SupermicroAS-5126GS-TNRT-01-G2

A 5U chassis with space for eight GPUs that ships with two NVIDIA RTX PRO 6000 Blackwell cards, two AMD EPYC 9355 and 1.5 TB of memory, so you can start small and grow without changing machine.

Ask for price and lead timeSee the configuration
Supermicro AS-5126GS-TNRT-01-G2Rack 5U · Gold Series G2

The exact configuration

Official datasheet

Front view diagram: hover over each component.
Processor2× AMD EPYC 935532 cores · 3.55 GHz
GPU2× NVIDIA RTX PRO 6000 Blackwell Server Edition
Memory1.5 TBDDR5-6400
Storage1× 960 GB M.2 NVMe
Network1× dual 10 GbE RJ45
Form factorRack 5U
Warranty
3 years parts and labour
On-site engineer
Optional: next business day, for 3 years

Start with two GPUs and reach eight in the same chassis

Two RTX PRO 6000 cards give you 192 GB of GPU memory, enough to serve a model of around 70 billion parameters at 8 bits to a department or to fine-tune a smaller one on your own data. The chassis takes up to eight double-width GPUs wired straight to the processors, with no PCIe switch in between, and all six power supplies come fitted, so going from two cards to four or eight is a matter of slotting them in and validating the firmware. The 1.5 TB of DDR5 earns its place here too, because it lets you hold the fine-tuning datasets in memory instead of going to disk at every step.

The weak spot of the standard build is networking: a single 10 GbE adapter is fine for management and a pilot, but falls short as soon as your data lives on shared storage or you want to join two nodes. And if you already know you will need eight GPUs from day one, the SYS-422GA-NRT-02-G2 comes with them fitted and the networking sorted.

What we do

We install it with Ubuntu or RHEL, the NVIDIA GPU Operator and vLLM, or with Slurm if several people will be queuing training jobs on it. When the time comes to expand, we order compatible GPUs, update the BIOS and the BMC (the remote management controller) and revalidate the machine with your workload. If the Gold configuration does not quite fit, we build it as a custom server on the same chassis.

What your data centre needs

It is 5U and 786 mm deep. The official datasheet lists six 2,700 W Titanium power supplies in 4+2 redundancy, sized for a chassis full of GPUs even if it carries two today, so it pays to plan sockets and PDU for the full machine from the start. Cooling is by air, with up to ten high-performance fans; with two GPUs the room barely notices, with eight the heat coming out of the back is a different story.

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.
SYS-212GB-FNR-01-G22x RTX PRO 6000 · 512 GBThe most compact: 2U for a two-GPU service.
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 AS-5126GS-TNRT-01-G2

How many more GPUs can I add?

The chassis takes up to eight double-width GPUs, so you can add six more RTX PRO 6000 Blackwell Server Edition cards. Supermicro's compatibility list also includes the NVIDIA H200 NVL and the AMD Instinct MI350P, although mixing vendors in one machine complicates the software stack and we rarely recommend it.

Which AI model fits in 192 GB of GPU memory?

A 70-billion-parameter model takes about 140 GB at 16 bits, which fits but leaves little room for context, and about 70 GB at 8 bits, which is how we usually serve it. Beyond that size, or with many users at once, you need more GPUs.

Is the standard networking enough?

For managing the machine and running a pilot, yes. It has a dual-port RJ45 10 GbE adapter, which falls short if models and data are read from shared storage or if you want to join two nodes. In that case we build it to order with a faster card, or suggest the SYS-422GA-NRT-02-G2, which already has 200 and 400 GbE.

Is the AS-5126GS-TNRT-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