# Servers (compute instances)

Source: https://docs.quake.ai/docs/compute/concepts/servers
Markdown: https://docs.quake.ai/docs/compute/concepts/servers.md

---

# Servers (compute instances)

When documentation or the OpenStack CLI says **server**, it means a compute **[instance](/docs/compute/concepts/instances)** on Quake AI: a running virtual machine backed by Nova, not a physical rack server and not a **[server group](/docs/compute/concepts/server-groups)** placement policy.

## Why the word appears

OpenStack Nova names the resource `server` in API paths and CLI commands:

```bash
openstack server create ...
openstack server list
openstack server show <id>
```

Terraform and Ansible modules often mirror that name (`openstack_compute_instance_v2` still creates what Nova calls a server). Quake AI product language standardizes on **instance** in the Console and in customer-facing guides so the noun matches the service you are using.

## On Quake AI

| Surface | Noun you see |
|---|---|
| Console | Instance |
| Nova API / OpenStack CLI | Server |
| This documentation (default) | Instance |
| Ansible `openstack.cloud.server` module | Server (OpenStack resource type) |

A **server group** is a different object: it controls whether instances land on the same hypervisor (affinity) or on different hypervisors (anti-affinity).

## What to read next

- [Instances](/docs/compute/concepts/instances): canonical explanation of compute guests
- [Virtual machines (VMs)](/docs/compute/concepts/virtual-machines): VM vocabulary mapped to instances
- [Compute API reference](/reference/compute/instances-api): Nova `servers` endpoints and fields