2U GPU-Optimised AI Server — 4 × 400W PCIe Gen5 GPUs, Air-Cooled, DDR5 8 TB
The Dell PowerEdge R760xa is a purpose-built 2U AI/ML server designed to accommodate the widest range of GPU configurations in an air-cooled 2U chassis. Supporting up to 4 × 400W double-wide PCIe Gen5 GPU accelerators (NVIDIA, AMD, Intel), or up to 12 single-wide accelerators, with 4th/5th Gen Intel Xeon Scalable processors and DDR5 up to 8 TB, it accelerates AI training, inference, digital twins, and high-performance rendering.
Get Pricing — Speak to Our Team
Competitive pricing · Response within 4 hours
Front-facing accelerator design accommodates up to 4 × 400W double-wide PCIe Gen5 GPU cards (NVIDIA, AMD, Intel) or up to 12 single-wide (PCIe x8, 75W) accelerators. Front-facing placement improves airflow directly over GPU heatsinks — enabling higher TDP cards without liquid cooling.
Dual-socket design supports 4th Gen Intel Xeon Scalable (up to 56 cores, 350W) or 5th Gen Intel Xeon Scalable (up to 64 cores, 5600 MT/s DDR5). Intel QuickAssist Technology optional for encryption/compression offload alongside GPU workloads.
32 DDR5 RDIMM slots support up to 8 TB registered ECC memory at 5600 MT/s with 5th Gen Xeon. Provides the host memory bandwidth needed to feed multiple high-performance GPU accelerators simultaneously.
Up to 6 × EDSFF E3.S Gen5 NVMe (max 46.08 TB), 6 × 2.5" NVMe (max 92.16 TB), or 8 × 2.5" SAS/SATA/NVMe (max 122.88 TB) — sized to match GPU workload data throughput rather than archival capacity.
Unlike most 4-GPU 2U servers requiring liquid cooling, the R760xa's front-facing accelerator design maintains air-cooled operation for all GPU configurations — simplifying data centre infrastructure requirements significantly.
Full Dell silicon root of trust, SCV hardware integrity check, MFA, TPM 2.0, Secure Boot, and cryptographically signed firmware extend the PowerEdge security platform to GPU-accelerated AI workloads.
The Dell PowerEdge R760xa was designed to address a specific gap in the AI server market: enterprise AI workloads that require more than 2 GPUs but do not require — or cannot accommodate — a dedicated liquid-cooled DGX-class system. The R760xa achieves this by using a front-facing accelerator chassis design where GPU cards face the front of the server and draw cool air directly from the inlet, allowing 400W TDP GPUs to operate at full performance in standard air-cooled rack environments.
GPU configuration flexibility is the R760xa's defining characteristic. The same chassis can be configured with up to 4 × 400W double-wide PCIe Gen5 GPU cards for pure AI training workloads, or up to 12 × 75W single-wide PCIe x8 GPU cards for maximum GPU parallelism in inference tasks. This configurability allows a single server SKU to serve both training-optimised and inference-optimised deployment patterns.
The compute and memory subsystem is equivalent to the R760: two 4th or 5th Generation Intel Xeon Scalable processors with 32 DDR5 RDIMM slots and up to 8 TB of registered ECC memory. However, storage bays are deliberately limited (6–8 drives maximum) to leave maximum PCIe bandwidth available for GPU accelerators — reflecting the GPU-primary workload focus of this platform.
Power infrastructure scales to match GPU configurations: 2400W Platinum and 2800W/3200W Titanium hot-swap power supply options ensure the R760xa can deliver full rated power to 4 × 400W GPUs plus two 350W CPUs and 8 drive bays simultaneously, without the power derating that affects competitors in equivalent configurations.
4 × PCIe Gen5 GPU cards at 400W TDP each provide serious AI training throughput. Inference deployments can choose 12 × single-wide GPUs for maximum parallelism at lower power.
GPU-accelerated simulation, NVIDIA Omniverse digital twin rendering, and 3D visual computing workloads benefit from front-facing GPU cooling at full sustained TDP.
NVIDIA RTX virtual GPU partition across single-wide GPU cards delivers high-quality virtual desktop graphics for engineering, media production, and clinical imaging workloads.
Dual Intel Xeon Scalable processors with 4 PCIe Gen5 compute accelerators provide a cost-effective on-premises HPC cluster node for scientific simulation and computation.
Trusted UK IT reseller since 2003 — supplying businesses of all sizes with genuine, competitively priced technology.
Servnet is an authorised IT reseller with direct access to leading product lines at competitive pricing.
Our engineers will help you select the right product and advise on compatibility with your existing infrastructure.
Most products ship from UK stock. Express delivery available for urgent deployments.
All products are genuine, brand-new, with full manufacturer warranty and RMA support.
We can pre-configure your hardware before shipping — reducing on-site deployment time and cost.
Post-sale support from our certified engineers. We're here if anything goes wrong.
