Images are for illustration
The photos show an example build. The case, fans, RGB lighting and cable styling you receive may look different, depending on stock.
The core specification listed on this page (CPU, graphics card, RAM and storage) is exactly what you'll get.
Cortex AI Ultra X – Workstation PC
£8,499.90 Inc.VAT
- AMD Ryzen 9 9950X
- 96GB DDR5 RAM
- 6TB SSD Storage
- NVIDIA GeForce RTX 5090
- Windows 10 / 11 Ready
Designed for AI Professionals
ML Engineers
Train, evaluate and ship models locally without waiting on cluster time.
Data Scientists
Wide dataframes and GPU-accelerated modelling that stay in memory.
AI Researchers
Multi-day training runs with no per-hour meter running against you.
Indie Developers
A capable local assistant to build against, with no API bill per request.
AI Artists
High-resolution diffusion and long ComfyUI graphs rendered in batches.
Startups & Labs
Own the hardware, and keep proprietary training data on your premises.
Engineered for Machine Learning Workloads
AMD Ryzen 9 9950X3D
16 Zen 5 cores with 3D V-Cache and 144MB of total cache keep data loading, tokenisation and preprocessing from starving a 5090 during training.
NVIDIA RTX 5090 32GB
32GB of GDDR7 VRAM on a 512-bit bus holds quantised 30B-class models and single-GPU fine-tuning runs, with 5th-gen Tensor cores and FP4 support.
96GB DDR5 6000MHz
High-capacity memory stages tokenised datasets and optimiser state alongside VRAM, so system RAM never becomes the bottleneck.
2TB + 4TB NVMe SSD
A Samsung 9100 Pro boot drive for models and environments, plus a 4TB second NVMe for datasets and checkpoints, so neither competes for bandwidth.
Arctic Freezer III 420
A 420mm all-in-one liquid cooler that holds clocks steady through multi-day training runs, and stays quiet enough to sit beside.
Specifications
Processor (CPU)
16 Cores / 32 Threads | 144MB Cache
Up to 5.7 GHz Boost | AM5
Processor Cooling
420mm All-in-One Liquid Cooler
Triple 140mm P14 PWM Fans
Thermal Paste
Applied for Best Performance
Optimised Sustained-Load Heat Transfer
Motherboard
AMD X870E Chipset | PCIe 5.0 ATX Architecture
Wi-Fi 7 & High-Speed LAN
Memory (RAM)
Ultra-Fast 6000MHz Speed
High-Capacity Workstation Memory Pool
Graphics Card
Blackwell Architecture | 32GB GDDR7 VRAM
5th-Gen Tensor Cores for AI Acceleration
1st M.2 SSD Drive
PCIe 5.0 Solid State Drive
Boot Drive for OS, Models & Environments
2nd M.2 SSD Drive
Internal Solid State Drive
Dataset & Checkpoint Storage
Operating System
Home / Pro Edition
United Kingdom - English Language
Power Supply
80 Plus Platinum | Fully Modular
ATX 3.1 & PCIe 5.0 Ready
Case
Full-Tower ATX Chassis
High-Airflow Mesh Front Panel
Warranty
1 Year Collect & Return
1 Year Parts, 3 Year Labour
Built for Your Workflow
Local LLM Inference
Fine-Tuning & LoRA
Diffusion & Image Generation
RAG & Agent Pipelines
Data Science & Preprocessing
Optimised for Industry-Leading AI Frameworks
Why Choose Cortex AI?
Purpose-built for local inference, fine-tuning and diffusion work rather than adapted from a gaming range. Every Cortex AI workstation is professionally assembled, extensively tested and shipped with the CUDA data-science stack already configured.
Professionally Built
Hand-assembled by experienced UK technicians.
Stress Tested
Every build passes extended full-load testing.
Day-One Ready
CUDA, cuDNN and PyTorch installed and validated.
Fully Customisable
Tailor any spec to your exact model size.
Every Cortex AI Workstation Includes
Component Selection
Workstation-grade parts onlyExpert Assembly
Built by UK techniciansClean Cabling
Optimised for airflowStress Testing
Extended full-load runsCUDA & Driver Setup
PyTorch validated on a test jobFinal QC & Dispatch
Checked, packed, shippedWARRANTY
Tested for Today. Ready for Tomorrow.
Every Cortex AI workstation is backed by our comprehensive 3-year total warranty. We don't just assemble components—we hand-select and individually test every hardware part to ensure stability through multi-day training runs, while guaranteeing your system stays easily upgradable in the future as your models grow.
Before any workstation leaves our workshop it undergoes exhaustive full-load stress testing, and the CUDA stack is validated against a live PyTorch job. Everything is proven rock-solid, thermally balanced and quiet before being safely boxed and shipped to your door.
— Covered by The House of Computers UK Technical SupportFrequently Asked Questions
Which Cortex AI model should I choose?
Choose the Cortex AI for local inference and everyday assistant use, the AI Pro for RAG pipelines, diffusion work and light fine-tuning, the AI Ultra X — this build — for single-GPU 30B-class fine-tuning, and the AI Ultimate when you need multi-GPU training or 70B+ model serving. Not sure? Call us on 0207 096 5008 and we'll match a build to your framework and model size.
What model sizes will 32GB of VRAM run?
As a rough guide, the RTX 5090’s 32GB comfortably runs quantised 30B-class models at full context length, serves 70B at aggressive quantisation, and handles single-GPU QLoRA fine-tuning as well as SDXL, Flux and video diffusion work. Context length and quantisation both change the maths, so tell us your target and we'll size it precisely.
Is CUDA and PyTorch already installed?
Yes. Every Cortex AI ships with NVIDIA's data-science driver stack, with CUDA and cuDNN version-matched to PyTorch and validated against a live test job before dispatch, so you're training on arrival rather than untangling version mismatches.
Can I run Linux instead of Windows?
Yes. We supply Cortex AI workstations with Windows 11, Ubuntu LTS, or a dual-boot configuration, with the CUDA stack installed and validated on whichever you choose.
Can I add a second GPU later?
Yes. The X870E AORUS Master and Meshify 3 XL are chosen with expansion in mind, and the 1500W HX1500i already has the headroom for a second card. Tell us if multi-GPU is on your roadmap and we’ll lay the build out for it up front. Tell us if multi-GPU is on your roadmap and we'll specify the PSU for it up front.
Can I customise the specification?
Yes — every Cortex AI workstation is fully customisable. We can adjust GPU, memory, storage and cooling to suit your exact model size and budget before we build.
What warranty is included?
A 3-Year Standard Warranty: 1 year of Collect & Return service, 1 year of parts coverage and 3 full years of technical labour support.
Do you offer delivery or collection?
Both. Free UK delivery is included, or you can collect for free from our Central London store at 48 Tottenham Court Road.
Reviews
Clear filtersThere are no reviews yet.