AMD EPYC dedicated servers scale virtualization, databases, and analytics across dozens of server-class cores with wide memory bandwidth. Each server is single-tenant, with hardware RAID NVMe storage, a 10 Gbps uplink, and 100 TB of free data egress a month.
Up to 128 cores and 256 threads per socket, enough to replace a rack of servers with one machine.
Pay with 12+ cryptocurrencies, including Bitcoin, Ethereum, and USDC.
24/7 support via chat, email, or phone, with a dedicated point of contact at no extra cost.
Built for dense virtualization, in-memory databases, and AI inference at scale.
Log in to the Cherry Servers Client Portal using your credentials, or create an account if you are new.
Choose the location closest to your users and keep round-trip latency low.
Choose an instant EPYC plan to deploy quickly, or build your own custom server. Pick the ideal core count, the NVMe disks, and the RAID level for your workload.
Install a prebuilt Ubuntu, AlmaLinux, Debian, or Windows Server image, or mount your own ISO through the IP KVM console.
Add public and private IP addresses, enable IPv6, and link your servers to a private VLAN for internal traffic.
Confirm your specs and pick from our flexible billing cycles. Pay by card, crypto, or bank transfer, then deploy the server.
Access your server via SSH or RDP using the credentials issued during deployment, then start running your production workloads.
Two Gen5 NVMe disks provided by default on hardware RAID, with up to ten available depending on the processor.
ECC memory ships are standard on every server. Single-bit errors are caught and corrected in hardware.
100 TB of outbound traffic per server each month, with unmetered inbound.
A private uplink of up to 10 Gbps on every server, ready for sustained transfers.
Network-layer DDoS filtering on every server, blocking volumetric floods at no added cost.
Full root control over the OS and firmware, with no limits. Out-of-band KVM opens a BIOS-level console for recovery, even with no OS installed.
Single-tenant hardware under a 99.97% uptime SLA, with no noisy neighbors.
Every plan includes a free 100 GB backup volume, expandable to 2 TB. Separate network storage keeps backups safe when a primary disk fails.
Up to twelve DDR5 channels per socket move data to the cores in parallel. Memory-bound databases, analytics, and dense VM hosts run at full speed instead of queuing for bandwidth.
Every socket has all 128 PCIe Gen5 lanes. Attach NVMe arrays, fast NICs, and accelerators at once, with lanes to spare rather than a single shared bottleneck.
You get the whole machine. AMD SEV encrypts each VM’s memory with its own key: guests cannot read each other’s memory, and the server belongs to you alone.
EPYC holds high performance per watt, so it runs cool enough to keep its clocks under a full, continuous load. Long jobs finish at the rate they started, with no thermal throttling.
A private cloud, a container platform, or a VDI estate runs on a single machine. One EPYC host consolidates dozens of guests, with the entire server dedicated to you. No other customer workloads competing for the same cores, memory, or storage.
Data held in memory returns results with no disk I/O. EPYC provides up to 1152 GB of memory and wide DDR5 bandwidth to keep stores, caches, and columnar analytics resident, with cores to enable fast querying.
Serve models up to 13B parameters on the processor alone, from recommendation engines to fraud scoring, no GPU required. One socket runs several inference jobs simultaneously, each with its own cores and memory.
One server fields thousands of simultaneous users. A web tier, API fleet, or microservice mesh runs on multiple cores at once, with enough memory to hold sessions and connection pools at peak without requiring a swap to disk.
A pre-configured EPYC plan deploys in 12 minutes on average, while a custom build, which lets you pick the processor, RAM, and storage yourself, takes an average of 24 hours to deploy, depending on the location.
AMD EPYC servers run Ubuntu, Debian, AlmaLinux, openSUSE, and Windows Server from prebuilt images. You can install any other x86-64 system from your own ISO, mounted over the IP KVM console.
Yes. A custom build lets you select the EPYC model, RAM up to 1152GB, NVMe storage and RAID level, and optional GPU accelerators. Pre-configured plans ship with a fixed spec you can deploy right away.
No. Bare-metal hardware is fixed once deployed, so you cannot dynamically add cores, RAM, or storage to a running EPYC server. To scale up, you can either choose a custom bare-metal instance and specify your hardware requirements before deployment, or deploy a new EPYC server to the spec you need and migrate your workload using Terraform or Ansible.
Use our help center
Our customers consistently highlight speed, expertise, and personal attention from our team.
Fill out the form, and our team will get back to you!
