SupermicroSYS-212GB-FNR-01-G2

Two NVIDIA RTX PRO 6000 Blackwell cards, a Xeon 6731P and 512 GB of DDR5 packed into 2U, with 7.68 TB of E1.S storage to keep models and search indexes local.

Ask for price and lead timeSee the configuration
Supermicro SYS-212GB-FNR-01-G2Rack 2U · Gold Series G2

The exact configuration

Official datasheet

Front view diagram: hover over each component.
Processor1× Intel Xeon 6731P32 cores · 2.5 GHz
GPU2× NVIDIA RTX PRO 6000 Blackwell Server Edition
Memory512 GBDDR5-6400
Storage1× 960 GB M.2 NVMe2× 3.84 TB E1.S
Network1× dual 10 GbE RJ45
Form factorRack 2U
Warranty
3 years parts and labour
On-site engineer
Optional: next business day, for 3 years

Inference in 2U for a specific service

This is the machine for rolling out an internal assistant, a search tool over company documentation using retrieval-augmented generation (RAG, where the model answers from your own documents) or an agent workflow serving one department. The 192 GB across its two GPUs serves a 70-billion-parameter model at 8 bits, and the two E1.S drives, the compact NVMe format designed for dense servers, keep the weights and the vector database on the machine itself, so starting up or switching models does not depend on the network.

The price of that density is a single processor, with less system memory and fewer PCIe lanes than the dual-socket models. It takes up to four GPUs, but each card can draw up to 600 W and in 2U that leaves little thermal margin, so if you plan to go beyond two you are better off with a 4U or 5U chassis.

What we do

We set it up with RHEL or Ubuntu, the NVIDIA drivers and vLLM serving the model behind an OpenAI-compatible API, which is what almost every application and agent framework expects. If you already run Kubernetes, we join it to the cluster with the GPU Operator; if not, we leave it running as a standalone server with monitoring and configuration backups in place.

What your data centre needs

It takes 2U at 900 mm deep and, according to the official datasheet, carries two redundant 2,700 W Titanium power supplies. It is air-cooled by five 8 cm fans, so it asks nothing special of the room beyond a cold aisle that works, although it packs a lot of heat into little height and stacking several without checking the rack's thermal load is not a good idea.

Compared with the rest of the family

ModelWhat it hasWhen to choose it
SYS-822GS-NB3RT-01-G2HGX B300 8-GPU · 2 TBThe HGX B300 with Intel; the choice if your standard is Xeon.
AS-8126GS-NB3RT-01-G2HGX B300 8-GPU · 3 TBSame HGX B300 with AMD, 3 TB and a separate storage network.
AS-5126GS-TNRT-01-G22x RTX PRO 6000 · 1.5 TBTwo GPUs today, space for eight and memory to spare.
SYS-422GA-NRT-01-G24x RTX PRO 6000 · 1 TBFour GPUs that MIG splits into up to sixteen instances for several teams.
SYS-422GA-NRT-02-G28x RTX PRO 6000 · 1.5 TBThe most PCIe can do: eight GPUs and 400 GbE networking.
SYS-C542i-11302URTX 50-ready · 32 GBDevelopment tower without a standard GPU; you pick the card.
ARS-511GD-NB-LCC-01-G2Grace + B300 · 748 GBA GB300 node in a tower for teams that fine-tune models without booking a cluster.

See all AI and GPU servers

What the delivery includes

From the factory

  • Configuration assembled and tested by Supermicro
  • Warranty: 3 years parts and labour

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 the full 3 years

How we build it

What runs on top of these servers, explained in detail:

AI inference with Ceph, OpenStack and KubernetesAll our AI services

Support and training for your team

Independent support from the people who installed itLocal LLM and GPU infrastructure courseDeep learning course with PyTorch and TensorFlowRed Hat OpenShift training

Questions about the SYS-212GB-FNR-01-G2

How does it differ from the AS-5126GS-TNRT-01-G2?

Both ship with two RTX PRO 6000 cards. This one fits in 2U, has a single Xeon with 512 GB and comes with 7.68 TB of E1.S storage for the models; the AS-5126GS takes 5U, carries two EPYCs with 1.5 TB and takes up to eight GPUs. For a well-defined service where rack space matters, this one; if you will be adding GPUs, the other.

Does a single processor hold the GPUs back?

For inference, hardly. The Xeon 6731P has 32 cores and connects each GPU over PCIe 5.0 x16, which is enough to keep them busy serving a model. Where it shows is in workloads that preprocess a lot of data on the CPU before handing it to the GPU, such as training on large image datasets.

Can I fit four GPUs?

The official datasheet allows up to four double-width cards, each on PCIe 5.0 x16. But each RTX PRO 6000 can draw up to 600 W, and with four in 2U fed by two 2,700 W supplies the thermal and electrical margins get tight, so if the plan is to go beyond two we suggest the AS-5126GS-TNRT-01-G2 or the SYS-422GA-NRT-01-G2.

Is the SYS-212GB-FNR-01-G2 right for what you're building?

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

Ask for price and lead timeWhatsApp+34 91 198 02 43