The Gen12 lineup spans 1U and 2U rack servers, a tower, and accelerated GPU and NVIDIA CPU platforms — all on one HPE iLO 7 security, efficiency, and management foundation. Narrow it down by form factor or by workload below, compare the models side by side, and we will help you land on the exact build.
Rack height and socket count are the fastest way to narrow the field. Pick the shape that fits your rack, your power envelope, and how much you need to expand.
Compact, rack-optimized 2P server that maximizes density while delivering the compute, memory, and hybrid-cloud performance of a mainstream workhorse.
Explore DL360Single-socket 1U rackmount that maximizes rack utilization while mitigating virtualization risk in power-constrained and edge environments.
Explore DL320The flexible 2P workhorse: up to 8 TB memory, up to 36 EDSFF E3.S NVMe drives, and GPU headroom for data analytics and hybrid cloud.
Explore DL380Single-socket 2U with reduced power draw, better thermals, and more PCIe lanes — tuned for data- and storage-centric IaaS, PaaS, and SaaS.
Explore DL340Robust 2P tower with an optional rackable chassis — datacenter-class compute, expandability, and reliability for offices and edge sites without a data center.
Explore ML3504U 2P accelerator-optimized platform supporting up to eight double-wide (or sixteen single-wide) GPUs for AI training and inference and other GPU-intensive workloads.
Explore DL380aAlready know what the server has to do? Jump straight to the model built for that job. Not sure? Use these as a starting point and we will confirm the fit.
Consolidated VMs, business apps, and mixed workloads that need a dependable, balanced foundation.
Best fit: DL360 Gen12 (1U) or DL380 Gen12 (2U)
Memory- and storage-hungry databases, analytics, and hybrid-cloud platforms that scale.
Best fit: DL380 Gen12 — up to 8 TB memory and 36 EDSFF NVMe
Model training and inference, and other workloads that live and die on GPU and memory bandwidth.
Best fit: DL380a Gen12, DL384 Gen12, or DL394 Gen12
Remote and space- or power-limited locations that still need enterprise security and management.
Best fit: DL320 Gen12 (1U) or DL340 Gen12 (2U); for NEBS Level 3 telco/cell-site environments, the short-depth DL110 Gen12
Workloads that need more drive bays and PCIe lanes per socket without a second CPU.
Best fit: DL340 Gen12 — 2U single-socket with expanded PCIe
On-premises compute where there is no rack — but the same reliability is expected.
Best fit: ML350 Gen12 tower (optional rackable chassis)
Every model below shares HPE iLO 7 with a dedicated secure-enclave processor, Compute Ops Management, and Gen12 efficiency. Use this to shortlist, then talk to us to spec the exact configuration.
| Model | Form factor | Sockets | Processor | Best for |
|---|---|---|---|---|
| DL110 Gen12 | 1U rack (short depth) | 1P | Intel Xeon 6 SoC | NEBS Level 3 telco/edge — vRAN & 5G edge sites |
| DL320 Gen12 | 1U rack | 1P | Intel Xeon 6 | Dense, power-efficient rack & edge compute |
| DL340 Gen12 | 2U rack | 1P | Intel Xeon 6 | Data- & storage-centric workloads needing more PCIe |
| DL360 Gen12 | 1U rack | 2P | Intel Xeon 6 | Dense general compute, virtualization, hybrid cloud |
| DL380 Gen12 | 2U rack | 2P | Intel Xeon 6 | Data analytics, hybrid cloud, expandable enterprise apps |
| DL380a Gen12 | 4U rack | 2P | Intel Xeon 6 | Accelerated AI — up to 8 double-wide (16 single-wide) GPUs |
| DL384 Gen12 | 2U rack | — | NVIDIA Grace CPU + Hopper GPU | Large-model AI inference & HPC |
| DL394 Gen12 | 2U rack | — | NVIDIA Vera CPU | Agentic AI & reinforcement learning |
| ML350 Gen12 | Tower (rackable) | 2P | Intel Xeon 6 | Offices & edge sites without a data center |
Work through these in order. Once you can answer them, there is usually one clear model — and we can confirm it and price it in a single conversation.
Talk to an ExpertA rack points you to a DL model; an office or edge closet points to the ML350 tower. Tight on rack units or power? Favor 1U (DL320/DL360).
One socket (DL320, DL340) is efficient and cost-effective; two sockets (DL360, DL380, ML350) add cores, memory, and expansion for heavier consolidation.
Need lots of NVMe and PCIe? The 2U DL380 (up to 36 EDSFF) and DL340 lead here. Compact nodes trade capacity for density.
Light GPU use fits the DL380. Heavy AI points to the accelerated DL380a, the Grace-Hopper DL384, or the Vera-based DL394.
Tell us what the server has to run, where it lives, and how you want to buy it. We will recommend the right Gen12 build, get you authorized-partner pricing, and stay on the order until it is racked.
Start with form factor and socket count, then match the workload. Choose a 1U DL360 (2P) or DL320 (1P) for dense rack compute, a 2U DL380 (2P) or DL340 (1P) for balanced or storage-centric workloads, an ML350 tower for offices and edge sites, and an accelerated DL380a, DL384, or Vera-based DL394 for AI and GPU workloads. Every Gen12 model shares the same HPE iLO 7 security, efficiency, and Compute Ops Management foundation, so the decision is about form factor and workload rather than protection.
For accelerated AI, the 4U DL380a Gen12 supports up to eight double-wide (or sixteen single-wide) GPUs, the DL384 Gen12 pairs NVIDIA Grace CPUs with Hopper GPUs, and the DL394 Gen12 uses the NVIDIA Vera CPU for agentic AI. For GPU-assisted general workloads, the 2U DL380 Gen12 supports up to three double-wide or eight single-wide GPUs.
The 1U single-socket DL320 Gen12 maximizes rack utilization in power-constrained environments, and the 2U single-socket DL340 Gen12 adds PCIe lanes and storage for data-centric edge workloads. Where there is no data center at all, the ML350 Gen12 tower delivers datacenter-class capability in an office-friendly chassis.
Both are two-socket Intel Xeon 6 servers. The DL360 Gen12 is a compact 1U system that maximizes rack density, while the DL380 Gen12 is a 2U system with more storage bays (up to 36 EDSFF E3.S NVMe), more memory (up to 8 TB), and more GPU and PCIe expansion for data analytics and hybrid cloud.
The first wave of HPE ProLiant Gen12 rack and tower servers (DL320, DL340, DL360, DL380, DL380a, and ML350) is powered by Intel Xeon 6 processors. HPE also offers AMD EPYC-based Gen12 models. Tell us your workload and we will map you to the right processor family and configuration.
Yes. The full Gen12 portfolio is available as a standalone purchase or as a service via HPE GreenLake, so you can align server capacity with how your workloads and budget grow.
Tell us about your workloads and a ServerComputeWorks specialist will help you choose, configure, price, and order the right HPE ProLiant Gen12 system.
Request a Quote*Intel and Intel Xeon are trademarks of Intel Corporation in the U.S. and other countries. NVIDIA, Vera, Grace, and Hopper are trademarks and/or registered trademarks of NVIDIA Corporation. All other third-party trademark(s) is/are property of their respective owner(s). Specifications and availability are subject to change.