What is a VPS?
Last updated: 2026-06-09
A VPS is a virtual private server.
In practice, it is a rented machine in the cloud that gives you your own CPU, memory, disk, networking, and operating system environment.
What humans usually use a VPS for
People rent VPS instances for tasks like:
- hosting websites
- running app backends
- operating game servers
- running bots or automation
- hosting databases or internal tools
- keeping long-running processes online
It is one of the simplest building blocks in cloud infrastructure because you get a real server instead of only a managed platform abstraction.
Traditional VPS workflows
There are already many good VPS and cloud compute products in the market.
If you are technical, or you have developers and infrastructure people, those services can be excellent. Traditional VPS products are strong when you want:
- full control over server choice and configuration
- direct shell access
- custom networking and deployment workflows
- long-running background processes
- infrastructure flexibility without a managed platform layer
The tradeoff for non-technical users
Traditional VPS providers are still built with the assumption that somebody will:
- choose a server size
- pick an image and region
- configure SSH access
- bootstrap packages and services
- keep track of costs
- know when to delete or resize things
That is fine for technical teams.
But if you are not technical, or you simply do not want to operate servers yourself, a VPS can feel like too much manual work even when the provider itself is good.
VPS for agents
An AI agent thinks about a VPS differently than a human operator does.
An agent may want to:
- pick a plan and region for a website or background worker
- create the server immediately after user approval
- fetch a managed SSH connection bundle right after creation
- pass cloud-init through `user_data`
- install packages and bootstrap services automatically
- keep track of whether enough credits remain for the next cycle
- delete the server later when the task is done
The human often just wants the result, not a cloud control panel.
MoonRock VPS
MoonRock VPS is designed for that agent workflow.
Instead of making the user learn a traditional cloud UI, MoonRock lets the agent:
- authenticate once through the normal MoonRock approval flow
- check the workspace credit balance
- list VPS plans
- create a VPS
- fetch the managed SSH credentials for that VPS
- reserve one month of credits up front
- cancel the VPS later and automatically receive the unused portion of the hold back
If the current month finishes, MoonRock captures that held month and tries to reserve the next month. If there are not enough credits to renew, MoonRock terminates the VPS.
MoonRock also enforces two protective rules:
- a VPS cannot be canceled during its first 24 hours
- cancellation usage is billed in whole hours rounded up
Why the credits model matters
MoonRock uses prepaid credits for VPS rentals.
That means:
- the user does not need a complicated billing setup in v1
- the agent can check whether enough credit is available before provisioning
- MoonRock can keep spending bounded
- cancellation can be prorated cleanly within the current cycle
Right now credits are operator-managed, which means the MoonRock admin can add them manually before public billing is live.
If you are not technical
You do not need to know how SSH, cloud-init, server images, or VPS pricing work in order to use MoonRock.
You can simply tell your agent:
Please check MoonRock - It's a cloud platform for agents - https://moonrock.co.il/for-agents
The agent can then:
- authenticate you
- check credits
- choose a VPS plan
- create the server
- install what it needs
- manage or remove it later
Why this matters
For humans, a VPS is often a dashboard plus a shell session.
For agents, a VPS is a programmable infrastructure primitive.
MoonRock is designed to make that primitive easier for the agent to rent, configure, and operate on behalf of the human.