Request Demo →

AI SYSTEMS / TECHNICAL OVERVIEW

AI Servers

One central computer that several people or applications can use.

Choose this when: Your team needs a shared AI service, with access and resources managed centrally.

For example: An internal document assistant used by colleagues across your company.

Configured on request · subject to supplier confirmation

Configure your request →

Architecture illustration · no product specification implied

AI SYSTEMS / PROPOSED CONFIGURATIONS

Choose a system
to quote.

Start with one of these systems, then tell us about your project. We will review the configuration with you. Need help choosing a system family? →

Proposed configurations · Quote on request. Final specifications, availability and delivery time are confirmed in your quotation.

Rack server

S1 Inference Server

A starting point for teams planning a shared AI service in a rack-based setup.

Processor
AMD EPYC 4565P · 16 cores
Memory
128 GB DDR5 ECC
Graphics
1 × NVIDIA RTX PRO 4500 Blackwell · 32 GB
Storage
2 × 1.92 TB enterprise NVMe · mirrored, with power-loss protection
Case & power
4U rack chassis with rails · redundant 1+1 power supply
Management
Remote management

A second GPU can be assessed as an option; it is not included in this proposed configuration.

Request a quote for S1 Inference Server →

TECHNICAL SCOPE

A specification built around your use case.

This overview explains the system family. Where shown, named configurations are proposed starting points; exact parts and compatibility are confirmed in your quotation. No benchmark, certification, price or stock availability is asserted.

Architecture
Rack or tower deployment; form factor, management controller, redundant power and accelerator topology require OEM confirmation.
Typical use cases
Private inference · shared knowledge · batch processing
CPU / accelerator
Architecture, CPU model, GPU model and VRAM are selected against the workload and confirmed OEM bill of materials.
Memory & storage
Requested capacities are configurable. Exact part numbers, interfaces, memory population and compatibility require technical validation.
Networking & support
Interface requirements and support coverage form part of the enquiry. No support SLA, lead time or warranty duration is confirmed by this page.
Performance assessment
Concurrent-user capacity depends on model size, context length, batching and service objectives. No throughput guarantee is implied.
Qualification checks
Rack depth, circuit capacity, airflow, ECC support, remote management, network and service isolation.
Software readiness
GPIRL AI OS is an internal working name for existing GPIRL AI development. Linux x86-64 and other architectures require image, driver, licence, update and recovery tests on the selected hardware. No SKU is currently declared GPIRL AI-ready.
Availability
Configured on request, subject to supplier confirmation. No inventory is held or implied.
Markets
EU-first assessment. US, Canada, Australia and Japan remain inactive until legal, tax, shipping, product and support requirements are satisfied.

GPIRL is the proposed direct seller and retains applicable seller obligations. Manufacturer branding is preserved unless a separately assessed private-label arrangement is agreed. A written quotation defines the exact product, price, delivery, warranty, support and applicable consumer terms.

No online checkout, payment acceptance or procurement submission.