New

LLM Inference AI Workstation – Threadripper 9970X, x1/x2 RTX 5090 32GB, 128GB ECC, 4TB Gen5

Name: LLM Inference AI Workstation - Threadripper 9970X, x1/x2 RTX 5090 32GB, 128GB ECC, 4TB Gen5
Brand: PRO SEKRET
SKU: GSK-AIWS-W002
Availability: InStock

SKU: GSK-AIWS-W002

Price range: AED 106,500.00 through AED 152,500.00

Pay as low as AED 10,383.75/month or 4 interest-free payments with Tabby.

Pay as low as AED 10,383.75/month or 4 interest-free payments with Tamara. No late fees.

Graphics Cards Configuration

Clear

Whatsapp for Inquiries

Add to wishlist

7 People watching this product now!

Payments Accepted:

Pick up from Store: Computer Plaza, M28. Bur Dubai

Call +971549966403 to arrange for Pick-up

Free Pick-Up

Courier delivery

UAE: 24-48 Hrs, GCC Countries: 2-5 Days

Free Delivery
(On Orders over Aed. 500)
UAE Only

Free 14-Day returns

More details

AMD Ryzen Threadripper 9970X — 32 cores / 64 threads, up to 5.4GHz boost, 128MB L3 cache, sTR5 platform
ASUS TUF Gaming GeForce RTX 5090 32GB OC — NVIDIA Blackwell, 21,760 CUDA cores, GDDR7, FP4 support for local LLM inference
32GB GPU VRAM — Run large language models entirely on-premise with no API costs and full data privacy
128GB DDR5 5600 ECC Registered Memory — 4 × 32GB Kingston modules for error-corrected stability in long inference sessions
4TB Samsung 9100 PRO NVMe SSD — PCIe 5.0 M.2 storage for fast model loading and massive local model libraries
ASUS Pro WS TRX50-SAGE WIFI A Motherboard — 3 × PCIe 5.0 x16 Safeslots, 4 × M.2 slots, WiFi 7, USB4, 10Gb + 2.5Gb LAN
SilverStone XE360-TR5 Liquid Cooling — 360mm AIO engineered for sTR5 Threadripper heat loads
2000W Cooler Master X Mighty Platinum Power — ATX 3.1, 2 × native 12V-2×6 for dual RTX 5090s, MasterCTRL monitoring, zero-RPM fan mode
Cooler Master HAF 700 Full Tower — High-airflow SSI-EEB chassis with 5 included aRGB fans and up to 480mm radiator support

Your own private AI, running at full speed. The LLM Inference AI Workstation pairs the 32-core AMD Ryzen Threadripper 9970X with the ASUS TUF Gaming RTX 5090 32GB, 128GB of ECC Registered DDR5 memory, and 4TB of PCIe 5.0 storage — purpose-built to run large language models locally, with no subscriptions, no rate limits, and no data leaving your network.

32GB of VRAM for Serious Local Models

The RTX 5090’s 32GB of GDDR7 memory and 21,760 CUDA cores accelerate quantized 30B–70B-class LLMs with room for long context windows — from coding assistants and document analysis to RAG pipelines and multi-agent workflows. NVIDIA Blackwell’s FP4/FP8 support maximizes tokens per second, while the 4TB Samsung 9100 PRO PCIe 5.0 SSD loads multi-gigabyte model weights in seconds and holds an extensive local model library.

A Threadripper Foundation That Never Bottlenecks

Model loading, tokenization, embeddings, and data preprocessing run on the Threadripper 9970X — 32 cores and 64 threads at up to 5.4GHz — while 128GB of Kingston DDR5 5600 ECC Registered memory keeps long sessions error-free and supports CPU-offloaded layers for even larger models. The ASUS Pro WS TRX50-SAGE WIFI A offers three PCIe 5.0 x16 slots and four M.2 slots for additional GPUs or storage, with WiFi 7, USB4, and 10Gb Ethernet for fast model transfers across your network.

Engineered for 24/7 Inference

The SilverStone XE360-TR5 360mm liquid cooler keeps the 350W Threadripper at sustained boost clocks, and the Cooler Master HAF 700 moves serious air with two 200mm and three 120mm aRGB SickleFlow fans included. The Cooler Master X Mighty Platinum 2000W PSU delivers fully modular, 80 PLUS Platinum power with two native 12V-2×6 connectors for dual-GPU configurations, Infineon industrial-grade components, and real-time MasterCTRL monitoring — ready for continuous, around-the-clock AI workloads.

Perfect For:

AI developers running local LLM inference, fine-tuning, and RAG pipelines
Businesses deploying private chatbots and document AI with full data sovereignty
Researchers and power users who want API-class models without recurring API costs

The LLM Inference AI Workstation arrives fully assembled and tested — your private AI server, ready to run from day one.

Product Type	Custom-Built LLM Inference AI Workstation
Brand / Model	IXGAMERS LLM Inference AI Workstation (Single RTX 5090)
Processor	AMD Ryzen Threadripper 9970X, sTR5, 32 cores / 64 threads, 4.0GHz base, up to 5.4GHz boost, 128MB L3 cache, 350W TDP (P/N: 100-100001594WOF)
Motherboard	ASUS Pro WS TRX50-SAGE WIFI A, AMD TRX50, 4 × DDR5 ECC R-DIMM slots, 3 × PCIe 5.0 x16 Safeslots, 4 × M.2 slots, WiFi 7, USB4, 10Gb + 2.5Gb LAN (P/N: 90MB1LE0-M0EAY0)
Memory	128GB (4 × 32GB) Kingston DDR5 5600, CL46, 288-pin, ECC Registered
Storage	Samsung 9100 PRO 4TB PCIe 5.0 NVMe M.2 SSD
Cooling	SilverStone XE360-TR5 360mm All-In-One Liquid CPU Cooler (sTR5)
Graphics Card	ASUS TUF Gaming GeForce RTX 5090 32GB OC Edition, GDDR7, 512-bit, PCIe 5.0, 21,760 CUDA cores, Blackwell architecture, FP4/FP8 support, 32GB VRAM
Power Supply	Cooler Master X Mighty Platinum 2000W, 80 PLUS Platinum, ATX 3.1 / PCIe 5.1, fully modular, 2 × native 90° 12V-2×6 (600W) cables, Infineon chipset, MasterCTRL digital monitoring, zero-RPM fan mode, 10-year manufacturer warranty (P/N: MPS-K001-AFBP-NBUK; 200–240V input, C19 power cord)
Chassis	Cooler Master HAF 700 full tower (P/N: H700-IGNN-S00), tempered glass side panel, 2 × 200mm + 3 × 120mm aRGB SickleFlow fans included, up to 480mm radiator & 17 fan support (66.6 x 29.1 x 62.6 cm, L x W x H)
Intended Use	Local LLM inference, RAG pipelines, private AI assistants, model fine-tuning
Weight	30.4 kg (assembled unit)
Dimensions	78 x 40 x 73 cm (packaged)
Warranty	Backed by a 1 Year Service Warranty + Lifetime Phone Support. Component brands offer 3-10 Year extended warranty (subject to registration on the particular brands website)
Care Instructions	Keep in a well-ventilated area away from direct sunlight; clean dust filters regularly and clear vents with compressed air. Power off before opening the chassis.