"Dedicated servers are delivered within 24 to 72 hours" is such a standard line in this industry that most people no longer question it. We do. An in-stock dedicated server on cheapserv.com is online within 60 seconds of payment confirmation, the same as a VPS, and there is no trick involved: only inventory, cabling and automation done ahead of time instead of after the order.
Where the days usually go
The traditional dedicated workflow starts when the order arrives. Someone checks stock, a technician pulls a chassis from a shelf, mounts it, runs power and network, a network engineer configures the switch port and the IP allocation, and finally an installer image is pushed. Each step is short; the waiting between them is not. Add a manual fraud review at the front of the queue and a weekend, and 72 hours is an honest estimate.
None of that work depends on knowing who the customer is. So we do all of it before there is a customer.
Inventory that is already racked
Every dedicated configuration we list is physically installed in every location where it is offered: mounted, powered, cabled to its switch port and to the IPMI network, with firmware and BIOS settings already applied. A server waiting for an order draws idle power in the rack rather than sitting in a box. That costs us money for every hour it stays unsold, which is exactly the point: it makes overselling expensive and honesty cheap.
The switch side is prepared the same way. Each port has its VLAN, its rate limit and its DDoS protection profile configured in advance; a public IPv4 address and a /64 IPv6 block are reserved for every slot. Provisioning a server therefore never touches the network configuration. It only records which customer now owns a slot that was already fully wired.
Imaging in seconds, not minutes
Installing an operating system over the network normally means downloading an image from a central repository. Across a continent, that alone can take longer than our entire delivery time. We keep an image cache in every rack: a storage node that holds the current build of every OS we offer, refreshed nightly. The install path is short and always the same:
- The order is confirmed and a slot is assigned.
- Through IPMI, the server is set to boot from the network and power-cycled.
- It PXE-boots a minimal environment from the rack cache, which partitions the NVMe drives and writes the chosen image from the same cache at local line speed.
- Your SSH key, hostname, network settings and root password are injected, the boot order is restored and the machine reboots into the installed system.
The full cycle, including the two reboots, typically completes in 40 to 55 seconds on Ryzen and EPYC platforms. Custom ISO installs and rescue mode use the same IPMI path, so you can reinstall or boot your own media from the dashboard at any time without opening a ticket.
Why you see "Sold out" so often
Because each configuration is physically racked, stock on the dedicated page is a count of real machines, not a promise to buy one later. When the last DS-7950X3D in Frankfurt is taken, the plan shows Sold out until the next batch is installed. We could keep the button green and quote "3 to 5 business days" like everyone else, but that would turn a 60-second product into a 72-hour one for the people who need it most.
Sold-out plans return as batches land, usually within days, and popular configurations get restocked first. If you cannot wait, a comparable plan is often available in another location at the same price; the locations page shows what each site carries.
Custom builds are the exception
Not everything fits in a pre-racked slot. A specific RAID card, 24 drives, a particular NIC or more than 1 TB of memory is a custom build, and for those we quote individually and deliver within 3 business days. The difference is visible up front: custom builds are ordered through support, never through the instant checkout, so a 60-second expectation is never attached to a 3-day product.
When hardware fails
The same principle covers repairs. Every location keeps an on-site spare pool sized against the installed base: drives, memory modules, power supplies and a few complete chassis per platform. A failed component is replaced within 4 hours of the fault being confirmed, and a full chassis swap keeps your existing drives whenever the failure allows it. Hardware replacement is part of the SLA, not a paid extra.
Pick an in-stock configuration on the dedicated servers page, pay it from your balance, and watch the dashboard. From the click to an SSH prompt, the clock rarely passes one minute.