Axe series 4U GPU servers
The AXE series 4U AI GPU servers bring data centre-grade computing to edge and industrial environments. The range includes platforms based on Intel Xeon W, Intel Xeon Scalable, and AMD EPYC processors, allowing organisations to match CPU architecture to workload requirements. With both full-depth and short-depth 4U chassis options, the series suits standard racks as well as space-constrained installations such as factory cabinets and mobile deployments. Depending on the model, the servers support up to four double-width GPU or accelerator cards for AI training, inference, vision, and analytics workloads. Rich PCIe Gen5 expansion enables integration of GPUs, high-speed networking, frame grabbers, and other specialist I/O.
Local storage options include combinations of SATA and NVMe bays to keep datasets close to the point of processing. Robust cooling and server-grade power designs support continuous operation under demanding thermal and compute loads. By processing AI workloads on-site, the AXE series helps reduce latency, improve data privacy, and support reliable real-time decision-making.
The AXE series is built for engineers, system integrators, and operators who need powerful GPU acceleration outside traditional data centres. These 4U platforms combine enterprise-class processing, flexible expansion, and edge-friendly mechanical designs to support AI deployment where data is created.
Use cases include large language model inference, industrial machine vision, medical image analysis, smart manufacturing, autonomous systems, and other latency-sensitive workloads that benefit from keeping compute and data on-site.
Range features
A high level overview of what this range offers
- Flexible processor platforms: Available with Intel Xeon W-3400, Intel Xeon Scalable 4th/5th Gen, or AMD EPYC 9004/9005 processors to align performance, core count, and platform features with the intended AI workload.
- Multi-GPU acceleration: Supports up to four full-length, double-width GPU or FPGA cards on selected models, with the short-depth AXE-7420GWA supporting up to two GPUs for compact deployments.
- Extensive PCIe Gen5 expansion: Up to 11 or 12 PCIe slots depending on model, enabling installation of accelerator cards, high-speed NICs, frame grabbers, and other edge I/O devices.
- Short-depth deployment options: 550 mm and 650 mm chassis variants simplify installation in industrial cabinets, telecom racks, and other constrained environments.
- High-capacity on-site storage: Configurations include up to eight SATA bays or mixed NVMe and SATA storage to keep datasets local and accessible with low latency.
- Industrial-grade thermal design: Multiple hot-swap or modular fan assemblies help maintain stable operation with high-power CPUs and GPUs under continuous load.
- Reliable power architecture: Select systems support 1+1 redundant power supplies for improved uptime, while compact models use robust industrial ATX power solutions.
- Optimised for edge AI: Designed to run AI inference, training, analytics, and vision workloads on-premises to improve responsiveness, privacy, and operational resilience.
What’s in this range?
All the variants in the range and a comparison of what they offer
| Specification | AXE-7440GW | AXE-7420GWA | AXE-7440TR | AXE-7400SR |
|---|---|---|---|---|
Processor support | 1× Intel Xeon W-3400 series, up to 350 W TDP | 1× Intel Xeon W-3400 series, up to 350 W TDP | 1× AMD EPYC 9004/9005 series | 1× Intel Xeon Scalable 4th/5th Gen |
Chassis depth | Full-depth 4U, approximately 800 mm | Short-depth 4U, 550 mm | Full-depth 4U, approximately 800 mm | Short-depth 4U, 650 mm |
Max GPU cards | Up to 4× double-width GPUs | Up to 2× double-width GPUs | Up to 4× double-width GPUs | Up to 4× double-width GPUs |
Expansion slots | 11 PCIe slots total, including 4× Gen5 x16 plus x8/x4 via MCIO | 12 PCIe slots total, including 3× Gen5 x16 plus multiple x8/x4 via MCIO | 7 PCIe slots total, max 4× Gen5 x16 and 3× Gen5 x8 via MCIO | 6 PCIe slots total, max 4× Gen5 x16 and 2× Gen5 x8 via MCIO |
Storage bays | 8× 2.5"/3.5" SATA drive bays | 8× 2.5"/3.5" SATA drive bays + 2× SlimSAS SSD bays + 1× slim ODD | 6× 2.5" hot-swap NVMe SSD + 2× 2.5" hot-swap SATA SSD bays | 6× 2.5"/3.5" hot-swap SATA drive bays |
Cooling | 3× high-flow internal hot-swap fans + optional 2× rear fans | 3× internal hot-swap fans + optional 2× rear fans | 3× internal fan modules + optional 2× rear fan modules | 3× internal fan modules + optional 2× rear fan modules |
Power supply | Single or 1+1 redundant ATX PSUs, up to 1600 W total | Single industrial ATX PSU | Dual redundant server PSUs | 1+1 redundant 2000 W CRPS power supplies |
Notable features | Full-length GPU board support and extra PCIe expansion capability | 550 mm short-depth design for compact industrial racks | AMD platform for high core-count AI workloads | NVIDIA-Certified design with BMC, IPMI 2.0, and Redfish |
FAQs
for Axe series 4U GPU servers
The two Intel Xeon W-based models share a similar platform approach, but they target different deployment constraints. The AXE-7440GW uses a full-depth 4U chassis and supports up to four double-width GPUs for maximum accelerator density. The AXE-7420GWA uses a 550 mm short-depth 4U chassis and supports up to two double-width GPUs, making it better suited to cabinets and racks with limited depth.
The series includes multiple CPU platform options to match different performance profiles. AXE-7440GW and AXE-7420GWA support Intel Xeon W-3400 series processors. AXE-7400SR supports a single Intel Xeon Scalable 4th or 5th Gen processor. AXE-7440TR is based on a single AMD EPYC 9004/9005 series processor. This gives buyers a choice between workstation-class and server-class platforms for AI and edge computing.
GPU capacity depends on the model. The AXE-7440GW, AXE-7440TR, and AXE-7400SR support up to four full-length, double-width GPU or accelerator cards. The AXE-7420GWA supports up to two double-width GPUs due to its short-depth chassis. This allows the range to cover both compact inference deployments and high-density accelerator configurations.
Storage configurations vary by model. The AXE-7440GW and AXE-7420GWA offer eight SATA drive bays for local storage. The AXE-7440TR combines six hot-swap NVMe SSD bays with two SATA SSD bays for high-throughput workloads. The AXE-7400SR provides six hot-swap SATA bays and is positioned for reliable edge deployments that need local data retention and RAID-oriented storage workflows.
Yes. The AXE-7400SR is described with integrated server management through a BMC supporting IPMI 2.0 and Redfish, enabling remote monitoring and control. This is especially useful for unmanned or distributed edge sites. The other models are more focused on compute and expansion, and their management capabilities are less explicitly defined in the provided product information.
MCIO refers to a high-density connector approach used to extend PCIe connectivity from the motherboard to additional slot or expansion board implementations. In practical terms, it helps these single-socket servers expose more PCIe slot options for GPUs, networking, and specialist cards without overcrowding the main board layout.
Yes. The AXE series is designed specifically for edge AI deployment in locations such as factory backends, medical facilities, telecom sites, and mobile or remote installations. Features such as robust 4U rackmount construction, strong active cooling, short-depth chassis options, and redundant power support on selected models make the range suitable for demanding on-site computing scenarios.
The short-depth AXE-7420GWA and AXE-7400SR are intended for installations where standard full-depth servers are difficult to deploy. Their reduced chassis depth helps them fit into compact cabinets, industrial racks, telecom enclosures, and other constrained spaces while still delivering strong CPU performance, GPU support, and expansion for edge AI workloads.
Yes, selected models do. The AXE-7440GW supports single or 1+1 redundant ATX power supply configurations, and the AXE-7400SR is specified with 1+1 redundant 2000 W CRPS power supplies. The AXE-7440TR is also positioned with redundant server PSU support, while the AXE-7420GWA uses a single industrial ATX PSU to suit its compact short-depth design.
Typical use cases include industrial machine vision, medical imaging analytics, smart manufacturing, AI inference, large language model deployment, security and video analytics, and telecom edge computing. The series is intended for workloads that need high-performance CPUs and GPUs close to the data source to reduce latency, improve privacy, and maintain operational independence from the cloud.
