Supermicro SYS-A22GA-NBRT NVIDIA HGX B200 8-GPU AI Server – 10U, Dual Intel Xeon 6900P

The Supermicro SYS-A22GA-NBRT is a 10U air-cooled AI server built around the NVIDIA HGX B200 8-GPU board – the same eight-GPU Blackwell design NVIDIA uses in its DGX B200 – paired with two Intel Xeon 6900P processors, the largest Xeon 6 parts with up to 128 performance cores each. It is the Intel-based workhorse of Supermicro's HGX B200 range: lots of host CPU, the fastest memory option on the market (DDR5-8800 MRDIMM) and twelve PCIe 5.0 slots for a full InfiniBand or Ethernet AI fabric.

Choose it when you train or fine-tune large language models, run mixed AI and HPC workloads (the B200 still delivers 296 TFLOPS of FP64) or need serious CPU power next to the GPUs for data preparation, tokenisation and retrieval pipelines. Mineshop.eu supplies it as a configured, quote-based system with EU-wide delivery.

Explore AI & GPU Servers and compare SYS-822GS-NB3RT (HGX B300, Intel) with AS-A126GS-TNBR (HGX B200, AMD).

Key features

• 8× NVIDIA B200 SXM on the HGX B200 board – 1.4 TB HBM3e, NVLink 1.8 TB/s per GPU
• Dual Intel Xeon 6900P with P-cores – up to 128 cores and 504 MB cache per CPU
• 24 DIMM slots – up to 6 TB DDR5-6400 RDIMM or DDR5-8800 MRDIMM
• 10× hot-swap 2.5" PCIe 5.0 NVMe (U.2) + 2× M.2 NVMe with RAID
• 10× PCIe 5.0 x16 LP + 2× FHHL slots – ready for a 1:1 GPU-to-NIC 400G fabric
• 6× 5250 W Titanium PSUs (3+3 redundant), air-cooled 10U chassis

Workloads and use cases

• LLM training and fine-tuning: 1.4 TB of HBM3e and 14.4 TB/s of NVLink bandwidth keep 70B–400B-parameter models on one node for fine-tuning and distributed pre-training.
• AI + HPC in one box: 296 TFLOPS FP64 makes the B200 the better Blackwell choice for simulation, CFD, molecular dynamics and scientific computing.
• CPU-heavy data pipelines: Up to 256 Intel cores and MRDIMM-8800 bandwidth for ETL, tokenisation, vector search and pre-processing beside the GPUs.
• Sovereign / on-prem AI: Keep sensitive data in your own EU rack instead of a public cloud, with predictable cost per GPU-hour.

NVIDIA HGX B200 8-GPU performance

System-level figures published by NVIDIA for the eight-GPU board; real application performance depends on model, precision and software.

SYS-A22GA-NBRT full specifications

Configuration options

Every HGX server is configured to order. Typical options are listed below; final compatibility and part numbers are confirmed in your quote.

Delivery and lead time

• Estimated lead time: Confirmed in your quote; custom HGX builds can take up to ~30 weeks depending on GPU allocation.
• Quote: written B2B quote excluding VAT within 1 business day, including the confirmed delivery window.
• Shipping: palletised freight to your data centre in all 27 EU countries; delivery to other countries on request and subject to export-control checks.
• Before shipping: optional burn-in test with report, firmware update and OS pre-installation.

Data-centre checklist

• Power: 6× 5250 W Titanium PSUs in 3+3 redundancy on 200–240 VAC – plan two independent feeds/PDUs. The exact power budget is part of the quote.
• Space: 843 mm deep 10U chassis – a 1000–1200 mm deep rack.
• Weight: 133 kg net – use a lift for installation and check rack and floor load ratings.
• Cooling: front-to-back airflow, inlet air 10–35 °C; hot/cold-aisle containment recommended.
• Network: plan leaf switches, optics and cables for the GPU fabric (8× 400G per server) plus management and storage networks.
• Service: Supermicro lists this model as onsite-service-required – a manufacturer onsite service plan is included in every quote.

Buying from Mineshop.eu

• New systems sourced through European IT distribution, with manufacturer warranty and onsite service options.
• B2B invoicing excluding VAT, EU reverse charge for VAT-registered businesses, prices in EUR.
• Hardware specialists who help you size GPUs, memory, storage and networking for your models.
• One supplier for the complete stack: servers, GPUs, 400G/800G networking, server memory and NVMe storage.

Ordering process

1. Click Request a quote and tell us the configuration, quantity, delivery country and timeline.
2. A hardware specialist confirms the configuration with you – by email or a short call.
3. You receive a written quote within 1 business day, with lead time, service plan and payment terms.
4. You confirm with a purchase order; we handle export-control and end-user paperwork.
5. Your server is built, tested and delivered; optional rack integration and onsite installation.

Compare Supermicro HGX servers

Frequently asked questions

What is the price of the Supermicro SYS-A22GA-NBRT?

The SYS-A22GA-NBRT is sold on quotation because NVIDIA B200 GPU pricing and allocation change frequently and every system is configured (CPU, memory, storage, networking, service plan). Send a request and you receive a written B2B quote, excluding VAT, within 1 business day.

Is the SYS-A22GA-NBRT the same as an NVIDIA DGX B200?

Both use the NVIDIA HGX B200 8-GPU board with eight B200 GPUs and NVLink Switch. DGX B200 is NVIDIA's own branded, fixed system with NVIDIA's software and support bundle; the Supermicro SYS-A22GA-NBRT lets you choose the CPUs, memory, storage, network adapters and service level, and is often available on different lead times.

Should I choose HGX B200 or HGX B300?

B300 (Blackwell Ultra) has 288 GB per GPU and higher dense FP4 throughput for inference and reasoning models. B200 has 180 GB per GPU but far stronger FP64 (296 vs 10 TFLOPS), so it is the better fit for mixed AI + HPC and simulation work, and often the more budget-friendly Blackwell option.

What does the server need in my data centre?

A 200–240 V supply for six 5250 W Titanium PSUs (3+3 redundant), 10U of rack space in a deep rack, a floor and rack rated for 133 kg, and inlet air between 10 and 35 °C. We confirm the exact power budget, cords and airflow requirements in your quote.

Why is an onsite service plan required?

Supermicro lists this model as 'onsite service required', so a manufacturer onsite service plan is part of every quote. It covers on-site diagnosis and part replacement for a system of this weight and value.

How long is delivery?

Lead time is confirmed in your quote. Custom HGX builds can take up to ~30 weeks depending on GPU allocation; ask us about faster pre-configured alternatives such as the Gold Series HGX B300 systems.

NVIDIA, HGX, NVLink, ConnectX and BlueField are trademarks of NVIDIA Corporation. Supermicro is a trademark of Super Micro Computer, Inc. Specifications from the manufacturers; subject to change without notice.