know that at the start.
Destroy It When You Are Done
Charging follows existence, not activity. A powered-off VX server still holds its memory and storage allocation and still bills. When a job is finished, snapshot anything you need and destroy the server — that is what stops the meter.
Sizing Honestly
It is cheaper to start on a plan that fits and redeploy than to pay for headroom you never touch. Memory is the figure that usually forces the decision: a job that exceeds it fails or swaps, while spare cores merely sit idle.
FAQs
Have A Question?
If you can't find the answer you are looking for our support is just an email away.
Contact UsHow is this different from Optimized Cloud?
Mostly billing and scale. Optimized Cloud plans are billed monthly and are tuned in four directions for steady, always-on workloads. VX plans are billed hourly and go considerably larger, which suits work that is intense but temporary.
If the server will run permanently, compare the monthly total of both before choosing.
What does it cost for a full month?
Multiply the hourly rate by roughly 730. The plan table shows the hourly figure because that is what you are actually charged; the monthly equivalent is arithmetic rather than a separate price.
Is anything charged when I order?
No. There is nothing to collect up front on an hourly plan. You add a payment method to authorise the order and it is charged when the invoice for your recorded hours is raised.
Can I resize a VX server?
You can move up within the same family — more cores and memory on the same machine — but not across into a different family. Storage attached at deployment stays with the server, so choose the disk size with a little room to grow.
Does stopping the server stop the charges?
No. The allocation is held for you while it is stopped, so it continues to bill. Destroying the server is what ends it. Take a snapshot first if you want to restore the same disk later.
Which one should I pick for a database?
Memory Optimized, in almost every case. Databases are usually limited by how much of the working set fits in RAM, and the Memory plans carry twice the memory per core. Match the memory figure to your dataset and let the core count follow.