Supermicro Gold Series G2

Supermicro servers, pre-configured and installed across Europe

29 configurations that Supermicro has already assembled and tested, from a 1U node for a remote office to an 8U system with eight B300 GPUs. We install them in your data centre with the operating system and everything on top up and running, and support comes from the same people who installed them.

Ask for price and lead time or see all 29 configurations
Supermicro SYS-822GS-NB3RT-01-G2 with NVIDIA HGX B300 SYS-822GS-NB3RT-01-G2 · 8 NVIDIA B300 GPUs · 8U

By family

Rack servers SupermicroRack servers12 modelsAMD EPYC 9004, 9005, 4004 and 4005 series and Intel Xeon 6, in 1U and 2U, plus GrandTwin and MicroCloud multi-node systems for virtualisation, databases and hosting.See the familyAI and GPU servers SupermicroAI and GPU servers8 modelsFrom two RTX PRO 6000 Blackwell cards up to the eight-GPU HGX B300, plus the Super AI Station GB300 in a tower chassis.See the familyStorage servers SupermicroStorage servers2 modelsTwo 2U chassis for Ceph, backup and archive: twenty-four 22 TB drives, or a mixed layout with NVMe.See the familyEdge servers SupermicroEdge servers4 modelsCompact systems for the factory floor, the shop or the office comms cabinet, including a fanless unit with Jetson Orin NX.See the familyBusiness computers SupermicroBusiness computers3 modelsCompact mini towers with Ryzen or EPYC 4000 series processors for the desk.See the family

What we build with these servers

These are the four projects we're asked for most, with the Gold Series hardware we'd use for each.

Artificial intelligence

Training and fine-tuning language models in your own data centre, inference for large numbers of users, and RAG over data that can't leave your network. It's the alternative to renting cloud GPUs when the project will run for more than a year. How we build it

ArchitectureGPU nodes, a control node and shared NVMe storage
HardwareSYS-822GS-NB3RT-01-G2 or AS-8126GS-NB3RT-01-G2 with HGX B300 for training; SYS-422GA-NRT-02-G2 with eight RTX PRO 6000 for inference; AS-1115CS-TNR-01-G2 to feed data to the GPUs
Network800 GbE or XDR800 between GPU nodes, which the Gold Series HGX B300 systems already include
SoftwareOpenShift AI, RHEL AI or Kubernetes with the NVIDIA GPU Operator, and vLLM to serve the models

High-performance computing

Scientific clusters, fluid dynamics, genomics or risk engines: lots of cores, a low-latency network and a job scheduler that spreads the work properly.

ArchitectureControl and login node, identical compute nodes and parallel storage
HardwareAS-2115GT-HNTR-01-G2, four EPYC nodes in 2U with 100 GbE each, so it already has the network the cluster needs
NetworkRoCEv2 (RDMA over Ethernet, which cuts latency between nodes) at 100 GbE
SoftwareSlurm and OpenHPC on Rocky Linux or RHEL, with Lustre or BeeGFS

Storage with Ceph

Replacing NetApp or Dell PowerStore arrays with your own Ceph cluster, which serves block, S3 object and file storage from the same hardware, and petabyte-scale backups with no per-terabyte licensing.

ArchitectureAt least three nodes with 3× replication, or six with 4+2 erasure coding
HardwareASG-2015S-E1CR24L-01-G2, with 24 × 22 TB drives in 2U, or AS-2015CS-TNR-01-G2, which mixes drives and NVMe
NetworkTwo separate networks, public and replication, at 25 or 100 GbE depending on the model
SoftwareCeph on RHEL or Ubuntu, Ceph built into Proxmox VE, or Rook inside OpenShift

Private cloud and moving off VMware

For teams leaving VMware who need more than a hypervisor: a private cloud in your data centre with multiple internal tenants, quotas, images and an API, running virtual machines, containers and bare-metal servers on one platform. How we build it

ArchitectureThree highly available controllers, compute nodes and Ceph for storage
HardwareAS-1126HS-TN-01-G2, SYS-222H-TN-01-G2 or AS-2126HS-TN-01-G2 as compute nodes, depending on how many cores you need
Network2× 25 GbE minimum, software-defined networking with OVN (OpenStack's virtual network layer) and a separate management network
SoftwareRed Hat OpenStack Services on OpenShift, Canonical or SUSE OpenStack, or Proxmox VE if a hypervisor is enough

Why buy from us

We deliver the machine installed, and if it ever fails you call the same team that put it into service.

We rack it, configure remote management (BMC) and install RHEL, SUSE, Ubuntu or Proxmox, with whatever runs on top already installed and tested: a Ceph cluster, OpenStack, Kubernetes with the NVIDIA GPU Operator, or Slurm.

When something fails you speak to the people who know that machine because they commissioned it, and if you need it there is 24×7 on-call cover with a contractual SLA. It's the same independent support we provide for IBM Power, Linux and Ceph estates.

If you need different memory, drives or networking, we start from the same base model. Tell us what you need in the custom server configurator and we'll call you back within 24 hours with a proposal.

We are an IBM Business Partner and a Red Hat, SUSE and Canonical partner, which is what counts when the server has to slot into an environment with existing systems and licences.

We work across Europe in all three languages, support included, so if your company has offices in Madrid and Brussels, the same team looks after both.

What the delivery includes

From the factory

  • Configuration assembled and tested by Supermicro
  • 3-year warranty on servers; 1 year on edge systems, desktops and the SYS-C542i

From us

  • Rack mounting
  • Remote management (BMC) and up-to-date firmware
  • Operating system and the stack that runs on top
  • Testing with your workload before it goes into production

If you need it

  • 24×7 on-call support with SLA
  • On-site engineer next business day for 3 years

Frequently asked questions

How long does a Gold Series server take to arrive?

Less time than a build-to-order server, because the configuration is already assembled and tested. We confirm the exact lead time in writing with the quote.

What warranty do they come with?

Rack, AI and storage servers come with 3 years parts and labour; edge systems, desktops and the SYS-C542i tower come with 1 year. All of them can be covered by an on-site engineer next business day for 3 years, and on the Super AI Station GB300 that service is mandatory and always included in the quote.

How much does a Supermicro Gold Series server cost?

It mainly depends on the cost of memory and drives at the time of ordering. We give you a fixed figure in the quote, together with the lead time.

What if none of the configurations fits?

Then the server is configured to order on the same base model, with the memory, drives and networking you need, and it takes the lead time of a build-to-order system. You can start with the custom server configurator or tell us about the workload directly.

Which one fits what you're building?

Tell us about the workload and the data centre it's going into, and we'll send you a quote with a confirmed lead time and what it would take to get it running.

Ask for price and lead time WhatsApp +34 91 198 02 43