UK’s trusted IT infrastructure partner since 2003
sales@servnetuk.com
0800 987 4111
Servnet
ConfiguratorGet in Touch
← Comparison Tool

Dell XE9680 vs XE9680L vs XE9640

AI-powered analysis across 19 matched specifications

Dell PowerEdge XE9680 8-way GPU AI server, 6U rack
PowerEdge XE9680
Dell
9.2
Overall Score
Best for: Flexible, multi-vendor AI infrastructure in air-cooled data centres requiring maximum GPU options and expansion capabilities.
View Full Details
Dell PowerEdge XE9680L liquid-cooled AI server, 4U rack
PowerEdge XE9680L
Dell
9.1
Overall Score
Best for: Maximum-performance NVIDIA Blackwell deployments in liquid-cooled environments where rack density and energy efficiency are critical.
View Full Details
Dell PowerEdge XE9640 4-GPU dense AI server, 2U rack
PowerEdge XE9640
Dell
8.3
Overall Score
Best for: High-density inference and distributed training workloads where rack space optimisation and liquid cooling efficiency are paramount.
View Full Details

Performance Overview

Scores based on quantifiable specification values (1-10 scale)

ComputeMemoryStorageNetworkingExpandabilityManagement
PowerEdge XE9680
PowerEdge XE9680L
PowerEdge XE9640
Compute
PowerEdge XE9680
9.5
PowerEdge XE9680L
9.5
PowerEdge XE9640
9.0
Memory
PowerEdge XE9680
10.0
PowerEdge XE9680L
10.0
PowerEdge XE9640
10.0
Storage
PowerEdge XE9680
9.0
PowerEdge XE9680L
8.0
PowerEdge XE9640
7.0
Networking
PowerEdge XE9680
8.0
PowerEdge XE9680L
7.0
PowerEdge XE9640
7.0
Expandability
PowerEdge XE9680
9.5
PowerEdge XE9680L
10.0
PowerEdge XE9640
7.5
Management
PowerEdge XE9680
9.0
PowerEdge XE9680L
9.0
PowerEdge XE9640
8.5

Detailed Specifications

Specification
PowerEdge XE9680
Dell
PowerEdge XE9680L
Dell
PowerEdge XE9640
Dell
Key Metrics
Form Factor6U Rack Server4U Rack Server (Liquid Cooled)2U Rack Server (Liquid Cooled)
GPU Accelerators per Node8× (H200/H100/H20/MI300X/Gaudi3)8× (B200/H200)4× (H100/H200)
Total GPU Memory (Max Config)1.12 TB (8× H200)1.44 TB (8× B200)564 GB (4× H200)
GPU Interconnect Bandwidth900 GB/s (NVLink full-mesh) / Infinity Fabric (AMD)NVLink full-mesh (B200/H200)900 GB/s (NVLink full-mesh)
Rack Density AdvantageAir-cooled — no liquid required33% higher GPU density per rack vs 6U air-cooled2× GPU core density vs air-cooled 6U
Cooling Energy Saving--45% lower energy cost vs air45% energy saving over air cooling
Compute
ProcessorDual Intel Xeon Scalable 5th Gen (up to 64 cores) or 4th Gen (up to 56 cores)Dual Intel Xeon Scalable 5th Gen (up to 64 cores) @ 5600 MT/sDual Intel Xeon Scalable 5th Gen (up to 64 cores) or 4th Gen (up to 56 cores)
Memory
System Memory32 DDR5 DIMMs, up to 4 TB @ 5600 MT/s (5th Gen)32 DDR5 DIMMs, up to 4 TB @ 5600 MT/s (1 DPC)32 DDR5 DIMMs, up to 4 TB @ 5600 MT/s
Storage
Storage Options8× 2.5" NVMe/SAS/SATA or 16× E3.S NVMe — up to 122.88 TB8× 2.5" NVMe SSDs, up to 122.88 TBUp to 8 NVMe drives
Networking
Networking1× OCP 3.0 + 2× 1GbE LOM; PCIe InfiniBand/Ethernet via slots----
GPU / Accelerators
GPU Options8× NVIDIA HGX H200 (141 GB) / H100 (80 GB) / H20; 8× AMD MI300X (192 GB); 8× Intel Gaudi3 (128 GB)8× NVIDIA HGX B200 (180 GB, 1000W) or H200 (141 GB, 700W)4× NVIDIA H100 SXM5 (80 GB) or H200 SXM5 (141 GB) — full NVLink
GPU Memory1.12TB GPU Memory (8× H200 NVLink pool)1.44 TB HBM3e (8× B200) / 1.13 TB HBM3 (8× H200)320 GB (4× H100) or 564 GB (4× H200)
Expansion / PCIe
PCIe Expansion10× PCIe Gen5 x16 (8 with Intel Gaudi3)12× PCIe Gen5 x164× PCIe Gen5 x16
Management
ManagementiDRAC9 Enterprise / Datacenter, Redfish API, CloudIQ, OpenManage EnterpriseiDRAC9 Datacenter, Redfish API, CloudIQ, OpenManage EnterpriseiDRAC9 Enterprise / Datacenter, Redfish API, OpenManage Enterprise
Power
Power Supplies6× 3200W Titanium (277V) / 2800W Titanium (200–240V), 3+3 FTR3000W Titanium (B200) / 2800W Titanium (H200)--
Physical / Environmental
CoolingAir-cooled, 16 fans, 1200 CFM, 10–35°C ambientDirect liquid cooling — CPUs, GPUs, NVLink switchesDirect liquid cooling — CPUs and GPUs
Dimensions10.36" H × 18.97" W × 39.71" D (with bezel)----
WeightUp to 251 lbs / 114 kg (configuration dependent)----
Security
SecuritySilicon Root of Trust, Secure Boot, Secure Component Verification, SEDs----

Expert Analysis

AI-generated based on published specifications

These three Dell PowerEdge servers represent a sophisticated tiered approach to GPU-accelerated computing, each optimised for different deployment scenarios within the data centre. The XE9680 offers exceptional flexibility with its 6U air-cooled design supporting eight GPUs across three vendor architectures (NVIDIA, AMD, Intel), making it ideal for organisations requiring multi-vendor AI/ML strategy or those operating in facilities without liquid cooling infrastructure. Its 10 PCIe Gen5 slots provide outstanding expansion for high-speed networking adapters. The XE9680L represents the pinnacle of density and efficiency with its 4U liquid-cooled chassis supporting eight of NVIDIA's latest Blackwell B200 GPUs, delivering 1.44TB of HBM3e memory and 12 PCIe slots while reducing cooling energy costs by 45% compared to air-cooled alternatives.

The XE9640 occupies a distinct position as a 2U liquid-cooled system with four GPUs, offering the highest rack density for inference workloads and distributed training where space efficiency is paramount. While it provides half the GPU count of its larger siblings, it delivers twice the GPU core density per rack compared to air-cooled 6U alternatives. The trade-offs are clear: the XE9680 provides maximum flexibility and vendor choice, the XE9680L delivers peak performance and efficiency for homogeneous NVIDIA deployments, and the XE9640 maximises rack density for cost-sensitive high-density deployments. Organisations must weigh GPU count requirements against rack space constraints, cooling infrastructure availability, and workload-specific performance needs when selecting between these platforms.

PowerEdge XE9680
Best for: Flexible, multi-vendor AI infrastructure in air-cooled data centres requiring maximum GPU options and expansion capabilities.
PowerEdge XE9680L
Best for: Maximum-performance NVIDIA Blackwell deployments in liquid-cooled environments where rack density and energy efficiency are critical.
PowerEdge XE9640
Best for: High-density inference and distributed training workloads where rack space optimisation and liquid cooling efficiency are paramount.

Ready to proceed?

Want to compare different products or add more to this comparison?

Open Interactive Comparison Tool →